Search

Abstract: Aiming at the problem that paper smoothness cannot be measured on line, a new method for soft measurement of paper smoothness is put forward on the basis of studying paper quantization distribution. A soft measurement system is established, by which, paper quantization noise is processed by time–frequency analysis and non-linear transformation, so that online paper smoothness can be obtained through a Radial Basis Function Neural Networks (RBFNNs). The results show that this soft computing method is a feasible way with high measuring accuracy. [Copyright &y& Elsevier]

• A novel water transparency calculation algorithm is introduced, employing deep neural network and image processing technology. This algorithm obviates the need for a water gauge, allowing the calculation of water transparency solely through the analysisof Secchi disk videos. Notably, it demonstrates heightened detection efficiency and success rates. • An enhanced Siamese tracker, termed SiamDCFF (Siamese tracker based on dual correlation feature fusion), is proposed for Secchi disk recognition. Distinguishing features of SiamDCFF include a dual correlation module and a feature fusion module. On the Secchi disk dataset, SiamDCFF achieves the highest success rate, outperforming other methods detailed in this paper. • The simplified 3D-ResNet [41] is leveraged to assess the state of the Secchi disk, transforming the critical position determination problem into an optimization challenge. The objective function involves maximizing the number of visible and invisible Secchi disk images on both sides of the critical position. This method facilitates adaptive critical position determination. • A novel Gated Recurrent Unit based on auto-regression (ARGRU) is proposed, taking into account the motion characteristics of the Secchi disk. ARGRU achieves a higher success detection rate for water transparency compared to the water gauge recognition method. Notably, ARGRU outperforms RNN [42], GRU, and LSTM [43] on the dataset presented in this paper. Water transparency is traditionally assessed through visual inspection using a lowered Secchi disk (SD) into the water, with the disappearance depth of the SD recorded as the measure of transparency. However, manual and visually-dependent frequency measurements of the SD render this process labor-intensive and time-consuming. This paper presents a comprehensive machine-vision-based algorithm designed for the automatic computation of water transparency using Secchi disk videos. The algorithm leverages multiple deep neural networks, including an enhanced Siamese tracking model, a 3D-ResNet, and an improved GRU network. Trained models directly identify the SD in the video, calculating water transparency by processing pixel information. Experimental results demonstrate the algorithm's efficacy in estimating water transparency across diverse natural environments, achieving commendable accuracy (MAE = 3.6 cm MSE = 21.5 cm RMSE = 4.6 cm), rapid processing speed (average 6.87 s), and robust stability. In comparison with the former water-gauge-based algorithm, our proposed algorithm exhibits heightened efficiency and a superior detection success rate. [ABSTRACT FROM AUTHOR]

• This paper suggests a grating Moiré fringe-based error correction approach that uses an artificial neural network (ANN) and the quartile method (QM). • The approach uses QM for data cleaning operations to eliminate outliers and creates the ANN model with a structure of fully connected layer-normalization layer-fully connected layer for regression training in order to achieve lightweight and stability. • The proposed QM-ANN error correction model's mean absolute error (MAE) can reach 3.3 nm once the QM data has been cleaned and the ANN model has been trained. Many interference factors of the complex and changeable working environment of the grating sensor are the main factors affecting the accuracy; a grating Moiré fringe error correction model based on the quartile method (QM) and artificial neural network (ANN) is proposed in this paper. Firstly, the grating sensor's signals in the working process are collected, and three kinds of interference signals (temperature, humidity and vibration) from the grating sensor working environment are collected simultaneously. Secondly, the QM is used to detect outliers from the collected data and delete data outliers directly. Subsequently, the processed data are input into the ANN model for training. The model is lightweight and stable, because it adopts the cascaded structure of full connection layer-normalization layer-full connection layer, which does not need deploy the complex convolution layer module etc. Finally, the corrected data are output after being processed by the QM-ANN error correction model. The experimental results show that the mean absolute error (MAE) of the proposed QM-ANN error correction model is better than QM-convolutional neural network (CNN), QM-long short-term memory (LSTM) and ANN, CNN, LSTM models without QM. It shows that the proposed model has better robustness and practicability. [ABSTRACT FROM AUTHOR]

Hysteretic system behavior is ubiquitous in science and engineering fields including measurement systems and applications. In this paper, we put forth a nonlinear state–space system identification method that combines the state–space equations to capture the system dynamics with a compact and exact artificial neural network (ANN) representation of the classical Prandtl–Ishlinskii (PI) hysteresis. These ANN representations called PI hysteresis operator neurons employ recurrent ANNs with classical activation functions, and thus can be trained with classical neural network learning algorithms. The structured nonlinear state–space model class proposed in this paper, for the first time, offers a flexible interconnection of PI hysteresis operators with a linear state–space model through a linear fractional representation. This results in a comprehensive and flexible model structure. The performance is validated both on numerical simulation and on measurement data. • Exact recurrent ANN representations of the stop, play & generalized play PI operators • Simultaneous identification of hysteretic & system dynamics using state–space models • The use of rich signals allows to capture the system dynamics and hysteretic effects [ABSTRACT FROM AUTHOR]

The prediction of wear in cutting tools is pivotal for boosting productivity and reducing manufacturing costs. Although current data-driven models in machine learning and deep learning have advanced predictive capabilities, they often lack generality and demand substantial data training. This paper presents a novel approach using Bayesian Regularized Artificial Neural Networks (BRANNs) to precisely forecast wear in milling tools. Unlike conventional machine learning models, BRANNs merge the strengths of artificial neural networks (ANNs) and Bayesian regularization, yielding a more robust and generalized predictive model. We utilized three open-access datasets from the literature alongside an in-house dataset generated by our milling setup. Initially, we assessed the model's predictive ability by training and testing it against individual open-access datasets. We investigated the impact of input features, training data size, hidden units, training algorithms, and transfer functions on the model's predictive capability. Subsequently, we trained the model using three open-access datasets and tested it against our in-house data. Our findings demonstrate that the developed model surpasses existing state-of-the-art models in accuracy and transferability [Display omitted] • Milling tool wear prediction using Bayesian Regularized Artificial Neural Networks. • Proposed BRANN model outperforms existing models in predicting tool wear. • Process parameters and monitoring signals are used as input to predict tool wear. • Training based on open-access data to predict wear for in-house experimental data. [ABSTRACT FROM AUTHOR]

• A data augmentation process based on contrast enhancement techniques is proposed. • A EfficientNet-b0 via Self-attention, is developed with additional layers and minimization weights. • The model is trained using optimized hyperparameters through Bayesian Optimization. • Fused features and selected best features using an improved tree growth optimization algorithm and fuzzy function. One of the most significant issues in the agricultural and remote (RS) sensing industry is detecting infected plants at the early stages. Deep learning-based methods for classifying and detecting agricultural diseases have shown remarkable achievements in technologically sophisticated horticultural research in recent decades. The shortage of imaging data for training a deep neural network model is challenging for accurately classifying fruit disease. Moreover, the remote sensing imaging data includes complex patterns of plant areas that are difficult to identify using conventional techniques. Remote sensing technology efficiently gathers fruit health data, detects disease signs, and monitors citrus orchards for early diagnosis and prevention. The fuzzy deep neural network techniques performed better for classifying remote sensing and digital imaging data for agriculture. This paper proposes a fuzzy deep learning and optimization-based novel framework for citrus fruit disease and agriculture land cover recognition. The Mendeley dataset and NWPU-RESISC45 are employed in this work for the experimental process. The challenge is that these datasets contain a limited number of images. Also, the classes are imbalanced, degrading the learning capability of the proposed models. Therefore, in the first step, we proposed a contrast enhancement technique based on brightness preserving histogram and entropy that generated the improved images that later merged with original data as an augmentation step. We modified the EfficientNet-b0 model in the next step by adding a few convolutional and self-attention layers. Bayesian Optimization has initialized hyperparameter values such as learning rate and momentum. The modified model is trained separately on original and enhanced images to keep distinct fuzzy information at the output layer. After that, deep features are extracted and fused using an Entropy-Serial approach called improved serial fusion. The fused features set observed a few irrelevant information that was further optimized using an improved tree growth optimization algorithm with a fuzzy function. The selected features are finally classified using shallow neural networks and machine learning classifiers. The experimental process obtained an improved average accuracy of 98% and 96.5% on the Mendeley and NWPU datasets, respectively. A t -test is also conducted to check the means of two classifiers (best and worst). In addition, a comparison is performed with recent techniques, showing improved precision and recall rate. [ABSTRACT FROM AUTHOR]

• Low-cost measurement platform for NO2 monitoring developed. • Machine-learning-based calibration procedure proposed. • Combination of neural network and kriging surrogates employed. • NO2 sensor correction applied using reference data and environmental parameters. • Satisfactory correlation between reference and calibrated sensor data demonstrate. Air quality significantly impacts human health, the environment, and the economy. Precise real-time monitoring of air pollution is crucial for managing associated risks and developing appropriate short- and long-term measures. Nitrogen dioxide (NO 2) stands as a common pollutant, with elevated levels posing risks to the human respiratory tract, exacerbating respiratory infections and asthma, and potentially leading to chronic lung diseases. Notwithstanding, precise NO 2 detection typically demands complex and costly equipment. This paper explores NO 2 monitoring using low-cost platforms, meticulously calibrated for reliability. An integrated measurement unit is first presented that contains primary and supplementary nitrogen dioxide sensors, as well as auxiliary detectors for evaluating outside and inside temperature and humidity. The calibration process utilizes data acquired over the period of five months from various reference stations. Employing machine learning with an artificial neural network (ANN)-based and kriging interpolation surrogate models, the correction strategy integrates additive and multiplicative enhancement, predicted by the ANN through auxiliary sensor data such as temperature, humidity, and the sensor-detected NO 2 levels. Extensive verification studies showcase that this calibration approach notably enhances monitoring precision (coefficient of determination surpassing 0.85 concerning reference data, and RMSE of less than four μg/m3), rendering low-cost NO 2 detection practical and dependable. [ABSTRACT FROM AUTHOR]

• Multi-sensor fusing and soft computing in CO 2 leakage measurement. • Integration of BP-ANN, RF, and LS-SVM models. • Feature extraction and analysis of AE and temperature signals. • Effect of impurity gas on CO 2 leakage. CO 2 leakage from carbon capture and storage (CCS) networks may lead to ecological hazards, bodily injury and economic losses. In addition, captured CO 2 often contains impurities which affect the leakage behavior of CO 2. This paper presents a method for continuous and quantitative measurements of CO 2 leakage flowrate and the volume fraction of impurities by combining data-driven models with acoustic emission (AE) and temperature sensors. Three data-driven models based on artificial neural network (ANN), random forest (RF), and least squares support vector machine (LS-SVM) algorithms are established. The outputs from the three data-driven models are then integrated to give improved results. Experimental work was conducted on a purpose-built CO 2 leakage test rig under a range of conditions. N 2 was injected to the CO 2 gas stream as an impurity medium. Results show that the integrated model yields a relative error within ±4.0 % for leakage flowrate and ±3.4 % for volume fraction of N 2. [ABSTRACT FROM AUTHOR]

• This article constructs a deep concentric Siamese network (DCSN) for mining hard samples in small-sample fault diagnosis of marine engines. • DCSN transforms the multi-classification problem under small-sample conditions to a binary-classification problem, which allows for deep models to learn more inter-class discriminative information. • Under the supervision of concentric loss, DCSN imposes strong forces on hard sample pairs, while completely neglects easy sample pairs that can already be correctly classified. • DCSN promotes intra-class aggregation and inter-class separability via shrinking inner boundary. • The efficacy of the constructed DCSN has been validated in a publicly available ship main engine fault dataset. Fault samples of marine engine are extremely scarce, and there are unavoidably some hard samples with small inter-class differences, which pose a serious challenge to fault diagnosis of marine engines. This paper proposes a deep metric learning method, namely deep concentric Siamese network (DCSN), to apply strong forces to hard samples towards their corresponding correct distribution areas under small-sample conditions. First, DCSN is committed to learn discriminative information from limited fault samples through a carefully designed metric learning strategy. Then, DCSN distinguishes hard samples using inner and outer boundaries, and applies strong forces to them, making the deep model more focusing on the correct classification of hard samples. Third, DCSN shrinks the distribution area of intra-class samples, which improves intra-class compactness and inter-class separability. Finally, the experimental results on the marine engine fault dataset show that the proposed DCSN yields higher diagnostic performance compared to the considered competitive methods. [ABSTRACT FROM AUTHOR]

Graphical abstract Highlights • Feature characteristics vary with the bearing's rotational speed. • This paper proposes a reliable fault diagnosis scheme based on acoustic spectral imaging (ASI) of acoustic emission signals. • ASI provides a visual representation of acoustic emission spectral features in images. • The proposed approach provides a robust classifier technique with high diagnostic accuracy. Abstract Incipient fault diagnosis of a bearing requires robust feature representation for an accurate condition-based monitoring system. Existing fault diagnosis schemes are mostly confined to manual features and traditional machine learning approaches such as artificial neural networks (ANN) and support vector machines (SVM). These handcrafted features require substantial human expertise and domain knowledge. In addition, these feature characteristics vary with the bearing's rotational speed. Thus, such methods do not yield the best results under variable speed conditions. To address this issue, this paper presents a reliable fault diagnosis scheme based on acoustic spectral imaging (ASI) of acoustic emission (AE) signals as a precise health state. These health states are further utilized with transfer learning, which is a machine learning technique, which shares knowledge with convolutional neural networks (CNN) for accurate diagnosis under variable operating conditions. In ASI, the amplitudes of the spectral components of the windowed time-domain acoustic emission signal are transformed into spectrum imaging. ASI provides a visual representation of acoustic emission spectral features in images. This ensures enhanced spectral images for transfer learning (TL) testing and training, and thus provides a robust classifier technique with high diagnostic accuracy. [ABSTRACT FROM AUTHOR]

• The multi-channel input fosters the precision and robustness of wear indicator (WI). • Exponentially decaying learning rate can greatly improve the efficiency of network training. • Weighted evaluation metric is more persuasive than one evaluation metric. • WI constructed by the EMDCNN does not need a lot of prior knowledge. Wear indicators (WIs) attempt to identify historical and ongoing degradation processes by extracting features from acquired data. The quality of the constructed WIs affects the validity of the data-driven prediction directly to a great extent. The main problems of the existing WI construction methods are as follows: (1) the existing WI construction methods are based on the single channel sensor signal, resulting in the insufficient use of the measured data; (2) the existing WI construction based on deep learning is using a fixed learning rate, leading to low training efficiency. To solve the above problems, a multi-channel deep convolutional neural network with exponentially decaying learning rate (EMDCNN) is proposed to evaluate the health of rolling bearings. In this paper, the original multi-channel signals are input to the proposed network. Exponentially decaying learning rate is proposed to train the neural network efficiently. Moreover, a weighted evaluation criterion is proposed in this paper. The validation results show that the proposed method is superior to the compared four WI construction methods in monotonicity, trendability, robustness, and the value of weighted criterion is 15.3%, 10.8%, 19.0%, 14.8% higher than that of ECNN-WI, FCNN-WI, NN-WI and SOM-WI respectively. [ABSTRACT FROM AUTHOR]

This paper presents a new nanoforce sensor based on a suspended carbon nanotube gate field-effect transistor. To do so, a numerical investigation of Suspended Gate Silicon-on-Insulator MOSFET (SG-SOIMOSFET) is carried out using ATLAS 2D simulator. Based on the relationship between the nanotube’s deflection and the applied force, a comprehensive study of the proposed nanoforce sensor behavior is performed. Moreover, we describe the evolution of the drain current characteristics as a function of the applied force while examining the influence of capacity variation of the insulating gate on the drain current in the saturation region. It is found that the sensor has a good sensitivity of 230.68 ln(A)/pN. Our second contribution in this paper is to develop a model based on artificial neural networks (ANNs). We successfully integrate our neural model of nanoforce sensor as a new component in the ORCAD-PSPICE electric simulator library; this component must accurately express the behavior of the sensor. A second model based on neural networks, which deals with correction and linearization of the sensor output signal, is designed and implemented into the same simulator. The proposed device can be considered as a potential alternative for CMOS-based nanoforce sensing. [ABSTRACT FROM AUTHOR]

The adhered old mortar paste of recycled concrete aggregates (RCA) plays an important role in production of new concrete with RCA and has positive environmental impacts. A larger number of research papers are available on the properties of concrete where different percentage of natural aggregates (NA) is being replaced by RCA. The outcomes of those research papers have shown that if the good quality of certain percentage (up to 30–50%) RCA is used; the properties of new concrete don’t change when it is compared with concrete made from 100% NA. However, the properties of RCA vary significantly from its source to source. Therefore, it is very important to know the optimum level of RCA that can be used in new concrete. This paper conducted experimental and numerical studies to reveal the probable relationship between the parameters such as percentage of RCA replacement, water to cement ratio, aggregate to cement ratio, percentage of air content in the concrete mix and relate them to the mechanical strength of RCA concrete. Finally, automated neural network search (ANNS) analysis was applied to predict the mechanical strength of RCA concrete when other parameters were used as the inputs of the model. It is observed that proposed relation predicts the compressive and splitting strength of RCA concrete that closely matches the experimental results. [ABSTRACT FROM AUTHOR]

[Display omitted] • Aerial thermographic inspection is done by thermal cameras embedded in UAVs. • A new approach based on image analysis with two consecutive Neural Networks. • Detect thermal patterns associated with superficial faults in an IoT platform. • A real case study is proposed with thermograms from three PV solar plants. • Accuracy of 99 % for PV panel detection, and 96 % for hot spot detection. • False positives are reduced in comparison with other studies. Aerial thermographic inspection is performed with thermal cameras embedded in unmanned aerial vehicles, being one of the most relevant monitoring techniques for photovoltaic panels. This technique allows the detection of thermal patterns associated with faults in the photovoltaic panels, although image analysis and fault detection require novel processing methodologies. The volume and variety of aerial thermal data together with the need for faster and more efficient analysis for early maintenance operations require efficient Internet of Things architectures to provide fast and effective diagnostics with easy access. This paper presents a new approach based on image analysis with two consecutive convolutional neural networks to detect hot spots in an Internet of Things platform and reduce the number of false positives. The architecture of the platform is designed to automatically process the received data through the implementation of different Convolutional Neural Networks. A real case study is proposed with thermograms from three photovoltaic solar plants with different sizes, shapes of panels and a wide temperature range. The results show an accuracy of 99 % for panel detection and 96 % for hot spot detection with a reduction of false positives compared to other studies, such as Support Vector Machine or different Artificial Neural Networks, demonstrating the robustness of the method. [ABSTRACT FROM AUTHOR]

Detecting and tracking the position of multiple targets indoors is a challenging measurement problem due to the inherent difficulty to cluster correctly the sensor data associated to a given target and to track the position of each cluster with adequate accuracy. This problem is critical especially in rooms filled with fixed or moving objects hampering target detection and whenever the paths of different targets cross one another. In this paper, a robust Multiple Targets Tracking (MTT) algorithm exploiting the clouds of points collected from a mmWave-FMCW radar is presented. The proposed solution consists of four main steps. First, the possible outliers of a raw radar data set are removed using a neural network model. Next, the cleaned-up radar data are clustered using the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm. Then, a Kalman Filter (KF) is used to track the position of the centroid of each cluster. Finally, a Structured Branching Multiple Hypothesis Testing (SBMHT) algorithm is applied and updated over reasonably short time intervals to decide which detected tracks are supposed to be confirmed and which ones instead should be discarded. The proposed MTT technique was validated experimentally using the data sets collected from a 60-GHz TI IWR6843 radar platform. The reported results show that the developed algorithm, if properly tuned, is faster and returns more accurate results than other MTT techniques. In particular, the percentage of detection errors is negligible and the planar positioning accuracy is within about 30 cm with 90% probability when up to five targets move freely within the same room. [Display omitted] • A Multiple Targets Tracking (MTT) algorithm exploiting indoor radar data is proposed. • Robust outlier removal and clustering based on a Neural Network and DBSCAN. • A Structured Branching Multiple Hypothesis Testing (SBMHT) algorithm used to confirm/discard tracks. • Planar positioning accuracy (with a 60-GHz radar) is within ±30 cm when 5 people move in the room. • Probability of detection errors and processing latency lower than other MTT algorithms. [ABSTRACT FROM AUTHOR]

Modeling and monitoring generally face numerous challenges such as complex characteristics of multi-unit, temporal correlations and strong interaction among subblocks in large-scale industrial processes. To handle those challenges, a novel process monitoring framework combined temporal feedback autoencoder and multilevel correlation analysis is proposed in this paper. Firstly, large-scale industrial processes are decomposed into multiple subblocks in spatial and temporal order based on process knowledge. Secondly, an improved autoencoder with temporal feedback mechanism is constructed as local monitoring model to capture the important latent variables of each subblock. Then, considering the sequential transmission and correlations among subblocks in series, a multilevel correlation analysis method is employed to efficiently extract the unique features of each subblock and the joint features of the whole process. Finally, the irregular contribution indices of the unique features and the joint features are designed for hierarchical process monitoring. The superiority of the proposed framework can be verified by Tennessee Eastman process and a real hot strip mill process. • A hierarchical process monitoring framework is proposed considering the sequential transmission for large-scale processes. • A novel deep neural network model named temporal feedback autoencoder (TFAE) is designed to capture the potentially important information. • A multilevel correlation analysis (MCA) method, which extracts the joint features and the unique features, is proposed for the first time. [ABSTRACT FROM AUTHOR]

• Low-cost measurement platform for particulate matter monitoring developed. • Artificial intelligence field calibration procedure of low-cost sensor proposed. • Combination of affine scaling and optimized neural network surrogate employed. • Correlation coefficient improved to 0.86 for particulate matter of size 1 μm. • Root-mean-squared error of 3 μm/m3 (1 μm particles) and 4.9 μm/m3 (10 μm particles) Due to detrimental effects of atmospheric particulate matter (PM), its accurate monitoring is of paramount importance, especially in densely populated urban areas. However, precise measurement of PM levels requires expensive and sophisticated equipment. Although low-cost alternatives are gaining popularity, their reliability is questionable, attributed to sensitivity to environmental conditions, inherent instability, and manufacturing imperfections. The objectives of this paper include (i) introduction of an innovative approach to field calibration for low-cost PM sensors using artificial intelligence methods, (ii) implementation of the calibration procedure involving optimized artificial neural network (ANN) and combined multiplicative and additive correction of the low-cost sensor readings, (iii) demonstrating the efficacy of the presented technique using a custom-designed portable PM monitoring platform and reference data acquired from public measurement stations. The results obtained through comprehensive experiments conducted using the aforementioned low-cost sensor and reference data demonstrate remarkable accuracy for the calibrated sensor, with correlation coefficients of 0.86 for PM 1 and PM 2.5 , and 0.76 PM 10 (particles categorized as having diameter equal to or less than 1 μm, 2.5 μm, and 10 μm, respectively), along with low RMSE values of only 3.1, 4.1, and 4.9 µg/m3. [ABSTRACT FROM AUTHOR]

The accurate determination of remaining useful life (RUL) of bearings is of immense importance in the condition-based maintenance of any rotating machinery. In this paper, a data driven prognostic approach based on nonlinear autoregressive neural network with eXogenous Inputs (NARX-NN) in combination with wavelet-filter technique is applied to the RUL estimation of bearings. Firstly, the vibration signals generated in an experimental test rig are processed with the proposed wavelet-filter to augment the impulsive characteristics of bearing signals and improve the quality of fault feature extraction. Secondly, a variety of time-domain features are extracted from the processed bearing signals. However, these features exhibit a highly non-monotonic behavior as the bearing condition degrades. To overcome this drawback, a new health indicator (HI) based on Mahalanobis distance (MD) criterion and cumulative sum (CUMSUM) chart is proposed in this paper. Thirdly, the NARX-NN is first designed as a time delay neural network (TDNN). Then, the derived HI and the age of the bearing are used as inputs with life percentage of the bearing as output in order to train the TDNN model, which unlike the usual artificial neural networks (ANNs) performs a one-step ahead prediction of the bearing RUL. The results suggest that the proposed method can effectively predict the RUL of bearings with an acceptable degree of accuracy, and outperforms the use of self-organizing map-based indicator and the traditional FFNNs for RUL inference. [ABSTRACT FROM AUTHOR]

This paper investigates the sensitivity of statistical fingerprints to different phase resolution (PR) and amplitude bins (AB) sizes of partial discharge (PD) φ - q - n (phase-amplitude-number) patterns. In particular, this paper compares the capability of the ensemble neural network (ENN) and the single neural network (SNN) in recognizing and distinguishing different resolution sizes of φ - q - n discharge patterns. The training fingerprints for both the SNN and ENN comprise statistical fingerprints from different φ - q - n measurements. The result shows that there exists statistical distinction for different PR and AB sizes on some of the statistical fingerprints. Additionally, the ENN and SNN outputs change depending on training and testing with different PR and AB sizes. Furthermore, the ENN appears to be more sensitive in recognizing and discriminating the resolution changes when compared with the SNN. Finally, the results are assessed for practical implementation in the power industry and benefits to practitioners in the field are highlighted. [ABSTRACT FROM AUTHOR]

• A new ensemble-based classification model for breast cancer detection is proposed. • Benchmark clinical data set is utilized to check the effectiveness of the model. • Most important risk factors of breast cancer were extracted using SVM. • Quite interpretable and accurate clinical decision support is yielded by the model. This paper presents a new data mining technique for an accurate prediction of breast cancer (BC), which is one of the major mortality causes among women around the globe. The main objective of our study is to expand an automatic expert system (ES) to provide an accurate diagnosis of BC. Both, Support Vector Machines (SVMs) and Artificial Neural Networks (ANNs) were applied to analyze BC data. The well-known Wisconsin Breast Cancer Dataset (WBCD), available in the UCI repository, was examined in our study. We first tested the SVM algorithm using various values of the C , ɛ and γ parameters. As a result of the first experiment, we were able to observe that the adjustment of these regularization parameters can greatly improve the performance of the traditional SVM algorithm applied for BC detection. The highest obtained accuracy at the first step was 99.71%. Then, we performed a new BC detection approach based on two ensemble learning techniques: the confidence-weighted voting method and the boosting ensemble technique. Our model, called CWV-BANNSVM, combines boosting ANNs (BANN) and two SVMs, using optimal parameters selected during the first experiment. The performance of the applied methods was evaluated using several popular metrics, such as specificity, sensitivity, precision, FPR, FNR, F 1 score, AUC, Gini and accuracy. The proposed CWV-BANNSVM model was able to improve the performance of the traditional machine learning algorithms applied to BC detection, reaching the accuracy of 100%. To overcome the overfitting issue, we determined and used some appropriate parameter values of polynomial SVM. Our comparison with the existing studies dedicated to BC prediction suggests that the proposed CWV-BANN-SVM model provides one of the best prediction performances overall. [ABSTRACT FROM AUTHOR]

This paper proposes a novel ANN-based wind speed forecasting method based in the introduction of low-quality measurements as exogenous information, processed by six prediction models to perform one-hour-ahead enhanced forecasting. The models evaluated are classified in two groups: first, persistence and ARIMA, which are used as references, and secondly, the remaining four, based on neural networks. Model comparison is realized by applying two procedures. On the one hand, four quality indexes are assessed (the Pearson's correlation coefficient, the index of agreement, the mean absolute error and the mean squared error), and the other hand, an ANOVA test and multiple comparison procedure are conducted. A backpropagation network with nine neurons in the hidden layer obtains improvements couples (mean absolute - mean squared error) of 23.92-47.48%, and 23.19-45.54% for the persistence and the ARIMA models, respectively. The paper provides strong practical evidence that traditional agricultural measurements are potentially capable of improving estimates and forecasts. [ABSTRACT FROM AUTHOR]

77 GHz radar has become a promising approach to enhance automotive safety by quickly detecting and identifying targets around the vehicle, especially in harsh weather conditions. This requires 77 GHz radars to provide environmental imaging with high resolution and reliability. However, radar images are easily blurred by sidelobes and background noises, which makes it difficult to extract real target information. In this paper, a sidelobe suppression algorithm based on the point spread function (PSF) and complex-valued neural network has been proposed to discriminate and suppress unwanted sidelobes while maintaining mainlobes referring to targets. To overcome the scarcity of real-world 77 GHz multiple-input multiple-output (MIMO) radar datasets, this paper derives the formula of PSF for 77 GHz MIMO radars in detail and exploits the PSF to generate simulated datasets for training. In addition, to be compatible with the complex-valued radar datasets, a customized neural network model has also been established in this paper. The well-trained neural network is further adopted to suppress sidelobes on real-world radar images. Comprehensive simulations and measurements have proved the superior performance of the proposed method, especially in cases with low signal-to-noise ratio (SNR), large channel mismatches, small target Radar Cross-Section (RCS) and large observation angle. • This paper derived the point spread function for 77 GHz MIMO radars. • A complex-valued neural network model was established to cope with raw radar data. • A 77 GHz radar model based on PSF was established to train the neural network. [ABSTRACT FROM AUTHOR]

• This paper proposes an ACM-based non-contact, low magnetization level, and high sensitivity detection technique for the deformation of oil and gas pipelines. • By analyzing the ACM waveform signals, the peak value, integrated area, first-order differential peak and trough, and peak and trough length of the first-order differential are proposed as the features. • The proposed SSA-BP algorithm can characterize the critical deformation dimensions (height, length, tilt angle) within the mean relative error of 10%. Accurate and quantitative characterization of deformation in oil and gas pipelines is essential. This paper proposed a novel ACM (alternating current magnetization) based technique to detect the deformation of oil and gas pipelines. Numerical simulations and experiments reveal the relationships between the deformation factors (height, length, tilt angle) and the detected waveform signals. Meanwhile, the peak value, integral area, first-order differential peak and valley value, peak and valley length of the waveform signals are selected as the features. In addition, a BP neural network model optimized by SSA (sparrow search algorithm) was introduced to identify the deformation of the pipelines. The results show that the waveform signals corresponding to the deformation due to external stress and corrosion are distributed in the mountain peak and basin shape, respectively. With features as input, the proposed SSA-BP algorithm can efficiently characterize the deformation within the mean relative error of 10%. [ABSTRACT FROM AUTHOR]

Abstract: This paper is submitted upon a research to enhance the output of an accelerometer widely used in inertial navigation units. In our study we introduce relatively simple and effective test beds to collect accurate and diverse reference data. We also carried out calibration runs with a highly accurate electromechanical shaking table. The proposed test beds compare well with sophisticated counterparts. The collected data is used to train artificial neural networks (ANNs) which would improve the accelerometer outputs by estimating the reference data from the actual sensor outputs. The ANN performance is compared with classic low pass filtering methods to provide a relative performance criterion. In this paper we focus on test beds rather than to give the details of the whole study. The test beds introduced in this research can be used for acquiring reference data for implementation of other different filter methods as well. [Copyright &y& Elsevier]

• Spectral analysis results in ppm for metallic elements mixed to the engine lubricant. • Determining maintenance time and/or engine failure of the diesel locomotives. • Measurement of the electrical properties of the engine lubricant samples. • Artificial neural network study as a different approach for locomotive maintenance. In this paper, we proposed an approach for locomotive maintenance systems by observing engine lube oil. The mechanical particles in lube oil give information about locomotive engine system condition. The engine lubricant is monthly monitored by a spectral analyzer (SA) to detect engine system failure and routine maintenance time. However, this old fashioned technique has many disadvantages such as non-real time measuring, high cost and time consumption. A novel approach is proposed to eliminate these disadvantages. The new method determines the lubricant sample conditions with respect to electrical characteristics by using artificial neural network (ANN). The study focuses on a relationship between mechanical particles (in ppm) and dielectric characteristics of the lube oil samples. Therefore, ANN method is applied to observe linear relation between observed and predicted dielectric constant and loss factor values of the engine oil samples. The electrical characteristics of the samples are observed at four frequency points (2.40 GHz, 5.80 GHz, 7.40 GHz and 9.60 GHz). ANN studies are realized by using data at these frequency points. The regression (R) coefficients are obtained as 0.7239, 0.7951, 0.8513 and 0.7463 for dielectric constant and 0.7627, 0.7196, 0.8015 and 0.7334 for dielectric loss, respectively. Moreover, the mean square error (MSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) are calculated and examined. The obtained results are very sufficient and this approach can be applied to a sensor device having low cost and real time working mechanism in the future. [ABSTRACT FROM AUTHOR]

• To improve of the quality of lung image and diagnosis of lung cancer. • The lung CT images are collected from CIA dataset. • Noises are eliminated by applying weighted mean histogram equalization approach. • To enhance the quality of the image, the IPCT is used. Automatic lung disease detection is a critical challenging task for researchers because of the noise signals getting included into creative signals amid the image capturing process which may corrupt the cancer image quality thusly bringing about the debased performance. So as to evade this, Lung cancer preprocessing has turned into an imperative stage with the key parts as edge detection, lung image resampling, lung image upgrade and image denoising for improving the nature of input image. Image Denoising is a critical pre-processing task preceding further preparing of the image like feature extraction, segmentation, surface examination, and so forth which elminates the noise whereas retaining the edges and additional complete features to the extent possible. This paper deals with improvement of the quality of lung image and diagnosis of lung cancer by reducing misclassification. The lung CT images are collected from Cancer imaging Archive (CIA) dataset, noise present in the images are eliminated by applying weighted mean histogram equalization approach which successfully removes noise from image, also enhancing the quality of the image, using improved profuse clustering technique (IPCT) for segmenting the affected region. Various spectral features are derived from the affected region. These are examined by applying deep learning instantaneously trained neural network for predicting lung cancer. Eventually, the system is examined by the efficiency of the system using MATLAB based simulation results. The system ensures that 98.42% of accuracy with minimum classification error 0.038. [ABSTRACT FROM AUTHOR]

• A convolutional neural network (CNN) based method for recognizing motion blurred visual targets was proposed. • A motion blur model based on inner-frame path superposition imaging was established. • An automatic blur image synthesizing method was put forward for generating the large amount of training data of the CNN. • An optimization framework was set up to reconstruct the 3D target moving path during camera exposure. Vision-based geometry measurement plays a crucial role in many science and industrial areas. Plenty of researches devoted to measuring static objects, while few focused on motion blurred situations, which inevitably arise when the object being measured moves fast relative to the camera(s). Motion blur usually invalids the vision-based measurement algorithms designated for static objects. In this paper, we devote to accurate three dimensional (3D) reconstruction of moving objects from motion blurred stereo image pairs. A convolutional neural network (CNN) based method is first proposed to recognize the motion blurred visual targets. A motion blur model based on inner-frame path superposition imaging is then established. Finally, an optimization framework is set up to reconstruct the 3D target motion path during the camera exposure. Experiments are involved to demonstrate the validity and accuracy of the method. [ABSTRACT FROM AUTHOR]

• The evolutionary correlated gravitational search algorithm (ECGS) for selecting the optimized features. • Analyzes each diabetic feature according to the correlation and mutual information is selected. • The selected features are processed by GHNN for predicting the diabetic related features effectively. • The efficiency of the system is implemented using MATLAB tool. • It utilizes the Pima Indian Diabetic Dataset for analyzing the efficiency of introduced diabetic expert system. In worldwide 415 million of peoples are affected by diabetics in the year of 2015, that is increased from the year of 2012. Based on the survey, it clearly shows the diabetics are one of the dangerous diseases because it leads to create several risk of early death. Due to the seriousness of the diabetic, it has been detected in early stage by creating expert system. During this process, the expert system has several issues such as accuracy of prediction due to the huge dimension of the diabetic feature that reduce the entire efficiency of the system. So, in this paper introduced the evolutionary correlated gravitational search algorithm (ECGS) for selecting the optimized features. The introduced method analyzes each diabetic feature according to the correlation and mutual information is selected with minimum computation time and cost. The selected features are processed by genetic optimized Hopfield neural network (GHNN) for predicting the diabetic related features effectively. Then the efficiency of the system is implemented using MATLAB tool that utilizes the Pima Indian Diabetic Dataset for analyzing the efficiency of introduced diabetic expert system. The efficiency of the system is evaluated in terms of using mean square error rate, F-measurer, accuracy, confusion matrix and ROC curve. [ABSTRACT FROM AUTHOR]

• Performance degradation trend affects the useful life of ultrasonic motor. • Hybrid neural network model has a significant improvement in prediction ability. • Condition monitoring data manifests the performance degradation exteriorly. • Different characteristics parameters of ultrasonic motor are highly correlated. In this paper, a data-driven prediction method combining condition monitoring data and Elman neural network is proposed, this method obtains the remaining useful life of ultrasonic motor by predicting the tendency of motor performance degradation index. Firstly, the improved particle optimization algorithm is employed to enhance the prediction accuracy of Elman neural network. Principal component analysis is used to extract the motor degradation index from condition monitoring data. Then Elman neural network prediction model is established to predict the variation trend of the degradation index, and the motor failure threshold λ is applied to evaluate the value of motor remaining useful life. Finally, the proposed model is used to perform the prediction test on three PMR60 ultrasonic motors and compare with three benchmark models. Experimental results indicate that the proposed method is a new effective method for estimating the remaining useful life of ultrasonic motor. [ABSTRACT FROM AUTHOR]

• Discarding lifetimes of rotation resistant rope were determined experimentally. • Novel multiple linear regression model was devised. • Novel theoretical discarding lifetime prediction equation was presented. • Discarding lifetimes were predicted by using artificial neural networks. • Results shed light on failure formation cycles for this rope. In this paper, theoretical and experimental studies are conducted to exhibit the discarding fatigue lifetime of a rotation resistant rope exposed to alternate bending over sheave (BoS) fatigue. Experimental studies are fulfilled to determine discarding lifetimes of a rotation resistant rope exposed to BoS fatigue. Multiple linear regression analysis is performed and novel theoretical discarding lifetime prediction formula is determined by using the least square method. Furthermore, discarding lifetimes of rotation resistant ropes exposed to BoS fatigue is predicted by using artificial neural network (ANN). There is a vigorous correlation among the results acquired by regression model, ANN and experimental data. [ABSTRACT FROM AUTHOR]

• To analyze the tooth decay and acid erosion from European teeth biomedical data. • To collect information from children having age 5. • The teeth decay activities are monitored by EMOCA algorithm with RTRFNN. • To analyze the changes & characteristics of children biomedical teeth data. Now-a-days most of the children faced tooth decay and acid erosion problem in their teeth because of continuous bacterial infection, acid segregation, presents of food particles in teeth and so on. Especially, children are more affected by tooth decay, that leads to create severe problem like gingivitis, teeth loss and teeth pain. Due to the importance of tooth decay it needs to predict in earlier condition for eliminating children teeth problem such anorexia and bulimia disorders. Hence the bacterial infection of teeth is critical to be predicted from affected teeth. So, in this paper we analyze the tooth decay and acid erosion from European teeth biomedical data portal which collects information from children having age 5. The teeth decay activities are monitored by evolutionary multi-objective cuckoo feature selection (EMOCA) algorithm with Ruzzo–Tompa optimized regulatory feedback neural network (RTRFNN) that successfully analyze the changes and characteristics of children teeth biomedical teeth data. The introduced method effectively evaluates children tooth data before making the final decision about tooth decay and acid erosion. Then the excellence of the system is evaluated with the help of the experimental results, Ruzzo–Tompa optimized regulatory feedback neural network recognize the abnormal dental features with 99.22% of accuracy. [ABSTRACT FROM AUTHOR]

• The proposed method can online approximate unknown functions in control process. • This method can ensure better real-time in deal with uncertainty and disturbance. • The ultimately uniformly bounded for this controller can be proved theoretically. • Experimental results show strong robustness w.r.t. uncertainties and disturbances. The electromagnet levitation control system is the core component of maglev trains, which has a significant influence on the performance of the maglev train. However, the control system suffers from the essential strong nonlinear and open-loop unstable. Moreover, the model uncertainty and many exogenous disturbances make the controller design even harder. In this paper, the mathematical model of maglev system is established firstly. Then, using the nonlinear transformation method, the affine nonlinear mathematical model of the maglev system is obtained without any linear approximation. Based on the presented model, we design a sliding mode controller based on the exponential reaching law preliminarily and the stability is proved. Since the control characteristics of the maglev system are highly uncertain and time varying with external disturbance, a radial basis function (RBF) neural network estimator is added to the proposed controller. To improve the convergence speed and better satisfy the requirements of real-time control, the minimum parameter learning method is adopted to replace the weights in the neural network without model information. The boundedness and convergence of the presented control law are proved by Lyapunov method. Finally, both simulation and experiment results are included to verify the effectiveness of the proposed control strategy. [ABSTRACT FROM AUTHOR]

• ANN is well capable to predict the compressive strength of boron-based geopolymers. • ANN-created prediction network resulted in regression coefficient of above 0.99. • GP predicted the compressive strength of BASG with solid accuracy and consistency. • Si and B ions are the most influential factors on the compressive strength of BASG. This paper employs artificial intelligence methods in order to create a function for compressive strength of the boroaluminosilicate geopolymers based on mixture proportion variables. Boroaluminosilicate geopolymers (BASGs), a group of boron-based alkali-activated materials, not only minimise the carbon footprint in the construction industry but also decrease the consumption of energy and natural resources. Australian fly ash and iron making slag are activated in sodium and boron-based alkaline medium in order to produce the geopolymer binders. The current study employs artificial neural network in order to classify the collected data into train, test, and validation followed by genetic programming for developing a function to approximate the compressive strength of BASGs. The independent variables comprise the percentage of fly ash and slag as well as ratios of boron, silicon, and sodium ions in the alkaline solution. The performance of each method is assessed by the acquired regression and the error parameters. The obtained results show that the percent of silicon and boron ions, with positive direct correlation and the largest power in the function respectively, have the most significant effects on the compressive strength of BASG. The assessment factors, including R-squared 0.95 and root-mean-square error 0.07 in the testing data, indicate that the model explains all the variability of the response data around its mean. It implies a high level of accuracy and reliability for the model. [ABSTRACT FROM AUTHOR]

• Monitoring biaxial bending structural deformations by optical fiber. • An improved conjugate beam method is proposed to calculate the deflection. • An algorithm is established by BP neural network to calculate the deflection. • Taking a concrete simple beam test to verify the above two methods. Concrete structures are often subjected to loads in both directions, resulting in deflections in both directions. To comprehensively and automatically monitor these continuous deformations, this paper uses a distributed optical fiber to monitor the strain of a structure and establishes the strain-deflection relationship by both an improved conjugate beam method and a BP neural network. The improved conjugate beam method applies to simple structures, such as simple beams, and the BP neural network applies to complex structures. A concrete simple beam model is used to verify the two methods. The results show that the relationships between the strain and deflection that are established by the two methods have high precision, and the results of the BP algorithm are closer to the actual deflections than the results of the explicit function method. [ABSTRACT FROM AUTHOR]

• Using microwave time series method to measure the water-to-liquid ratio of oil-water-gas three-phase flow. • A wavelet coefficient method was proposed to improve the predictive performance of the water-to-liquid ratio. • A CNN model was proposed to improve the predictive performance of the water-to-liquid ratio. This paper proposes a method for predicting the water-to-liquid ratio (WLR) of oil-water-gas three-phase flow based on microwave amplitude and phase time series. Two methods of extracting time series information have been proposed. One method is the wavelet coefficient method, which proves time series information does help to improve the predictive performance of the WLR. After that, based on the advantages of time series another method—a convolutional neural network model was proposed to fully extract time series information, and then fit the nonlinear relationship between measured values and WLR. The experimental results show that the convolutional neural network has good performance. [ABSTRACT FROM AUTHOR]

Highlights • Springback is influenced by various factors in the sheet metal forming process. • Neural network has the potential to solve finite element analysis complexity. • The existing neural network method is unable to represent all springback factors. Abstract The prediction of springback in sheet metal is vital to ensure economical metal forming. The latest nonlinear recovery in finite element analysis is used to achieve accurate results, but this method has become more complicated and requires complex computational programming to develop a constitutive model. Having the potential to assist the complexity, computational intelligence approach is often regarded as a statistical method that does not contribute to the development of a constitutive model. To provide a reference for researchers who are studying the potential application of computational intelligence in springback research, a review of studies into the development of sheet metal forming and the application of neural network to predict springback is presented in this research paper. It can be summarized as: (1) Springback is influenced by various factors that are involved in the sheet metal forming process. (2) The main complexity in FE analysis is the development of a constitutive model of a material that has the potential to be solved by using the computational intelligence approach. (3) The existing neural network approach for solving springback predictions is unable to represent all the factors that affect the results of the analysis. [ABSTRACT FROM AUTHOR]

• Innovative machine learning methods to predict the fracture energy of asphalt mixture. • Formulating fracture energy in terms of various predictor variables. • The GEP model appears to be more practical than the ANN/SA model. • The models can be used for pre-design purposes. This paper presents innovative machine learning methods called gene expression programming (GEP) and hybrid artificial neural network/simulated annealing (ANN/SA) to predict the fracture energy of asphalt mixture specimens. The GEP and ANN/SA models are developed using an experimental database including a number of disk-shaped compact tension (DC(T)) test results for fracture energy. The fracture energy is formulated in terms of various predictor variables such as asphalt binder performance grading (PG), asphalt content, aggregate size, aggregate gradation, reclaimed asphalt pavement (RAP) content, reclaimed asphalt shingles (RAS) content, crumb rubber content, and test temperature. A calculation procedure is presented to interpret the models and transform them into practical design equations. A sensitivity analysis is conducted to evaluate the effect of these predictor variables on the fracture energy. Based on the results, the proposed design equations accurately characterize the fracture energy of asphalt mixtures. The GEP model appears to be more practical than the ANN/SA model because of its better generalization and simpler functional structure. The models are recommended for pre-design purposes or as a means to determine asphalt mixture fracture energy when testing is not feasible. [ABSTRACT FROM AUTHOR]

• The optomechatronic system detect deformations in eggshell. • Laser projection and a cubic spline interpolation are used for carrying out the image processing. • The neural network i sable to clasify eggs without deformations. This paper describes a method for detecting defects on curved surfaces. In particular, this research focuses on defects in poultry eggs for damage identification in the shell as a result of thin-shelled eggs. The vision system is based on defect detection by scanning a laser pattern of structured light and imaging, highlighting the changes in geometry as a result of deformation of the laser transitions generated by scanning the egg surface. Then, the images are analyzed to obtain equidistant points along the curve and evaluated by creating a cubic spline interpolation. The interpolation allows for the extraction of descriptive metric characteristics to observe the disparity between curves, illustrating the defects by performing graph interposition. The obtained metric information is used to classify the defective samples by developing an algorithm using an artificial neural network, trained with a database composed of 200 images, wich achievies 97.5% efficiency during the evaluation of 150 egg samples. This technique can be applied to detecting corrugated, wrinkled, pimpled, odd- shaped and misshapen eggs. [ABSTRACT FROM AUTHOR]

• Precision plastic biaspheric lens having high and uneven thickness was developed using injection molding process. • A hybrid ANN and PSO technique was employed to predict the optimal process parameters. • Mould with aspheric profiles was prepared using single point diamond turning machine. • Surface characterisation like form, waveiness and surface roughness of mould and moulded bi-aspheric lens are measured. • Aberrations in the injection molded lens were studied using Shack-Hartmann Wavefront Sensor. Injection molding of bi-aspheric lens using polycarbonate material with minimum variation in volumetric shrinkage is crucial for optical quality and is more challenging task among the researchers. In this paper, a hybrid artificial neural networks (ANN) and particle swarm optimization (PSO) technique is used to predict the optimal process parameters of injection molding process of the bi-aspheric lens. The developed ANN network (7-13-6) was trained as well as tested with experimental data sampled from statistical methods. The well trained and tested ANN network was coupled with improved PSO algorithm as a hybrid ANN-PSO to optimize the injection molding process parameters. The optimized injection molding process parameters obtained from hybrid ANN-PSO algorithm are validated with experiments using J. S. W injection molding machine. It is observed from the lens quality parameters that the proposed hybrid ANN-PSO method optimized the injection molding process of the bi-aspheric lens with an optical power of 27.73 Diopter and the lens posses seventh order spherical aberrations. [ABSTRACT FROM AUTHOR]

Highlights • Method simulates effects of launch conditions on rail-gun current and armature speed. • The input five parameters are closely related to the rail-gun launchment. • Three neural networks predict the rail-gun current and armature speed with low error. • General Regression neural network is optimal on error, time cost, parameter design. Abstract In this paper, based on the advantages of artificial neural network, such as good tolerance of data noise, strong ability of nonlinear mapping, multi-dimensional input variables, fast operation, low error, etc., a method of using artificial neural network for data prediction is proposed for the research of rail-gun. The results show that it is feasible to use the Back Propagation Neural Network, the Radial Basis Function Neural Network and the General Regression Neural Network to realize the method of prediction and simulation of the rail-gun current and the armature speed curve through relevant parameters. The General Regression Neural Network has superiority in error performance and time cost of neural network training and simulation. [ABSTRACT FROM AUTHOR]

Highlights • Experimental study for ultrasonic welding process of polymer blends is undertaken. • Data comprising of two measured outputs (weld strength, heat generation) is obtained. • Artificial neural network is applied to build the predictive models for each output. • 2-D and 3-D plots of models give detailed insights into welding parameters. • Optimization of models using NSGA II results in maximum value of weld strength. Abstract This paper presents a study undertaken with an objective to establish ultrasonic welding process for joining polymer blends expressed to aid the eco-friendly qualities desired in manufacturing sectors. Polycarbonate (PC) and Acrylonitrile Butadiene Styrene (ABS) blends are welded after creating suitable parts with energy directors using injection molding techniques. It is imperative to estimate the performance of the weld preferred in industrial sectors to be expressed in terms of strength along with the maximum heat generated. Experiments are conducted by varying three of the process parameters namely amplitude, pressure and weld time with measurement of responses such as the tensile strength and heat generated. Artificial neural network (ANN) algorithm is then used to formulate models for each of the measured response. NSGA II is then applied for optimization of models for achieving higher weld strength created with an optimal level of heating. The weld strength of 6.02 N mm−2 is achieved with the welding parameters of amplitude (33.14 µm), pressure (4.03 bar) and weld time (3.35 s). The heat generated at the weld (146.20 °C) is achieved with the welding parameters of amplitude (40.89 µm), pressure (4.29 bar) and weld time (4.52 s). [ABSTRACT FROM AUTHOR]

• Image analysis was carried out by exploiting a trained artificial neural network. • Spray drying with suitable selection and share of carrier can be an effective solution. • Texture analysis could be used instead of physicochemical analysis for identification of spray-dried. In the paper an attempt was made to evaluate influence of type (maltodextrin, gum Arabic and inulin) and content (50, 60, 70% solids) of carrier on quality of raspberry powders. The different types of powders were compared taking into account the structure of microparticles, water activity and their humidity. The use of modern methods such as: low temperature spray drying, artificial intelligence together with visual technique supported by electron microscope are undoubtedly an innovation in this solution. The aim of this undertaking is monitoring the process in order to obtain raspberry powders characterized by high quality characteristics. The paper shows the created neural models, which allow to obtain homogeneity on the basis of microparticles of powders as well as their microbiological condition (indirectly via water activity and humidity). The created networks were characterized by low Root Mean Square and high effectiveness of classification on the level of 99%. [ABSTRACT FROM AUTHOR]

Abstract: The present paper describes the development of an ANN based phase correction system which has been employed in conjunction with a real automatic digital ac bridge. The proposed ANN-based phase corrector has been developed using backpropagation learning employing resilient backpropagation (popularly known as RPROP). Significant improvements have been obtained in the proposed phase correction system for measuring impedance and reported in the paper. [Copyright &y& Elsevier]

In the paper new methods of fault localisation and identification in linear electronic circuits (two-port or multi-port type) based on bilinear transformations in multi-dimensional spaces are presented. The novelty of these methods lies in transferring a family of identification loci from a plane to multi-dimensional spaces. It implies greater distances between the loci and, in consequence, better fault resolution as well as robustness against non-faulty component tolerances and measurement errors. The methods can be used for diagnosis of electronic circuits in conventional testing systems and neural networks. They may be also useful in parameter identification measurements of other multi-parameter objects modelled by electrical circuits.In the paper we present the idea of the new methods, with particular consideration of the 3D, 4D and 6D methods, algorithms of single-fault and double-fault diagnosis, some results of experimental verification of the 4D method and implementation of the 4D and 6D methods in neural networks. [Copyright &y& Elsevier]

• A modified, inverse-estimation analytical model of heat flux is presented; • The nonlinear inverse model is approximated by an artificial neural networks; • A hot-wall heat flux sensor for the nonlinear inverse estimation is developed. Compared with cold-wall heat flux measurement, hot-wall heat flux indicates more information from dynamic heat transfer between the aerothermodynamic boundary layer and the true model's surface; thus, hot-wall heat flux measurement has become predominant in the aerothermodynamic measurement field. In harsh aerothermodynamic in-flight or ground tests, the hot-wall heat flux must be determined from time history temperature measurements at one or more interior locations. Therefore, accounting for the temperature dependence of the thermophysical properties, hot-wall heat flux measurement essentially results in the solution of the nonlinear inverse heat conduction problem (IHCP). In this paper, a novel inverse estimation of hot-wall heat flux using nonlinear artificial neural networks (ANN) is presented. First, motivated by the hybrid method proposed by Clayton A. Pullins, David O. Hubble, and Tom E. Diller et al., [2010], a modified hybrid heat flux measurement method using two in-depth thermocouples is proposed, which avoid to directly measure surface temperature of gauge or model under harsh aerodynamic heating environments; accounting for the unknown temperature dependent thermophysical properties, a new heat flux inverse estimation model of a nonlinear ANN is proposed and identified to approximate the modified hybrid measurement method through calibration experiment. This heat flux inverse estimation method does not need to solve a first kind Volterra integral equation and to obtain the information about the thermophysical properties of heat conduction body, and the thermal inertia, locations of thermocouples. This paper presents xenon lamp calibration and arc-heated wind tunnel experiments that validate the new inverse estimation method combined with the fabricated hot-wall heat flux sensor and its probe. These experimental results show that, in general, the dynamical hot-wall heat flux estimated based on the proposed method agree well with the known calibrated values and the stagnation heat fluxes of the classic slug calorimeter. [ABSTRACT FROM AUTHOR]

• The instrument is used to study on pore structure of cement-based materials. • Standard resistance method is used to calibrate the modulus and phase of impedance. • SVM model is used to predict calibration value of modulus and phase of impedance. • SVM model has the best comprehensive performance among FT,ANN. The non-contact alternating current (AC) impedance method is widely used in the study of electrochemical properties of cement-based materials,the instrument of this paper provide a new way for the study of pore structure of cement-based materials. Due to the loss of magnetic materials and the parasitic capacitance of measuring circuit, the ratio difference and phase difference of current sensor inevitably exit. Traditionally the standard resistor calibration method is used to determine the ratio error and phase error under different frequencies based on polynomial fitting (FT). In this paper, a machine-learning support vector machine (SVM) model is used to predict the calibration data with a small test sample. For the accuracy assessment, SVM model has the best comprehensive performance among FT, artificial neural network (ANN). The actual impedance test of typical cement-based materials under different frequencies is given and the results verify the excellent performance of SVM. [ABSTRACT FROM AUTHOR]

Due to the complicated structure and varying operating conditions of machinery in various applications, intelligent identification of the health state based on the vibration data is still a great challenge in fault diagnosis. In this paper, a variant of the convolutional neural network, named dynamic ensemble convolutional neural network was proposed for fault diagnosis by intelligent fusion of the multi-level wavelet packet. First, wavelet packet transform was employed to construct multi-level wavelet coefficients matrixes for representing the nonstationary vibration signal comprehensively. Then, several paralleled convolutional neural networks with shared parameters were built, not only to learn the multi-level fault features automatically, but also to restrain the overfitting of the deep learning partially. At last, a dynamic ensemble layer was applied to fuse multi-level wavelet packet by assigning weights dynamically. The validation on two experimental datasets of the planetary gearbox under varying speed demonstrated that the developed method can fuse the fault features in multi-level wavelet packet thoroughly, and improve the effectiveness and robustness for fault diagnosis of gearbox under whether the sufficient or limited fault data conditions. [ABSTRACT FROM AUTHOR]

This paper presents a new fuzzy change detection and measurement approach to overcome the drawbacks of traditional thresholding methods in remote sensing. The proposed technique is taking the advantages of following concepts: (1) asymmetric thresholding in order to improve the reliability and accuracy of change detection in each spectral bands, (2) fuzzy measurement approach to consider ambiguity in thresholding of difference image, and represent the changes in fuzzy form and fusing the obtained change maps from various spectral bands, and (3) non-linear modeling with artificial neural networks for relative radiometric normalization (RRN). The performance of developed technique is evaluated on a pair of Landsat 5, 7 images which were taken over the southern part of Urmia Lake, Iran. The major merits of developed technique compared to prevalent thresholding techniques are: higher accuracy in change detection and measurement, higher capability in detection of multiple changes and lower dependence on quality of RRN process. [ABSTRACT FROM AUTHOR]

Computer-aided detection (CAD) provides an efficient way to assist doctors to interpret fundus images. In a CAD system, retinal vessel (RV) detection is an important step to identify the retinal disease regions automatically and accurately. However, RV detection is still a challenging problem due to variations in morphology of the vessels on a noisy background. In this paper, we formulate the detection task as a classification problem and solve it using a convolutional neural network (CNN) as a two-class classifier. The proposed model has 2 convolution layers, 2 pooling layers, 1 dropout layer and 1 loss layer. The contributions of the algorithm are two-fold. First, a new model of CNN is designed to automatically extract features and classify the retinal vessel region. Compared to traditional classification procedures, it is fully automatic and does not need preprocessing and manual extraction and description of features. Second, a novel reinforcement sample learning scheme is proposed to train the CNN with fewer iterations of epochs and less training time. The proposed model is trained and tested using the Digital Retinal Images for Vessel Extraction (DRIVE) and Structured Analysis of the Retina (STARE) data sets. The proposed CNN achieves better performance and significantly outperforms the state-of-the-art for automatic retinal vessel segmentation on the DRIVE data set with 91.99% accuracy and 0.9652 AUC score (area under ROC), and on the STARE data set with 92.20% accuracy and 0.9440 AUC value. We further compare our result with several state-of-the-art methods based on AUC values. The comparison shows that our proposal yields the second best AUC value. This demonstrates the efficiency of the proposed method without pre-processing and with high accuracy and training speed. [ABSTRACT FROM AUTHOR]

This paper proposes a fault distance estimation scheme for fixed series capacitor compensated parallel transmission lines using discrete wavelet transform and decision tree regression. The purpose of the data mining based scheme is to avoid the complicated equation based methods that have been suggested by researchers to overcome the drawbacks of conventional fault location scheme. Although decision tree has inherent advantage over other methods like artificial neural network and support vector machines to work with large data sets, it has not been used in fault location estimation in series compensated (SC) transmission line so far. Decision tree is chosen to locate the faults because of its ability to work with large data set and high accuracy in associating the fault pattern to the fault distance using regression analysis. The discrete wavelet transform processed signals makes the decision process of decision tree regression easy by providing appropriate features. The proposed method is evaluated with variation of fault location, fault type, pre-fault load angle, location of series capacitor, degree of series compensation, fault inception angle, line parameters, inter-circuit faults and fault resistance. The test result of decision tree regression based location estimation scheme ensures that, it can estimate the fault distance accurately. [ABSTRACT FROM AUTHOR]