Search

This paper investigates the deformation patterns of surrounding rock in the Xingzishan Tunnel Project, located in Yunnan Province, China, and a method that can respond to the deterioration pattern of the surrounding rock parameters is proposed. The surrounding rock in this project is carbonaceous slate, which possesses the characteristics of anisotropy, and it is hard to detect the mechanical parameters of the rock. A deformation back-analysis method that considers the deterioration effect of the surrounding rock is proposed by using the function approximation and pattern recognition functions of the back propagation (BP) neural network. The fitting equations for the deterioration pattern of the vertical and horizontal elastic moduli E of the surrounding rock are established. Based on the regression prediction of the back propagation (BP) neural network, it is found that the changes in the deformation parameters of the surrounding rock follow certain variation rules within 20 m from the excavation surface. Furthermore, by comparing the deterioration of the surrounding rock in the vertical and horizontal directions, it is considered that the degree of deterioration of the anisotropic surrounding rock varies in different directions. The result is a reference for studying tunnel deformation in anisotropic rocks, and provide a basis for further studies on the deformation mechanisms and control measures of rock surrounding tunnels. [ABSTRACT FROM AUTHOR]

Support vector regression (SVR) is a powerful kernel-based regression prediction algorithm that performs excellently in various application scenarios. However, for real-world data, the general SVR often fails to achieve good predictive performance due to its inability to assess feature contribution accurately. Feature weighting is a suitable solution to address this issue, applying correlation measurement methods to obtain reasonable weights for features based on their contributions to the output. In this paper, based on the idea of a Hilbert–Schmidt independence criterion least absolute shrinkage and selection operator (HSIC LASSO) for selecting features with minimal redundancy and maximum relevance, we propose a novel feature-weighted SVR that considers the importance of features to the output and the redundancy between features. In this approach, the HSIC is utilized to effectively measure the correlation between features as well as that between features and the output. The feature weights are obtained by solving a LASSO regression problem. Compared to other feature weighting methods, our method takes much more comprehensive consideration of weight calculation, and the obtained weighted kernel function can lead to more precise predictions for unknown data. Comprehensive experiments on real datasets from the University of California Irvine (UCI) machine learning repository demonstrate the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

To address issues such as insufficient fishing vessel data records, low energy efficiency, and high emissions, this study constructs and validates a predictive model for the maximum power limitation of fishing vessels. Using gillnet fishing vessels as a case study, the Energy Efficiency Existing Ship Index (EEXI) reference line formula for gillnet fishing vessels is fitted using the nonlinear least squares method. Sensitivity analysis reveals power's impact on fishing vessels' energy efficiency. The maximum power calculation method for fishing vessels is derived by combining the EEXI calculation formula, the reference line formula, and the power-speed relationship. Three regression prediction models--Decision Tree, Random Forest, and Gradient Boosting--are used to construct prediction models with gross tonnage, length between perpendiculars, and gross tonnage and length between perpendiculars as inputs, respectively. Results show that power significantly impacts the energy efficiency of fishing vessels. The EEXI reference line formula for gillnet fishing vessels has MAE, MSE, MAPE, RMSE, and R^2 values of 13.3518, 369.5200, 18%, 19.2229, and 0.6366, respectively. The Random Forest regression model with gross tonnage as input performs best in predicting the maximum power limitation of gillnet fishing vessels, with MAE, MSE, MAPE, RMSE, and R^2 values of 5.61423, 6152.40982, 1.90888%, 78.43730, and 0.633062393, respectively. This paper provides a reliable calculation method and prediction model for the maximum power limitation of fishing vessels, offering systematic technical support and decision-making references for limiting the maximum power of fishing vessels. [ABSTRACT FROM AUTHOR]

As industrial development drives the increasing demand for steel, accurate estimation of the material's fatigue strength has become crucial. Fatigue strength, a critical mechanical property of steel, is a primary factor in component failure within engineering applications. Traditional fatigue testing is both costly and time-consuming, and fatigue failure can lead to severe consequences. Therefore, the need to develop faster and more efficient methods for predicting fatigue strength is evident. In this paper, a fatigue strength dataset was established, incorporating data on material element composition, physical properties, and mechanical performance parameters that influence fatigue strength. A machine learning regression model was then applied to facilitate rapid and efficient fatigue strength prediction of ferrous alloys. Twenty characteristic parameters, selected for their practical relevance in engineering applications, were used as input variables, with fatigue strength as the output. Multiple algorithms were trained on the dataset, and a deep learning regression model was employed for the prediction of fatigue strength. The performance of the models was evaluated using metrics such as MAE, RMSE, R2, and MAPE. The results demonstrated the superiority of the proposed models and the effectiveness of the applied methodologies. [ABSTRACT FROM AUTHOR]

Furnace temperature and oxygen content are important parameters reflecting the combustion inside a circulating fluidized bed (CFB) boiler. Accurately predicting boiler output is a complex task due to the high noise and nonsmoothness of actual boiler input and output data. In this paper, a new hybrid convolutional neural network (CNN), bidirectional long short-term memory (biLSTM) network, and squeezing and excitation (SE) network prediction model is proposed to significantly improve the prediction accuracy of oxygen content and furnace temperature by combining the advantages of multiple deep learning networks. This network can extract spatiotemporal characteristics of input parameters such as coal feed to effectively predict boiler furnace temperature and oxygen content. CNNs can extract complex features such as dynamic and static nonlinearities between multiple variables affecting the furnace temperature and oxygen content, as well as high noise. The biLSTM network layer can efficiently handle the temporal information of irregular trends in modeling time series components; SE can extract the important information between channels through the feature relationships between channels for better overfeature extraction. The CNN-biLSTM-SE model can effectively solve the problem of nonlinear mapping complexity between inputs and outputs. Experiments show that the proposed CNN-biLSTM-SE model outperforms existing methods. The experimental results showed that the average MAPE errors for oxygen content prediction were CNN-biLSTM-SE (0.038), CNN-biLSTM with attention mechanism (AM) (0.043), CNN-biLSTM (0.051), CNN-LSTM (0.051), biLSTM (0.051), RNN (0.051), LSTM(0.0052), and CNN(0.0054). Extensive experiments in CFB boilers with oxygen content and furnace temperature show that the proposed CNN-biLSTM-SE model achieves better results in terms of goodness-of-fit, generalization ability and accuracy. [ABSTRACT FROM AUTHOR]

Through the use of recycled aggregates, the construction industry can mitigate its environmental impact. A key consideration for concrete structural engineers when designing and constructing concrete structures is compressive strength. This study aims to accurately forecast the compressive strength of recycled aggregate concrete (RAC) using machine learning techniques. We propose a simplified approach that incorporates a two-layer stacked ensemble learning model to predict RAC compressive strength. In this framework, the first layer consists of ensemble models acting as base learners, while the second layer utilizes a random forest (RF) model as the meta-learner. A comparative analysis with four other ensemble learning models demonstrates the superior performance of the proposed stacked model in effectively integrating predictions from the base learners, resulting in enhanced model accuracy. The model achieves a low mean absolute error (MAE) of 2.599 MPa, a root mean squared error (RMSE) of 3.645 MPa, and a high R-squared (R2) value of 0.964. Additionally, a Shapley (SHAP) additive explanation analysis reveals the influence and interrelationships of various input factors on the compressive strength of RAC, aiding design and construction professionals in optimizing raw material content during the RAC design and production process. [ABSTRACT FROM AUTHOR]

A large amount of biomass wastes have been generated during agricultural production nowadays. The combustion and gasification can be expected to convert the biomass into clean energy, such as gas, electricity, and heat. Among them, the characteristic temperature of ash fusion is one of the important indexes for the tendency of slagging. Since the biomass ash is a mixture of a variety of minerals, there is no fixed melting point within a certain temperature range. In general, ash fusion temperatures are commonly characterized by four characteristic temperatures, namely the deformation, softening, hemisphere, and flow temperature. Few studies have been conducted to predict the ash fusion temperature of biomass and coal co-combustion ash. Most of them only considered the role of alkaline components in the ash, particularly without the influence of acidic components (SO3 and P2O5). This investigation aims to fully consider the effects of acidic/alkaline components in ash on the ash fusion temperature. Seven oxides were taken as the variables, including Al2O3, SiO2, P2O5, SO3, K2O, CaO, and Fe2O3 in biomass and coal co-combustion ash. A prediction model was established for the ash fusion temperature of biomass and coal co-combustion ash using a multiple linear regression model in the Matlab software. Besides, taking the corn straw and Shenmu bituminous coal in rural areas of China as examples, the ash composition and ash fusion temperatures were measured under different mixing ratios, temperatures, and residence times. The results showed that the content of MgO, K2O, CaO, Na2O, Fe2O3, and MnO oxides increased with the increase of the content of corn straw from 25% to 75%. Especially for K2O, the content increased from 5.89% to 14.41%. The content of acidic oxides decreased gradually, such as Al2O3, P2O5, and SO3, among which the content of Al2O3 decreased from 12.05% to 7.78%, the content of P2O5 decreased from 3.66% to 1.07%, and the content of SO3 decreased from 7.70% to 1.48%. The Cl content decreased outstandingly with the increase of ashing temperature and the extension of residence time. A comparison was made between the prediction and the experimental values, according to the existing empirical formula. It was found that the influence coefficients of P2O5 and SO3 on the characteristic ash fusion temperature were large in the prediction model, which was consistent with the test. It inferred that there was a significant effect of acidic components, such as P2O5 and SO3, on the ash fusion temperatures. Actually, the ash fusion temperature was positively correlated with the content of oxides, such as Al2O3, SiO2, P2O5, and CaO, indicating a great contribution to the inhibition of ash melting and slagging. There was an error of less than 5% in the characteristic ash fusion temperatures between the prediction of the model and the experiment, indicating the better fitting of the multiple linear regression. The accuracy and reliability of the model were verified as well. [ABSTRACT FROM AUTHOR]

The rapid and precise screening of appropriate inorganic perovskite materials poses a formidable challenge in the field of materials science. Traditional methods for material screening are not only time-consuming but also demand a substantial workforce. In this study, we introduce a modified parallel residual network (PRN) for the purpose of predicting the band gap of lead-free inorganic double perovskite materials, utilizing the atomic composition as the input data. The predictive performance of PRN is assessed using root mean square error (RMSE) and Pearson correlation coefficient (r) as evaluation metrics. The PRN model yields a band gap prediction with an RMSE of 0.402 eV and an r value of 0.962, respectively. Notably, PRN outperforms various alternative models, including random forest regression (RFR), kernel ridge regression (KRR), support vector regression (SVR), extreme gradient boosting regression (XGBR), one-layer residual variant network (OLRN), and three-layer parallel residual variant network (TLPRN), clearly demonstrating its superior predictive accuracy. [ABSTRACT FROM AUTHOR]

Support vector regression (SVR) is a powerful kernel-based regression prediction algorithm that performs excellently in various application scenarios. However, for real-world data, the general SVR often fails to achieve good predictive performance due to its inability to assess feature contribution accurately. Feature weighting is a suitable solution to address this issue, applying correlation measurement methods to obtain reasonable weights for features based on their contributions to the output. In this paper, based on the idea of a Hilbert–Schmidt independence criterion least absolute shrinkage and selection operator (HSIC LASSO) for selecting features with minimal redundancy and maximum relevance, we propose a novel feature-weighted SVR that considers the importance of features to the output and the redundancy between features. In this approach, the HSIC is utilized to effectively measure the correlation between features as well as that between features and the output. The feature weights are obtained by solving a LASSO regression problem. Compared to other feature weighting methods, our method takes much more comprehensive consideration of weight calculation, and the obtained weighted kernel function can lead to more precise predictions for unknown data. Comprehensive experiments on real datasets from the University of California Irvine (UCI) machine learning repository demonstrate the effectiveness of the proposed method.

Through the use of recycled aggregates, the construction industry can mitigate its environmental impact. A key consideration for concrete structural engineers when designing and constructing concrete structures is compressive strength. This study aims to accurately forecast the compressive strength of recycled aggregate concrete (RAC) using machine learning techniques. We propose a simplified approach that incorporates a two-layer stacked ensemble learning model to predict RAC compressive strength. In this framework, the first layer consists of ensemble models acting as base learners, while the second layer utilizes a random forest (RF) model as the meta-learner. A comparative analysis with four other ensemble learning models demonstrates the superior performance of the proposed stacked model in effectively integrating predictions from the base learners, resulting in enhanced model accuracy. The model achieves a low mean absolute error (MAE) of 2.599 MPa, a root mean squared error (RMSE) of 3.645 MPa, and a high R-squared (R2) value of 0.964. Additionally, a Shapley (SHAP) additive explanation analysis reveals the influence and interrelationships of various input factors on the compressive strength of RAC, aiding design and construction professionals in optimizing raw material content during the RAC design and production process.

Support vector machine is a very classical and popular model for data prediction. Traditional support vector machines use grid search to determine its parameters. In order to improve the accuracy of prediction, more and more frameworks are proposed. Among them, the combination of support vector machine and intelligent optimization algorithm is the most commonly used solution at present. The optimization objective is to determine the optimal penalty factor and kernel parameters of support vector machine to improve the prediction performance. In this paper, 10 intelligent optimization algorithms that are widely used at present are used for the optimization research of support vector machine. The performance of these optimization algorithms in support vector machine parameter optimization is analyzed in detail. Short-term wind speed and network traffic are chosen as the research object, and detailed performance indicators are given to judge the advantages and disadvantages of these intelligent optimization algorithms in optimizing support vector machine performance. Finally, the performance indicators, optimization speed, running memory usage, optimization success rate of different optimized SVM models, and impact of data distribution are analyzed in detail, and some conclusions are drawn. For the parameters optimization of support vector machine, various indicators are comprehensively considered, grey wolf optimizer algorithm and squirrel search algorithm are recommended. [ABSTRACT FROM AUTHOR]

Copyright of Basic Sciences Journal of Textile Universities / Fangzhi Gaoxiao Jichu Kexue Xuebao is the property of Basic Sciences Journal of Textile Universities and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In order to reduce the energy consumption of air source heat pump drying, a prediction method of energy consumption was proposed to optimize the drying process by means of multivariate linear regression model (MLRM) and back propagation neural network (BPNN) model. On the basis of analyzing characteristics of the energy consumption and factors that affect it, the drying process was proposed to split into sections of equal time to reduce the difficulty of data acquisition. Eight characteristic parameters were set as input, and two parameters were set as output. Input parameters were set temperature in drying room, set humidity in drying room, initial temperature in drying room, initial humidity in drying room, mean ambient temperature, mean ambient humidity, weight of material, and initial moisture content of material. Output parameters were energy consumption and end moisture content of material. Before build predictive models, the variation of the drying power of the air source heat pump was analysed. From the air source heat pump control principle and the drying power curve, the drying process was mainly divided into a heating stage and a insulation stage. The insulation stage consisted of a number of circulating insulation units. As the weight of pea increased, the duration of the heating stage increased, the duration of the insulation stage shortened. Meanwhile, the time of the insulation unit increased and the total drying time increased. Orthogonal experiment was designed using unit energy consumption (energy consumption required to reduce the moisture content of 1 kg of material by 1%) and total energy consumption as the evaluation indexes. According to the results of orthogonal experiment, among the three influencing factors of weight of pea, set temperature in drying room, and set humidity in drying room, the weight of pea had the most significant influence on unit energy consumption and total energy consumption, and the set humidity in drying room had slight significant effect on the total energy consumption. Based on characterization of energy consumption and moisture content of air source heat pump drying, the energy consumption was predicted using MLRM model and BPNN model. The corrective coefficients of determination of the MLRM model and BPNN model for energy consumption and end moisture content were 0.739 and 0.931, respectively. BPNN model with activation function of Sigmoid had the highest coefficient of determination of energy consumption, and the coefficient was 0.828. BPNN model with activation function of Identity had the highest coefficient of determination of end moisture content, and the coefficient was 0.942. The fitting effect was good and met the actual needs of production. Taking rehydrated peas as the drying object, a complete variable temperature and humidity drying process with a mass of 65 kg and a duration of 4 h was designed for verification test. The verification test results showed that the tested total energy consumption of the experiment was 15.066 kW·h, and the total energy consumption values predicted by MLRM model and BPNN model were 14.476 kW·h and 15.183 kW·h, with prediction accuracy of 96.08% and 99.23%, respectively. The tested end moisture content was 8.541%, and predicted moisture content by MLRM model and BPNN model were 9.560% and 8.889%, with prediction accuracies of 88.07% and 95.93%, respectively. This study analyzed the energy consumption characteristics of air source heat pump drying, and proposed an effective means for the prediction of energy consumption for air source heat pump drying, and the prediction was experimentally verified with high reliability of accuracy, which was of great practical significance for the drying process optimization and energy consumption reduction. [ABSTRACT FROM AUTHOR]

Long short-term memory (LSTM) is regarded as one of the most popular methods for regression prediction of time series. In the memory unit of LSTM, since most values of gate structures are usually in the middle state (around 0.5), gate structures cannot effectively retain important information or discard trivial information. Furthermore, the information between two adjacent layers can not be sufficiently transmitted only through the hidden state. To address these issues, we propose a new LSTM structure based on memory cell (MC-LSTM for short) in this paper. First, a new gate stretching mechanism in memory unit is introduced to readjust the distributions of the gates values to push them away from the uncertainty of 0.5. Second, before the memory unit, we establish an interaction gate, in which the input information, the hidden state and the output memory cell of the previous layer interact with each other. By doing so, the information fusion between two adjacent layers can be enhanced and the long-term dependencies can be captured effectively as well. This new method can be used to process time series data, and its goal is to use historical data to predict the value of a future time period. Experimental results on one UCI dataset and eight Kaggle time series datasets validate that the proposed network structure is superior to the most advanced networks. [ABSTRACT FROM AUTHOR]

This study aimed to quickly assess lamb meat freshness using smartphones under various modified atmosphere packaging and storage times (0–10 days). Lamb meat images were collected across different periods and packaging types (50% O2 + 40% CO2 + 10% N2; 70% O2 + 20% CO2 + 10% N2; air). Key color features were extracted, and significant differences in five color features were identified. Three distinct color feature combinations were chosen for models: support vector regression (SVR), genetic algorithm‐back propagation (GA‐BP) neural network, and convolutional neural network (CNN). Results showed SVR models with varied color feature inputs outperformed BP and CNN models. The SVR model with 12 input features yielded the best results, enabling effective spoilage level analysis of lamb meat under different storage times. This work establishes a foundation for future smartphone app development, utilizing Raspberry Pi hardware to evaluate lamb meat freshness under diverse storage conditions. Practical applications: This study introduces a cell phone‐based method to rapidly determine lamb spoilage in air conditioning packages during different storage times. Using cell phone images, we collected data on lamb samples stored for 0–10 days under three packaging conditions. We extracted significant color features and selected three combinations as inputs for prediction models: SVR, GA‐BP neural network, and CNN. The SVR model with 12 feature values demonstrated the best performance. This research lays the foundation for a cell phone application based on Raspberry Pi hardware, enabling users to assess lamb freshness in various storage conditions. The potential benefits include efficient quality control in the food industry, increased consumer confidence, and improved preservation of lamb products through optimized storage and packaging. Overall, this study contributes to food quality evaluation and provides practical applications for professionals and consumers. [ABSTRACT FROM AUTHOR]

Abstract Objective Population ageing, as a hot issue in global development, increases the burden of medical resources in society. This study aims to assess the current spatiotemporal evolution and interaction between population ageing and medical resources in mainland China; evaluate the matching level of medical resources to population ageing; and forecast future trends of ageing, medical resources, and the indicator of ageing-resources (IAR). Methods Data on ageing (EPR) and medical resources (NHI, NBHI, and NHTP) were obtained from China Health Statistics Yearbook and China Statistical Yearbook (2011–2020). We employed spatial autocorrelation to examine the spatial–temporal distribution trends and analyzed the spatio-temporal interaction using a Bayesian spatio-temporal effect model. The IAR, an improved evaluation indicator, was used to measure the matching level of medical resources to population ageing with kernel density analysis for visualization. Finally, an ETS-DNN model was used to forecast the trends in population ageing, medical resources, and their matching level over the next decade. Results The study found that China's ageing population and medical resources are growing annually, yet distribution is uneven across districts. There is a spatio-temporal interaction effect between ageing and medical resources, with higher levels of both in Eastern China and lower levels in Western China. The IAR is relatively high in Northwest, North China, and the Yangtze River Delta, but showed a declining trend in North China and the Yangtze River Delta. The hybrid model (ETS-DNN) gained an R2 of 0.9719, and the predicted median IAR for 2030 (0.99) across 31 regions was higher than the median IAR for 2020 (0.93). Conclusion This study analyzes the relationship between population ageing and medical resources, revealing a spatio-temporal interaction between them. The IAR evaluation indicator highlights the need to address ageing population challenges and cultivate a competent health workforce. The ETS-DNN forecasts indicate higher concentrations of both medical resources and ageing populations in eastern China, emphasizing the need for region-specific ageing security systems and health service industries. The findings provide valuable policy insights for addressing a hyper-aged society in the future.

Accurate measurement of fire source heat release rate is crucial for comprehensively understanding the fire evolution process. However, the widely used oxygen consumption method requires expensive equipment, incurring high costs. This study proposes a comprehensive framework based on machine learning to predict fire source heat release rate using temperature as input. Firstly, fire scenarios with different parameters in ISO9705 room were simulated using FDS software to obtain temperature at various locations, establishing a fire database. Then, two recursive feature elimination algorithms based on the Lasso and the Random Forest (RF) models were employed separately for feature selection, resulting in two different low-dimensional feature subsets and a control group. Finally, different feature subsets were input to analyse and compare the prediction performance on the heat release rate of three typical algorithms: linear regression (LR), K-nearest neighbor (KNN), and lightGBM. Results indicate that the LightGBM model trained with the feature subset selected by the recursive feature elimination algorithm based on the Random Forest model exhibits the best predictive performance, with root mean square error (RMSE) and mean absolute error (MAE) of 23.89 kW and 15.49 kW respectively, and a coefficient of determination (R2) of 0.9916. This comprehensive framework based on machine learning demonstrates excellent predictive performance and is cost-effective, providing a new and effective approach for predicting fire source heat release rate.

After analyzing a large amount of related data, five indicators, such as mining thickness and mining depth, were selected as the main factors influencing the height of the hydraulic fissure zone. On this basis, first the convolutional neural network was trained and tested based on the measured hydraulic fracture zone development height in 40 mining areas to obtain the best convolutional neural network model. Next, the trained network model was used to predict the height of the hydraulic fracture zone of the 1301N working face in the Longgu coal mine. The predicted results were compared with the measured results and the value calculated using the gauge formula and the absolute and relative errors of the model-predicted results were less than the regulatory values. Finally, the convolutional neural network prediction results of the five test cases were compared with back-propagation (BP) neural network and multiple linear regression predicted results, and the absolute error and relative error of the convolutional neural network model prediction results were better than those of the other two predictive models. Thus, the convolutional neural network prediction model is suitable for predicting the height of the hydraulic fracture zone and can predict the developed height of the hydraulic fracture zone more accurately than other predictive methods. [ABSTRACT FROM AUTHOR]

The Bayesian network (BN) method has been successfully applied to evaluate earthquake liquefaction‐induced settlement of free‐field ground in recent years due to several specific advantages. However, the existing BN models need to discretize all continuous variables, and therefore, they can only predict the range value of the settlement (i.e., classification). Thus, information loss is inevitable in the process of discretization, which will largely reduce the prediction accuracy of the model. To realize the application of the BN method in the regression prediction of seismic liquefaction‐induced settlement, this study proposes a hybrid modelling method combining a hill‐climbing algorithm and domain knowledge to construct the structure of a continuous BN regression model based on historical liquefaction‐induced settlement data, and then the conditional linear Gaussian approach is used to learn the conditional probability distributions of parameters. A five‐fold cross‐validation test is used to demonstrate better generalization performance and advantages (such as considering model uncertainty and prior knowledge) of the continuous BN regression model compared with a discrete BN classification model, simplified methods such as the Tokimatsu & Seed and Ishihara & Yoshimine methods, and an artificial neural network model. Their advantages and disadvantages are discussed. In addition, the other two continuous BN models using the Arias intensity and cumulative absolute velocity instead of the peak ground acceleration perform slightly worse than the proposed continuous BN model, and the reason for this difference is discussed. [ABSTRACT FROM AUTHOR]

Abstract Background Preoperative planning of orthognathic surgery is indispensable for achieving ideal surgical outcome regarding the occlusion and jaws' position. However, orthognathic surgery planning is sophisticated and highly experience-dependent, which requires comprehensive consideration of facial morphology and occlusal function. This study aimed to investigate a robust and automatic method based on deep learning to predict reposition vectors of jawbones in orthognathic surgery plan. Methods A regression neural network named VSP transformer was developed based on Transformer architecture. Firstly, 3D cephalometric analysis was employed to quantify skeletal-facial morphology as input features. Next, input features were weighted using pretrained results to minimize bias resulted from multicollinearity. Through encoder-decoder blocks, ten landmark-based reposition vectors of jawbones were predicted. Permutation importance (PI) method was used to calculate contributions of each feature to final prediction to reveal interpretability of the proposed model. Results VSP transformer model was developed with 383 samples and clinically tested with 49 prospectively collected samples. Our proposed model outperformed other four classic regression models in prediction accuracy. Mean absolute errors (MAE) of prediction were 1.41 mm in validation set and 1.34 mm in clinical test set. The interpretability results of the model were highly consistent with clinical knowledge and experience. Conclusions The developed model can predict reposition vectors of orthognathic surgery plan with high accuracy and good clinically practical-effectiveness. Moreover, the model was proved reliable because of its good interpretability.

Copyright of Journal of Beijing University of Technology is the property of Journal of Beijing University of Technology, Editorial Department and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Well logs play a significant role in geological exploration and oil resource development. Due to the limitation of measurement tools and expensive economic cost, obtaining certain well logs can be challenging. The inhomogeneity of the medium and the complexity of the subsurface conditions also make it difficult for linear regression and traditional neural network methods to provide accurate predictions. Multiple existing methods have difficulty in obtaining and utilizing spatiotemporal dependencies among known well logs to predict unavailable well logs. In this letter, an adaptive spatiotemporal graph neural network (ASTGNN) is designed to capture the spatial dependencies and temporal dependencies among various available well logs. A graph convolution neural network (GCN) is introduced into the field of well logs prediction to extract non-Euclidean spatial characteristics among well logs. In addition, a temporal convolution network (TCN) is employed to capture the temporal correlations. Then, a prediction block (PB) is designed to map the hidden states mined by the GCN and TCN to the target predicted well logs. Experiments on single regression well logs prediction and multiple regression well logs prediction are conducted on the dataset from the Norwegian field and the dataset from Daqing, China, respectively. The results demonstrate the accuracy and superior performance of the designed approach. [ABSTRACT FROM AUTHOR]

Objective: Population ageing, as a hot issue in global development, increases the burden of medical resources in society. This study aims to assess the current spatiotemporal evolution and interaction between population ageing and medical resources in mainland China; evaluate the matching level of medical resources to population ageing; and forecast future trends of ageing, medical resources, and the indicator of ageing-resources (IAR). Methods: Data on ageing (EPR) and medical resources (NHI, NBHI, and NHTP) were obtained from China Health Statistics Yearbook and China Statistical Yearbook (2011–2020). We employed spatial autocorrelation to examine the spatial–temporal distribution trends and analyzed the spatio-temporal interaction using a Bayesian spatio-temporal effect model. The IAR, an improved evaluation indicator, was used to measure the matching level of medical resources to population ageing with kernel density analysis for visualization. Finally, an ETS-DNN model was used to forecast the trends in population ageing, medical resources, and their matching level over the next decade. Results: The study found that China's ageing population and medical resources are growing annually, yet distribution is uneven across districts. There is a spatio-temporal interaction effect between ageing and medical resources, with higher levels of both in Eastern China and lower levels in Western China. The IAR is relatively high in Northwest, North China, and the Yangtze River Delta, but showed a declining trend in North China and the Yangtze River Delta. The hybrid model (ETS-DNN) gained an R2 of 0.9719, and the predicted median IAR for 2030 (0.99) across 31 regions was higher than the median IAR for 2020 (0.93). Conclusion: This study analyzes the relationship between population ageing and medical resources, revealing a spatio-temporal interaction between them. The IAR evaluation indicator highlights the need to address ageing population challenges and cultivate a competent health workforce. The ETS-DNN forecasts indicate higher concentrations of both medical resources and ageing populations in eastern China, emphasizing the need for region-specific ageing security systems and health service industries. The findings provide valuable policy insights for addressing a hyper-aged society in the future. [ABSTRACT FROM AUTHOR]

Background: Preoperative planning of orthognathic surgery is indispensable for achieving ideal surgical outcome regarding the occlusion and jaws' position. However, orthognathic surgery planning is sophisticated and highly experience-dependent, which requires comprehensive consideration of facial morphology and occlusal function. This study aimed to investigate a robust and automatic method based on deep learning to predict reposition vectors of jawbones in orthognathic surgery plan. Methods: A regression neural network named VSP transformer was developed based on Transformer architecture. Firstly, 3D cephalometric analysis was employed to quantify skeletal-facial morphology as input features. Next, input features were weighted using pretrained results to minimize bias resulted from multicollinearity. Through encoder-decoder blocks, ten landmark-based reposition vectors of jawbones were predicted. Permutation importance (PI) method was used to calculate contributions of each feature to final prediction to reveal interpretability of the proposed model. Results: VSP transformer model was developed with 383 samples and clinically tested with 49 prospectively collected samples. Our proposed model outperformed other four classic regression models in prediction accuracy. Mean absolute errors (MAE) of prediction were 1.41 mm in validation set and 1.34 mm in clinical test set. The interpretability results of the model were highly consistent with clinical knowledge and experience. Conclusions: The developed model can predict reposition vectors of orthognathic surgery plan with high accuracy and good clinically practical-effectiveness. Moreover, the model was proved reliable because of its good interpretability. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Signal Processing is the property of Journal of Signal Processing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Missing data is a common and inevitable phenomenon. In practical applications, the datasets usually contain noises for various reasons. Most of the existing missing data imputing algorithms are affected by noises which reduce the accuracy of the imputation. This paper proposes a noise-aware missing data multiple imputation algorithm NPMI in static data. Different multiple imputation models are proposed according to the missing mechanism of data. Secondly, the method to determine the imputation order of multivariablesmissing is given. A random sampling consistency algorithm is proposed to estimate the initial values of the parameters of the multiple imputation model to reduce the influence of noise data and improve the algorithm's robustness. Experiments on two real datasets and two synthetic datasets verify the accuracy and efficiency of the proposed NPMI algorithm, and the results are analyzed. [ABSTRACT FROM AUTHOR]

Machine learning methods are effective tools for improving short-term climate prediction. However, commonly used methods often carry out classification and regression prediction modeling separately and independently. Such a single modeling approach may obtain inconsistent prediction results in classification and regression and thus may not meet the needs of practical applications well. To address this issue, this study proposes a selective Naive Bayes ensemble model (SENB-EM) by introducing causal effect and voting strategy on Naive Bayes. The new model can not only screen effective predictors but also perform classification and regression prediction simultaneously. After being applied to the area prediction of summer western North Pacific subtropical high (WNPSH) from 2008 to 2021, it is found that the accuracy classification score (a metric to assess the overall classification prediction accuracy) and the time correlation coefficient (TCC) of SENB-EM can reach 1.0 and 0.81, respectively. After integrating the results of different models [including multiple linear regression ensemble model (MLR-EM), SENB-EM, and Chinese Multi-model Ensemble Prediction System (CMME) used by National Climate Center (NCC)] for 2017–2021, the TCC of the ensemble results of SENB-EM and CMME can reach 0.92 (the highest result among them). This indicates that the prediction results of the summer WNPSH area provided by SENB-EM have a high reference value for the real-time prediction. It is worth noting that, except for the numerical prediction results, the SENB-EM model can also give the range of numerical prediction intervals and predictions for anomalous degrees of the WNPSH area, thus providing more reference information for meteorological forecasters. Overall, as a new hybrid machine learning model, the SENB-EM has a good prediction ability; the approach of performing classification prediction and regression prediction simultaneously through integration is informative to short-term climate prediction. [ABSTRACT FROM AUTHOR]

In this article, a novel design of a hierarchicalfuzzy system (HFS) based on a self-organized fuzzy partition and fuzzy autoencoder is proposed. The initial rule set of the system is empty, and all the fuzzy sets and fuzzy rules are generated by a self-organized fuzzy partition algorithm. By adopting an improved box plot data standardization method, the processed data can more accurately represent the distribution characteristics of the input data, which improve the accuracy and the rationality. A fuzzy autoencoder is used to train the HFS layer by layer, which can not only ensure the effectiveness of the fuzzy system's hidden layer variables but also provide interpretability. Compared with the traditional fuzzy logic system, the HFS reduces the total number of rules and the complexity. The proposed HFS is tested on three different regression datasets. The experimental results illustrate that the hierarchical self-organized fuzzy system still performs better in terms of regression accuracy indicators than the self-organized fuzzy system. [ABSTRACT FROM AUTHOR]

[Display omitted] • The prepared SnO 2 sensor has higher response than the commercial sensors. • SnO 2 sensors have superior gas sensing performance, including selectivity, response, etc. • The dataset with 20 concentrations was collected using SnO 2 and commercial sensors. • The dataset was processed by the FPNet and high evaluation metrics were obtained. The leakage of HCHO gas has direct and irreversible damage on both human health and the environment. However, the current commercial electronic nose (E-nose) has low accuracy in predicting the concentration of HCHO. In this work, a sensor array consisting of 6 gas sensors was fabricated and combined with advanced deep learning algorithms to achieve accurate prediction of HCHO concentration. High-performance SnO 2 sensing material with high specific surface area was synthesized by applying the Metal-Organic Frameworks (MOFs) sacrificial template, and the response of the SnO 2 gas sensor was as high as 15.098, which is much higher than that of commonly used commercial sensors. In addition, the sensing performance of the SnO 2 sensor, including response, selectivity and repeatability etc., were investigated. Afterward, a Formaldehyde Prediction Network (FPNet) model was proposed for processing the HCHO gas dataset and cosine annealing was added to gradually reduce the learning rate during training to help the model converge to the optimal solution. Furthermore, the residual structures with different number of layers were compared, and the 3-layer residual structure expressed the most accurate results of concentration prediction of the HCHO gas, with an MSE of 9.003, MAPE of 0.015, and R2 of 0.998. This study provides a new idea from sensor preparation to pattern recognition algorithm research, effectively solving the problems of the low response of commercial sensors to HCHO and the low accuracy of pattern recognition algorithms for E-nose datasets. [ABSTRACT FROM AUTHOR]

This paper proposes a new multi-kernel learning ensemble algorithm, called Ada- $L_{1}$ MKL-WSVR, which can be regarded as an extension of multi-kernel learning (MKL) and weighted support vector regression (WSVR). The first novelty is to add the $L_{1}$ norm of the weights of the combined kernel function to the objective function of WSVR, which is used to adaptively select the optimal base models and their parameters. In addition, an accelerated method based on fast iterative shrinkage thresholding algorithm (FISTA) is developed to solve the weights of the combined kernel function. The second novelty is to propose an integrated learning framework based on AdaBoost, named Ada- $L_{1}$ MKL-WSVR. In this framework, we integrate FISTA into AdaBoost. At each iteration, we optimize the weights of the combined kernel function and update the weights of the training samples at the same time. Then an ensemble regression function of a set of regression functions is output. Finally, two groups of the experiments are designed to verify the performance of our algorithm. On the first group of the experiments including eight datasets from UCI machine learning repository, the MAEs and RMSEs of Ada- $L_{1}$ MKL-WSVR are reduced by 11.14% and 9.08% on average, respectively. Furthermore, on the second group of the experiments including the COVID-19 epidemic datasets from eight countries, the MAEs and RMSEs of Ada- $L_{1}$ MKL-WSVR are reduced by 31.19% and 29.98% on average, respectively.

With the increasingly serious eutrophication of inland water, the frequency and scope of harmful cyanobacteria blooms are increasing, which affects the ecological balance and endangers human health. The aim of this study was to propose an alternative method for the quantification of cyanobacterial concentrations in water by correlating multispectral data. The research object was the cyanobacteria in Erhai Lake, Dali, China. Ten monitoring sites were selected, and multispectral images and cyanobacterial concentrations were measured in Erhai Lake from September to November 2021. In this study, multispectral data were used as independent variables, and cyanobacterial concentrations as dependent variables. We performed curve estimation, and significance analysis for the independent variables, and compared them with the original variable model. Here, we chose about four algorithms to establish models and compare their applicability, including Multivariable Linear Regression (MLR), Support Vector Regression (SVR), Long Short-Term Memory (LSTM), and Extreme Learning Machine (ELM). The prediction performance was evaluated by the coefficient of determination (R2), Root-Mean-Square Error (RMSE), and Mean Relative Error (MRE). The results showed that the variable analysis model outperformed the original variable model, the ELM was superior to other algorithms, and the variable analysis model based on the ELM algorithm achieved the best results (R2 = 0.7609, RMSE = 4197 cells/mL, MRE = 0.044). This study confirmed the applicability of cyanobacterial concentrations prediction using multispectral data, which can be characterized as a quick and easy methodology, and the deep neural network has great potential to predict the concentration of cyanobacteria. [ABSTRACT FROM AUTHOR]

The bucket fill factor is a core evaluation indicator for the optimization of the loader's autonomous shoveling operation. Accurately predicting the bucket fill factor of the loader after different excavation trajectories is fundamental for optimizing the loader's efficiency and energy cost. Therefore, this paper proposes a method for predicting the bucket fill factor of the loader based on the three-dimensional information of the material surface. Firstly, the co-simulation model of loader shoveling material is established based on the multi-body dynamics software RecurDyn and the discrete element method software (DEMS) EDEM, and the co-simulation is conducted under different excavation trajectories. Then, the three-dimensional material surface information before shovel excavation is obtained from DEMS, and the surface function of the material contour is fitted based on the corresponding shovel excavation trajectory information. Meanwhile, the volume of the material excavated by the loader is obtained by the numerical integration method, and it is divided by the rated bucket volume to obtain the estimated bucket fill factor. Finally, the actual volume of the material after the shovel excavation is divided by the rated bucket volume to obtain the accurate bucket fill factor. Based on this, the prediction model of the bucket fill factor is built. The experimental results show that the proposed method is feasible, with a maximum error of 4.3%, a root mean square error of 0.025 and an average absolute error of 0.021. The research work lays the foundation for predicting the bucket fill factor of construction machinery such as loaders and excavators under real working conditions, which is conducive to promoting the development of autonomous, unmanned, and intelligent construction machinery. [ABSTRACT FROM AUTHOR]

Highlights: What are the main findings? It's feasible to use AutoML for concrete compressive strength prediction. Concrete compressive strength model based on AutoML has stronger robustness than ML. What is the implication of the main finding? AutoML can reduce modeling time and reliance upon engineer modeling experience. Machine learning is widely used for predicting the compressive strength of concrete. However, the machine learning modeling process relies on expert experience. Automated machine learning (AutoML) aims to automatically select optimal data preprocessing methods, feature preprocessing methods, machine learning algorithms, and hyperparameters according to the datasets used, to obtain high-precision prediction models. However, the effectiveness of modeling concrete compressive strength using AutoML has not been verified. This study attempts to fill the above research gap. We construct a database comprising four different types of concrete datasets and compare one AutoML algorithm (Auto-Sklearn) against five ML algorithms. The results show that Auto-Sklearn can automatically build an accurate concrete compressive strength prediction model without relying on expert experience. In addition, Auto-Sklearn achieves the highest accuracy for all four datasets, with an average R 2 of 0.953; the average R 2 values of the ML models with tuned hyperparameters range from 0.909 to 0.943. This study verifies for the first time the feasibility of AutoML for concrete compressive strength prediction, to allow concrete engineers to easily build accurate concrete compressive strength prediction models without relying on a large amount of ML modeling experience. [ABSTRACT FROM AUTHOR]

Abstract Background Due to the high cost of data collection for magnetization detection of media, the sample size is limited, it is not suitable to use deep learning method to predict its change trend. The prediction of physical and chemical properties of magnetized water and fertilizer (PCPMWF) by meta-learning can help to explore the effects of magnetized water and fertilizer irrigation on crops. Method In this article, we propose a meta-learning optimization model based on the meta-learner LSTM in the field of regression prediction of PCPMWF. In meta-learning, LSTM is used to replace MAML’s gradient descent optimizer for regression tasks, enables the meta-learner to learn the update rules of the LSTM, and apply it to update the parameters of the model. The proposed method is compared with the experimental results of MAML and LSTM to verify the feasibility and correctness. Results The average absolute percentage error of the meta-learning optimization model of meta-learner LSTM is reduced by 0.37% compared with the MAML model, and by 4.16% compared with the LSTM model. The loss value of the meta-learning optimization model in the iterative process drops the fastest and steadily compared to the MAML model and the LSTM model. In cross-domain experiments, the average accuracy of the meta-learning optimized model can still reach 0.833. Conclusions In the case of few sample, the proposed model is superior to the traditional LSTM model and the basic MAML model. And in the training of cross-domain datasets, this model performs best.

Reservoir parameter prediction is of significant value to oil and gas exploration and development. Artificial intelligence models are developing rapidly in reservoir parameter prediction. Unfortunately, current research has focused on multi-input single-output prediction models. Meaning, these models use a large amount of logging or seismic data to predict the petrophysical properties of a single reservoir. Another prominent problem is that most mechanistic learning studies have focused on using logging data (e.g., gamma ray and resistivity) to make predictions of reservoir parameters. Although these studies have yielded promising accuracy, a great shortcoming is the inability to obtain such data in logs by seismic inversion. The value of our research work is to achieve a complete description of the reservoir using the elastic parameters from the seismic inversion. We developed a deep learning method based on gated recurrent neural network (GRNN) suitable for simultaneous prediction of porosity, saturation and shale content in the reservoir. GRNN is based on Gated Recurrent Unit (GRU), which can automatically update and reset the hidden state. The input parameters to the model are compressive wave velocity, shear wave velocity and density. The model is trained to fit nonlinear relationships between input parameters and multiple physical parameters. We employed two wells: one for testing and the other for training. 20% of the data in the training wells were used as the validation set. In preprocessing, we performed z-score whitening on the input data. During the training phase, the model hyperparameters were optimized based on the mean absolute error (MAE) box plots of the validation set. Experiments on the test data show that the model has superior robustness and accuracy compared to the conventional recurrent neural network (RNN). In the GRNN prediction results of the test set, the MAE is 0.4889 and the mean squared error (MSE) is 0.5283. Due to the difference in input parameters, our prediction is weaker than the research method using logging data. However, our proposed method has higher practical value in exploration work.

Accurate quality prediction can find and eliminate quality hazards. It is difficult to construct an accurate quality mathematical model for the production of small samples with high dimensionality due to the influence of quality characteristics and the complex mechanism of action. In addition, overfitting scenarios are prone to occur in high-dimensional, small-sample industrial product quality prediction. This paper proposes an ensemble learning and measurement model based on stacking and selects eight algorithms as the base learning model. The maximal information coefficient (MIC) is used to obtain the correlation between the base learning models. Models with low correlation and strong predictive power were chosen to build stacking ensemble models, which effectively avoids overfitting and obtains better predictive performance. To improve the prediction performance as the optimization goal, in the data preprocessing stage, boxplots, ordinary least squares (OLS), and multivariate imputation by chained equations (MICE) are used to detect and replace outliers. The CatBoost algorithm is used to construct combined features. Strong combination features were selected to construct a new feature set. Concrete slump data from the University of California Irvine (UCI) machine learning library were used to conduct comprehensive verification experiments. The experimental results show that, compared with the optimal single model, the minimum correlation stacking ensemble learning model has higher precision and stronger robustness, and a new method is provided to guarantee the accuracy of final product quality prediction. [ABSTRACT FROM AUTHOR]

Support vector machine (SVM) is a supervised machine learning algorithm based on statistical learning theory. Due to low requirements for data and excellent generalization ability in regression and classification modeling, SVM is used widely in data analysis and mining modeling of intelligent transportation system. This paper first introduces the basic principles and open source tools of SVM. Next, this paper summarizes the applications of SVM in regression prediction of passenger flow, traffic congestion, traffic accident and traffic carbon emission. After that, this paper summarizes the applications of SVM in classified prediction of traffic status estimation, traffic sign recognition and traffic incident detection. This paper compares SVM with other widely used algorithms in intelligent transportation system. The research status of the optimization algorithms and derivative algorithms based on SVM are analyzed. Finally, this paper prospects the optimization and application trend of SVM in the future intelligent transportation system.

Recent years witnessed a surge in the number of IoT cameras in smart cities. In this article, an ensemble learning-based prediction model for image forensics from IoT camera is proposed. In particular, our goal is to obtain human body measurements from 2D images taken from two views. Firstly, 24 body part features are extracted by the DensePose algorithm from the two views. Secondly, the features of the upper body part are integrated with height and body weight features. Ensemble learning is then performed with the LightGBM algorithm and a regression prediction model is constructed. The proposed noncontact image prediction method is simple and workable. Its feasibility and validity are verified on an experimental dataset. Experimental results demonstrate that the proposed method is highly reliable in the size prediction of different body parts. Specifically, the mean absolute errors of chest circumference, waistline and hip circumference are about 2.5 cm, while the mean absolute errors of other predictions are about 1 cm.

Missing data is a common and inevitable phenomenon. In practical applications, the datasets usually contain noises for various reasons. Most of the existing missing data imputing algorithms are affected by noises which reduce the accuracy of the imputation. This paper proposes a noise-aware missing data multiple imputation algorithm NPMI in static data. Different multiple imputation models are proposed according to the missing mechanism of data. Secondly, the method to determine the imputation order of multivariablesmissing is given. A random sampling consistency algorithm is proposed to estimate the initial values of the parameters of the multiple imputation model to reduce the influence of noise data and improve the algorithm’s robustness. Experiments on two real datasets and two synthetic datasets verify the accuracy and efficiency of the proposed NPMI algorithm, and the results are analyzed.