Search

The main objective of this article is to solve the trajectory following problem for lower limb exoskeleton robot by using a novel adaptive robust control method. The uncertainties are considered in lower limb exoskeleton robot system which include initial condition offset, joint resistance, structural vibration, and environmental interferences. They are time-varying and have unknown boundaries. We express the trajectory following problem as a servo constraint problem. In contrast to conventional control methods, Udwadia–Kalaba theory does not make any linearization or approximations. Udwadia–Kalaba theory is adopted to derive the closed-form constrained equation of motion and design the proposed control. We also put forward an adaptive law as a performance index whose type is leakage. The proposed control approach ensures the uniform boundedness and uniform ultimate boundedness of the lower limb exoskeleton robot which are demonstrated via the Lyapunov method. Finally, simulation results have shown the tracking effect of the approach presented in this article.

In many fields of today’s society, the value that blockchain big data can bring has attracted more and more people’s attention. Due to people’s concerns, the rapid development of blockchain big data-related technologies, and their extensive application in national strategic guidelines, academic exploration, and life production, many countries now include the development of big data in their national blueprint plans for the future. China’s progress in big data is obvious to everyone, and it plays an important role in exploring the construction and governance of smart city communities. This article explores the research of smart city community governance based on the blockchain big data platform. Through analyzing the role of the blockchain big data platform, using the PBFT blockchain consensus algorithm, technology method, smart community governance model, etc., we try to provide community development suggestions, to further explore community governance innovation from the perspective of smart communities, which will provide new ideas for innovative social governance methods and help improve social service skills as a whole. The experimental results of this paper show that 77%of smart community management is still in its infancy, but smart city community governance based on the blockchain big data platform promotes smart communities to better serve residents and promotes the widespread and effective promotion of distinctive community governance models.

Sensor scheduling for energy-efficient collaborative target tracking in wireless sensor networks (WSNs) is an important problem to deal with the limited network resources. With the recent development and emerging applications of energy acquisition technologies, it has become possible to overcome the bottleneck of battery energy in WSNs using the energy harvesting devices, where theoretically the lifetime of the network could be extended to the infinite. However, the energy harvesting WSN also poses new challenges for sensor scheduling algorithm over the infinite horizon under the limited sensor energy harvesting capabilities. In this article, a novel multistep prediction-based adaptive dynamic programming (MSPADP) approach is proposed for collaborative target tracking in energy harvesting WSNs to schedule sensors over an infinite horizon, according to the ADP mechanism. The “action” module of MSPADP is designed to obtain the sensor scheduling for multiple steps starting from the current step, and implemented by the minimal-cost first search (MCFS) decision tree scheme, and the “critic network” module of MSPADP is iteratively performed to optimize the performance for the remaining infinite steps using neural network. Extended Kalman filter (EKF) is adopted to predict and estimate the target state. The performance index is defined by the tracking accuracy derived from EKF and the energy consumption predicted by the candidate sensor schedule. Theoretical analysis shows the optimality of MSPADP, and simulation results demonstrate its superior tracking performance compared with single-step prediction-based ADP (SSPADP), multistep prediction-based dynamic programming (MSPDP), and multistep prediction-based pruning (MSPP) sensor scheduling approaches. Note to Practitioners —Collaborative target tracking is a typical problem in wireless sensor networks (WSNs) where the sensors need to be scheduled to address the constraints of the limited network resources, such as sensor energy usually supplied by the battery. In the recent years, energy harvesting device has been developed and applied to WSNs to overcome the energy restriction. As the energy harvesting capabilities of the sensors are limited, sensor scheduling remains as a challenging problem and is studied in this article. A novel multistep prediction-based adaptive dynamic programming (MSPADP) approach is proposed for collaborative target tracking, by scheduling sensors for the current time step based on the predictions of the subsequent steps over an infinite horizon. It runs iteratively in two modules: obtaining the previous optimal multistep sensor scheduling and updating the remaining infinite-step performance. Simulation results show its superior tracking performance compared with single-step prediction-based ADP (SSPADP), multistep prediction-based dynamic programming (MSPDP), and multistep prediction-based pruning (MSPP) approaches, and lay a good foundation for the practical applications.

Ubiquitous WiFi signals not only provide fundamental communications for a large number of Internet of Things devices, but also enable to estimate target’s location in a contactless manner. However, most of the existing device-free localization (DFL) methods only utilize the time dynamics of the received WiFi signals, leading to inaccurate DFL in the cluttered indoor environments. Because different layouts of environments and deployments of WiFi devices cause the different mathematical distributions of the data collected from the cluttered indoor environments. In this article, a multiple kernel representation-based extreme learning machine (ELM) is proposed, named integrated multiple kernel ELM (IMK-ELM), for strengthening the localization performance in the cluttered indoor environments utilizing spatiotemporal information. In the proposed IMK-ELM-based DFL, the whole data set is first divided into several subsets depending on their mathematical distributions through the $K$ -means clustering algorithm, and then a corresponding number of local DFL models are built for all the subsets to capture both the time dynamics and spatial properties of the data. Finally, a global DFL model is achieved by seamlessly integrating all the local DFL models due to the consistency mechanism. In addition, the Fresnel zone sensing theory is utilized for helping understand and explain the essence of indoor DFL. Comprehensive experiments indicate that the proposed IMK-ELM-based DFL outperforms state-of-the-art methods in the cluttered indoor environments.

Device-free localization (DFL) is becoming one of the attractive techniques in wireless sensing field, due to its advantage that the target does not need to be attached to any electronic device. However, most of the existed approaches for DFL can only obtain satisfactory localization performance in specific small area, they cannot function well when implemented to complex and large area. In order to tackle this issue, in this article, a hierarchical framework is developed for large-scale DFL based on data and knowledge twin driven integration, which consists of two phases, including the offline training phase and the online localization phase. In the offline training phase, the complex and large monitoring area is first divided into some subdomains by the K-means clustering algorithm, and then training corresponding number of broad learning (BL)-based DFL models for each subdomain using the Fresnel phase difference as the fingerprints. Meanwhile, a class-specific cost regulation extreme learning machine (CCR-ELM) classifier is also trained for determining the attribution of the reference points, which can alleviate the impacts of imbalanced data distribution on classification results. In the online localization phase, the attributions of the testing points are first judged through the trained CCR-ELM classifier, after that, estimating the target’s location in the corresponding subdomains using BL-based DFL models. The validity of the proposed hierarchical framework is evaluated both in small and larger areas, respectively.

Current methods based on the traditional i-vectors and deep neural network (DNN) have shown effectiveness on the speaker identification task, especially with the corpus of large scale. However, when the size of the training dataset is small, the overfitting problem may happen and lead to performance degradation. Besides, the robust identification still remains a challenging problem even under the less strict requirements. This paper proposes a novel approach, SpeakerGAN, for speaker identification with the conditional generative adversarial network (CGAN). It allows the adversarial networks for distinguishing real/fake samples and predicting class labels simultaneously. We configure the generator and the discriminator in SpeakerGAN with the gated convolutional neural network (CNN) and the modified residual network (ResNet) to obtain generated samples of high diversity as well as increase the network capacity. The multiple loss functions are combined and optimized to encourage the correct mapping and accelerate the convergence. Experimental results show that SpeakerGAN reduces the classification error rate by 87% and 16% compared with the traditional i-vector system and the state-of-the-art DNN based method. Under the scenario of limited training data, SpeakerGAN obtains significant improvement over the baselines. In the case of taking 1.6 s of each speaker for testing, SpeakerGAN achieves the identification accuracy of 98.20%, which suggests the promise for short-utterance speaker identification.

Hot metal silicon content is an important indicator for measuring the smooth operation of the blast furnace. However, the hot metal silicon content cannot be directly detected online. Hence, this paper proposes a prediction model of the hot metal silicon content based on the improved multi-layer online extreme learning machine (ML-OSELM). The improved ML-OSLEM algorithm is based on ML-OSELM, the variable forgetting factor (VFF) and the ensemble model. VFF is introduced to make the new coming data get more emphasis. The ensemble model can overcome the overfitting problem of ML-OSELM. This improved algorithm is named as EVFF-ML-OSELM. The real blast furnace production data are used to testify the established prediction model based on EVFF-ML-OSELM. Compared with the prediction models of the hot metal silicon content based on other algorithms, the simulation results demonstrate that the prediction model based on EVFF-ML-OSELM has better prediction accuracy and generalization performance.

Extreme learning machine (ELM) is an emerging machine learning algorithm for training single-hidden-layer feedforward networks (SLFNs). The salient features of ELM are that its hidden layer parameters can be generated randomly, and only the corresponding output weights are determined analytically through the least-square manner, so it is easier to be implemented with faster learning speed and better generalization performance. As the online version of ELM, online sequential ELM (OS-ELM) can deal with the sequentially coming data one by one or chunk by chunk with fixed or varying chunk size. However, OS-ELM cannot function well in dealing with dynamic modeling problems due to the data saturation problem. In order to tackle this issue, in this paper, we propose a novel OS-ELM, named adaptive OS-ELM (AOS-ELM), for enhancing the generalization performance and dynamic tracking capability of OS-ELM for modeling problems in nonstationary environments. The proposed AOS-ELM can efficiently reduce the negative effects of the data saturation problem, in which approximate linear dependence (ALD) and a modified hybrid forgetting mechanism (HFM) are adopted to filter the useless new data and alleviate the impacts of the outdated data, respectively. The performance of AOS-ELM is verified using selected benchmark datasets and a real-world application, i.e., device-free localization (DFL), by comparing it with classic ELM, OS-ELM, FOS-ELM, and DU-OS-ELM. Experimental results demonstrate that AOS-ELM can achieve better performance.

Extreme learning machine (ELM) is an emerging machine learning technique for training single hidden layer feedforward networks (SLFNs). During the training phase, ELM model can be created by simultaneously minimizing the modeling errors and norm of the output weights. Usually, squared loss is widely utilized in the objective function of ELMs, which is theoretically optimal for the Gaussian error distribution. However, in practice, data collected from uncertain and heterogeneous environments trivially result in unknown noise, which may be very complex and cannot be described well using any single distribution. In order to tackle this issue, in this paper, a robust ELM (R-ELM) is proposed for improving the modeling capability and robustness with Gaussian and non-Gaussian noise. In R-ELM, a modified objective function is constructed to fit the noise using mixture of Gaussian (MoG) to approximate any continuous distribution. In addition, the corresponding solution for the new objective function is developed based on expectation maximization (EM) algorithm. Comprehensive experiments, both on selected benchmark datasets and real world applications, demonstrate that the proposed R-ELM has better robustness and generalization performance than state-of-the-art machine learning approaches.

In this paper, an intelligent data-driven optimization scheme is proposed for finding the proper burden surface distribution, which exerts large influences on keeping blast furnace running smoothly in an energy-efficient state. In the proposed scheme, production indicators prediction models are first developed using a kernel extreme learning machine algorithm. To heel, burden surface decision is presented as a multiobjective optimization problem for the first time and solved by a modified two-stage intelligent optimization strategy to generate the initial setting values of burden surface. Furthermore, considering the existence of the approximation error of the created prediction models, feedback compensation is implemented to enhance the reliability of the results, in which an improved association rule mining method is developed to find the corrected values to compensate the initial setting values. Finally, we apply the proposed optimization scheme to determine the setting values of burden surface using actual data, and experimental results illustrate its effectiveness and feasibility.

Collaborative filtering (CF) is the most classical method for recommender system, but it is usually suffered from limited performance by the sparseness of user-to-item rating data. Recently, due to the powerful learning feature representation ability, deep learning components are used to leverage auxiliary information to assist recommendation. However, most existing models based on deep learning are incomplete so that merely extracting the item latent representation and ignoring the user parts. Besides, different data are not chosen from current models. This paper proposes a novel probability framework, named as joint matrix factorization (JMF). There are three components in JMF. Firstly, the modified multilayer crossing version of the factorization machine (MFM) is designed to extract the user latent factors based on user behavior information. Moreover, MFM is a general method which can be used to accomplish many tasks in terms of machine learning. Secondly, a modification of Long Short-Term Memory (LSTM), named as bidirectional LSTM (BLSTM), is used to extract the item latent factors of a document sequence from both front and back directions. Finally, we tightly integrate BLSTM and MFM into probabilistic matrix factorization (PMF) to form JMF. Compared with the classical matrix factorization and other integration models, JMF extracts document data as well as user behavioral data as item vectors and user vectors. Extensive experiments on five real-world datasets show the proposed model has better performance compared with the state-of-the-art recommendation methods.

Using cameras to capture information about a person's gait for identification is a new biometric technology, At present, the recognition accuracy at the angle of 90 degree has reached a high level, but the videos collected in real life are often from other different angles, and the existing methods for multiperspective recognition are not perfect, with a number of problems such as long training time and low accuracy. In view of the above problems, characteristics of Gait Energy Image (GEI) and barycenter features of gait sequence were fused using Kernel canonical correlation analysis (KCCA) in this paper, and Extreme learning machine (ELM) is proposed for classification, and Multi-perspective identification method based on Kernel canonical correlation analysis is proposed. Experimental results show that the feature fusion method can significantly improve the accuracy of recognition from non-90° angles.

Automatic heartbeat classification via electrocardiogram (ECG) can help diagnose and prevent cardiovascular diseases in time. Many classification approaches have been proposed for heartbeat classification, based on feature extraction. However, the existing approaches face the challenges of high feature dimensions and slow recognition speeds. In this paper, we propose an efficient extreme learning machine (ELM) approach for heartbeat classification with multiple classes, based on the hybrid time-domain and wavelet time-frequency features. The proposed approach contains two sequential modules: (1) feature extraction of heartbeat signals, including RR interval features in the time-domain and wavelet time-frequency features, and (2) heartbeat classification using ELM based on the extracted features. RR interval features are calculated to reflect the dynamic characteristics of heartbeat signals. Discrete wavelet transform (DWT) is used to decompose the heartbeat signals and extract the time-frequency features of the heartbeat signals along the timeline. ELM is a single-hidden layer feedforward neural network with the hidden layer parameters randomly generated in advance and the output layer parameters calculated optimally using the least-square algorithm directly using the training samples. ELM is used as the heartbeat classification algorithm due to its high accuracy training accuracy, fast training speed, and good generalization ability. Experimental testing is carried out using the public MIT-BIH arrhythmia dataset to perform a 16-class classification. Experimental results show that the proposed approach achieves a superior classification accuracy with fast training and recognition speeds, compared with existing classification algorithms.

The distribution of blast furnace gas flow has an important influence on the use of chemical and thermal energy. The temperature of the cross measuring points of blast furnace can directly reflect the distribution of gas flow. In order to predict the temperature of the cross measuring points, a prediction model based on GWO-ELM is established in this paper. The bias of the hidden layer and the weights between the input layer and hidden layer of extreme learning machine (ELM) are generated randomly, which results in a decrease in the accuracy of the model. To solve this problem, Grey Wolf Optimizer (GWO) with good global optimization capability is utilized to optimize the ELM prediction model. The experiment results show that this method improves the accuracy of temperature prediction model and the operation of blast furnace might be well guided by this prediction model.

Mild cognitive impairment (MCI) is an aging disease mainly caused by memory impairment after the occurrence of brain lesions. EEG analysis is an effective non-invasive method to recognize brain activity and MCI. Due to the highly non-stationary characteristics of EEG, it is a challenging task to extract features from EEG signals and further improve classification performance for MCI. In this paper, we will present a novel deep learning approach for MCI based on convolutional neural networks (CNN) using EEG signals, where the CNN is used for feature extraction from EEG signals in cognitive tasks, a softmax function is utilized as classifier and creatively the attention mechanism is applied to one of the convolution operations. Experimental results show that the CNN with attention mechanism has an average accuracy of 79.66% (validation) after the accuracy of the validation set has stabilized, which is significantly higher than that of traditional convolutional networks. Compared with the highest accuracy of 70.09% in the other four existing approaches, it shows an obvious advantage. The proposed approach can enrich the convolution features of EEG, improve the model fitting ability and generalization performance, and realize the classification of MCI effectively.

In the past decade, deep learning techniques have powered many aspects of our daily life, and drawn ever-increasing research interests. However, conventional deep learning approaches, such as deep belief network (DBN), restricted Boltzmann machine (RBM), and convolutional neural network (CNN), suffer from time-consuming training process due to fine-tuning of a large number of parameters and the complicated hierarchical structure. Furthermore, the above complication makes it difficult to theoretically analyze and prove the universal approximation of those conventional deep learning approaches. In order to tackle the issues, multilayer extreme learning machines (ML-ELM) were proposed, which accelerate the development of deep learning. Compared with conventional deep learning, ML-ELMs are non-iterative and fast due to the random feature mapping mechanism. In this paper, we perform a thorough review on the development of ML-ELMs, including stacked ELM autoencoder (ELM-AE), residual ELM, and local receptive field based ELM (ELM-LRF), as well as address their applications. In addition, we also discuss the connection between random neural networks and conventional deep learning.

Device-free localization (DFL) is becoming one of the new techniques in wireless localization field, due to its advantage that the target to be localized does not need to attach any electronic device. One of the key issues of DFL is how to characterize the influence of the target on the wireless links, such that the target’s location can be accurately estimated by analyzing the changes of the signals of the links. Most of the existing related research works usually extract the useful information from the links through manual approaches, which are labor-intensive and time-consuming. Deep learning approaches have attempted to automatically extract the useful information from the links, but the training of the conventional deep learning approaches are time-consuming, because a large number of parameters need to be fine-tuned multiple times. Motivated by the fast learning speed and excellent generalization performance of extreme learning machine (ELM), which is an emerging training approach for generalized single hidden layer feedforward neural networks (SLFNs), this paper proposes a novel hierarchical ELM based on deep learning theory, named multilayer probability ELM (MP-ELM), for automatically extracting the useful information from the links, and implementing fast and accurate DFL. The proposed MP-ELM is stacked by ELM autoencoders, so it also keeps the very fast learning speed of ELM. In addition, considering the uncertainty and redundant links existing in DFL, MP-ELM outputs the probabilistic estimation of the target’s location instead of the deterministic output. The validity of the proposed MP-ELM-based DFL is evaluated both in the indoor and the outdoor environments, respectively. Experimental results demonstrate that the proposed MP-ELM can obtain better performance compared with classic ELM, multilayer ELM (ML-ELM), hierarchical ELM (H-ELM), deep belief network (DBN), and deep Boltzmann machine (DBM).

Wireless rechargeable sensor network is a new type of sensor network which has been widely concerned recently. It has the function of charging nodes in network. We use the mobile energy replenishment device to actively move to nodes in the sensor network for charging. Research shows that different charging circuit order will get different charging performance. Therefore, we explore the design of scheduling scheme to achieve better charging performance. In this paper, three new concepts are proposed respectively: the start moment of charging, the energy exhaustion moment and the state of charging to evaluate the scheduling. Based on the above, we define a quadratic performance index as the evaluation of charging performance in sensor network called charging cost of network, the smaller the index is, the better charging performance of the network will be. Through theoretical analysis, firstly, we prove the problem in this paper is NP-complete. Then, considering the charging efficiency of each node to be charged, combined with the quantum behavior particle swarm optimization(QPSO) algorithm, we propose an optimal scheduling scheme for mobile energy replenishment device. Finally, some simulations and theoretical analyses can verify the correctness and effectiveness of the scheme which we proposed.

Micro-grid energy scheduling optimization is a complex nonlinear optimization problem. During the studies, the micro-grid is the low-voltage system whose resistance of the transmission line plays an important role in the system. Difference in traditional micro-grid control scheme, a new control scheme is proposed which the power flow of each transmission line is taken as decision variables, that is to say, the power flow of each line is controlled in order to get the optimal operating cost of micro-grid. Based on the above, in this paper we focus on the micro-grid optimal energy transmission scheduling. It is formulated as a nonlinear quadratic programming problem with quadratic constraints, due to infinite micro-grid operating cycle, it is also an infinite steps optimization problem. Traditional optimal scheduling algorithm is difficult to deal with. Therefore, we propose an adaptive dynamic programing (ADP) algorithm which is effective to solve the infinite steps optimization problem. ADP could avoid meeting the curse of dimensionality caused by the micro-grid optimization, and also has the neural networks which are updated by themselves during applications. By theoretical proof, we obtain an optimal control accuracy and operation efficiency. Finally, numerical simulation results show that the adaptive dynamic programming algorithm has less operating cost and better control scheme compared with the simulated annealing algorithm.

Grey wolf optimizer (GWO) is a meta-heuristic algorithm inspired by the hierarchy of grey wolves (Canis lupus). Fireworks algorithm (FWA) is a nature-inspired optimization method mimicking the explosion process of fireworks for optimization problems. Both of them have a strong optimal search capability. However, in some cases, GWO converges to the local optimum and FWA converges slowly. In this paper, a new hybrid algorithm (named as FWGWO) is proposed, which fuses the advantages of these two algorithms to achieve global optima effectively. The proposed algorithm combines the exploration ability of the fireworks algorithm with the exploitation ability of the grey wolf optimizer (GWO) by setting a balance coefficient. In order to test the competence of the proposed hybrid FWGWO, 16 well-known benchmark functions having a wide range of dimensions and varied complexities are used in this paper. The results of the proposed FWGWO are compared to nine other algorithms, including the standard FWA, the native GWO, enhanced grey wolf optimizer (EGWO), and augmented grey wolf optimizer (AGWO). The experimental results show that the FWGWO effectively improves the global optimal search capability and convergence speed of the GWO and FWA.

Instantaneous physiological signal rates related to cardiopulmonary activities are important indicators of human health assessment. Noncontact vital signals detection using microwave radar is preferable due to its zero disturbance to the subject. This paper presents a Wavelet Analysis (WA) based noncontact heartbeat and respiration signals detection algorithm using millimeter Frequency Modulated Continuous Wave (FMCW) radar. In WA, wavelet packet decomposition is applied to separate heartbeat and respiration signals from radar signal and continuous wavelet transform is used for time frequency analysis. Comparison experiments have been conducted with wearable devices on 10 subjects. Compared with the measurement result of the reference sensor, the average absolute error percentage is less than 2.0% and 3.5% for respiration and heart rate, respectively. In addition, the proposed method improves the accuracy of vital signals detection in comparison with Bandpass filter and Peak Detection (BPK).

Most current recommender systems used the historical behaviour data of user to predict user’ preference. However, it is difficult to recommend items to new users accurately. To alleviate this problem, existing user cold start methods either apply deep learning to build a cross-domain recommender system or map user attributes into the space of user behaviour. These methods are more challenging when applied to online travel platform (e.g., Fliggy), because it is hard to find a cross-domain that user has similar behaviour with travel scenarios and the Location Based Services (LBS) information of users have not been paid sufficient attention. In this work, we propose a LBS-based Heterogeneous Relations Model (LHRM) for user cold start recommendation, which utilizes user’s LBS information and behaviour information in related domains (e.g., Taobao) and user’s behaviour information in travel platforms (e.g., Fliggy) to construct the heterogeneous relations between users and items. Moreover, an attention-based multi-layer perceptron is applied to extract latent factors of users and items. Through this way, LHRM has better generalization performance than existing methods. Experimental results on real data from Fliggy’s offline log illustrate the effectiveness of LHRM.

The deep neural network (DNNs) has been applied in voice conversion (VC) system successfully. DNN shows its effectiveness especially with a large scale of training speech data, but on a small dataset, DNN suffers from the overfitting problem, resulting in performance degradation. Recently, many generative adversarial network (GAN) based VC methods have been developed, which prove to be a promising approach. However, the quality and similarity of generated speech are not entirely satisfactory. The converted speech usually consists of vague contents with buzzy noise and sounds robotic. In this paper, we study and make modification to the basic conditional generative network (cGAN). Focusing on establishing a robust framework for high-quality VC, the proposed method uses a cGAN with a L1 loss and a less explicit constraint, an identity loss. These constraints cooperate to generate speech of higher quality and similarity. We test our method on non-parallel datasets. An objective evaluation demonstrates that this approach achieves better performance and robustness than other GAN-based methods on the intra-gender task. A subjective evaluation shows that the naturalness and similarity are improved.

In the blast furnace production site, the disposable thermocouple is used to measure the hot metal temperature. However, this method is not only inconvenient for continuous data acquisition but also costly for the use of one-time thermocouple. Hence, this paper establishes a prediction model to predict the hot metal temperature. Before the prediction model is established, the corresponding factors of influencing the hot metal temperature are selected, and the noises of production data are removed. In this paper, multi-layer extreme learning machine (ML-ELM) is used as the prediction algorithm of the prediction model. However, the input weights, hidden layer weights and hidden biases of ML-ELM are randomly selected, and the solution of the output weights is based on them, which makes ML-ELM inevitably have a set of non-optimal or unnecessary weights and biases. In addition, ML-ELM may suffer from over-fitting problem. Hence, this paper uses the adaptive particle swarm optimization (APSO) and the ensemble model to improve ML-ELM, and the improved algorithm is named as EAPSO-ML-ELM. APSO can optimize the selections of the input weights, hidden layer weights and hidden biases, the ensemble model can alleviate the over-fitting problem, i.e., this paper combines several of the optimized ML-ELMs which have different input weights, hidden layer weights and hidden biases. Finally, this paper also uses other algorithms to establish the prediction model, and simulation results demonstrate that the prediction model based on EAPSO-ML-ELM has better prediction accuracy and generalization performance.

Extreme learning machine (ELM) was proposed for training single hidden layer feedforward neural networks (SLFNs), and can provide an efficient learning solution for regression problem. However, the prediction error of ELM is unavoidable due to its limited modeling capability, and the nonlinear and stochastic nature of the regression problem. In this paper, a novel ELM, residual compensation ELM (RC-ELM), is proposed for regression problem by employing a multilayer structure with the baseline layer for building the feature mapping between the input and the output, and the other layers for residual compensation layer by layer iteratively. Two real world applications, device-free localization (DFL) and gas utilization ratio (GUR) prediction in blast furnace, are used for experimental testing of the proposed RC-ELM. Experimental results show that RC-ELM has better generalization performance and robustness than other machine learning approaches, including the classic ELM, weighted K-nearest neighbor (WKNN), support vector machine (SVM), and back propagation neural network (BPNN).

Gas utilization ratio (GUR) is an important indicator reflecting the operating state and energy consumption of blast furnace (BF). Skilled operators usually refer to changing trends of GUR to guide the next step of production. For these reasons, this paper establishes a soft sensing scheme based on an improved extreme learning machine (ELM) to predict GUR. In order to enhance the modeling capability of ELM for industrial data, an improved ELM, named GR-ELM, is proposed based on grey relational analysis (GRA) and residual modification mechanism. In GR-ELM, considering the different effective information contained in each input attribute for modeling, the input attribute optimization is proposed combining with GRA and entropy weight method. Then, because the modeling capability of ELM is limited and the data collected from industrial process are usually contaminated, the residual modification mechanism is implemented to improve the reliability of the model. In addition, considering the influence of time delay in BF ironmaking process, generalized correlation coefficient method based on mutual information is used for time delay analysis to eliminate the influence. The real data collected from a BF are applied and validated the performance and effectiveness of the proposed soft sensing scheme. The experimental results show that the proposed soft sensing scheme is available and can achieve better performance than some state-of-the-art algorithms. The soft sensing scheme can provide effective decision support and guidance for further optimization operation.

Reasonable burden distribution matrix is one of important requirements that can realize low consumption, high efficiency, high quality and long campaign life of the blast furnace. This paper proposes a data-driven prediction model of adjusting the burden distribution matrix based on the improved multilayer extreme learning machine (ML-ELM) algorithm. The improved ML-ELM algorithm is based on our previously modified ML-ELM algorithm (named as PLS-ML-ELM) and the ensemble model. It is named as EPLS-ML-ELM. The PLS-ML-ELM algorithm uses the partial least square (PLS) method to improve the algebraic property of the last hidden layer output matrix for the ML-ELM algorithm. However, the PLS-ML-ELM algorithm may have different results in different trails of simulations. The ensemble model can overcome this problem. Moreover, it can improve the generalization performance. Hence, the EPLS-ML-ELM algorithm is consisted of several PLS-ML-ELMs. The real blast furnace data are used to testify the data-driven prediction model. Compared with other prediction models which are based on the SVM algorithm, the ELM algorithm, the ML-ELM algorithm and the PLS-ML-ELM algorithm, the simulation results demonstrate that the data-driven prediction model based on the EPLS-ML-ELM algorithm has better prediction accuracy and generalization performance.

The permeability index of the blast furnace is a significant symbol to measure the smooth operation of the blast furnace. This paper proposes a novel prediction model for permeability index of the blast furnace based on the multi-layer extreme learning machine (ML-ELM), the principal component analysis (PCA) method and wavelet transform (called as W-PCA-ML-ELM prediction model). This modified ML-ELM algorithm is based on the ML-ELM algorithm and the PCA method (named as PCA-ML-ELM). The PCA method is applied on the ML-ELM algorithm to improve the algebraic property of the last hidden layer output matrix which deteriorates its generalization performance due to the high multicollinearity. Because the production data of the blast furnace field contain noises, this paper applies the wavelet transform to remove the noise. Comparing with other prediction models which are based on the ML-ELM, the ELM, the BP and the SVM, simulation results illustrate that the better generalization performance and stability of the proposed W-PCA-ML-ELM prediction model.

Click-through rate (CTR) prediction plays a predominant role in the online advertisements. CTR prediction is a problem of binary classification with imbalanced data. Many existing approaches for imbalance learning only focus on over-sampling and under-sampling, but these methods definitely ignore some vital information of the original data. In this paper, we first propose a weighted output extreme learning machine (WO-ELM) to learn the imbalanced data. A hierarchical extreme learning machine (H-C-ELM) is proposed based on the proposed WO-ELM and the weighted extreme learning machine (W-ELM). The H-C-ELM has two levels in its structure. In the first level, the WO-ELM and the W-ELM are trained on different combined fields of the CTR (each field has some attributes). The two extreme learning machines (ELMs) output their predicted scores of the corresponding combined fields of the CTR. The WO-ELM and the W-ELM have different predicted results on the same combined fields because of the difference of the two ELMs. Therefore, in the second level, another ELM is applied based on the outputs of the two ELMs in the first level and the actual outputs in order to improve the prediction accuracy. The experimental results demonstrate that the proposed H-C-ELM method has better performance for the binary classification with imbalanced data than the other related algorithms on CTR prediction, such as the WO-ELM, the W-ELM, and the stacked autoencoder-logistic regression.

Extreme learning machine (ELM) is a fast learning algorithm for the single-hidden layer feedforward neural networks. However, usually we cannot guarantee the stability of the ELM because the parameters of the ELM are generated randomly, such as its biases of the hidden layer and the connecting weights between the input layer and the hidden layer. Besides, it is hard for a single model to achieve high predicted accuracy on the dataset with low-quality data. In this paper, we first propose a modified residual ELM (R-ELM) to improve the ELM’s learning performance. In R-ELM, the first ELM is trained by the original dataset and the $m$ -th ( $m > 1$ ) ELM will be trained by the residuals between the ground truths and the predicted results of the previous ensemble model (with $m-1$ ELMs). R-ELM (with $m$ ELMs) is built in the direction of the error reduction by calculating the $m$ -th ELM’s optimal weight which is determined by the loss function of the R-ELM. As a result, R-ELM can remember almost all information of the training set. However, this ability does not assure a similar performance on the testing dataset. In view of this problem, we add ${L_{2}}$ regularization to the loss function of the R-ELM (RR-ELM) to avoid the overfitting problem of R-ELM. In RR-ELM, ${L_{2}}$ regularization is employed to encourage each ELM to ignore the unnecessary information of the training set. In order to verify the effectiveness of the two proposed algorithms, the real data from the blast furnace are engaged to perform the experiments. Experimental results illustrate that the proposed RR-ELM and the R-ELM are more stable than the single ELM. These results also demonstrate that the two proposed methods are more accurate than the average outputs of a group of ELMs, the ELM, and the support vector regression.

Due to its much faster speed and better generalization performance, extreme learning machine (ELM) has attracted much attention as an effective learning approach. However, ELM rarely involves strategies for imbalanced data distributions which may exist in many fields. Existing approaches for imbalance learning only consider the effect of the number of the class samples ignoring the dispersion degree of the data, and may lead to the suboptimal learning results. In this paper, we will propose a novel ELM, class-specific cost regulation extreme learning machine (CCR-ELM), together with its kernel based extension, for binary and multiclass classification problems with imbalanced data distributions. CCR-ELM introduces class-specific regulation cost for misclassification of each class in the performance index as the tradeoff of structural risk and empirical risk. The performance of CCR-ELM is verified using a number of benchmark datasets and the real blast furnace status diagnosis problem. Experimental results show that CCR-ELM can achieve better performance for classification problems with imbalanced data distributions than the original ELM and existing ELM imbalance learning approach, and the kernel based CCR-ELM can improve the performance further.

Accurate click-through rate (CTR) prediction can not only improve the advertisement company’s reputation and revenue, but also help the advertisers to optimize the advertising performance. There are two main unsolved problems of the CTR prediction: low prediction accuracy due to the imbalanced distribution of the advertising data and the lack of the real-time advertisement bidding implementation. In this paper, we will develop a novel online CTR prediction approach by incorporating the real-time bidding (RTB) advertising by the following strategies: user profile system is constructed from the historical data of the RTB advertising to describe the user features, the historical CTR features, the ID features, and the other numerical features. A novel CTR prediction approach is presented to address the imbalanced learning sample distribution by integrating the Weighted-ELM (WELM) and the Adaboost algorithm. Compared to the commonly used algorithms, the proposed approach can improve the CTR significantly.

The research of gait parameters is of great significance to understand the law of human motion and the coordination between limbs. To obtain accurate gait parameters, it is necessary to study the contact states of the heel and the toe against the ground during walking, that is, Heel Strike (HS) and Toe Off (TO), which are the basis for dividing the starting point of gait and calculating the characteristic parameters such as gait period. This paper will present a new zero-crossing gait detection method based on a single accelerometer by attaching the accelerometer to the chest of the subject and collecting the triaxial acceleration data when the subject is walking. Equal ripple Finite Impulse Response (EFIR) band pass filter is adopted for data preprocessing, so that the characteristics of zero point in the signal is more obvious. The zero crossing method is used to detect HS, TO and other gait events, then the related gait temporal parameters are calculated. The parameters are compared with the gait parameters calculation system using lower limb motion capture device. The results show that the proposed method can achieve high PPV (Positive Predictive Values).

T-distributed stochastic neighbor embedding (t-SNE) dimension reduction technique are combined with Spectral clustering algorithm to mine and analyze the historical condition data of the blast furnace, and the clustering results are divided into different furnace conditions. The nonlinear dimensionality reduction algorithm tSNE finds the rule in the data by recognizing the observed pattern based on the similarity of data points with multiple features. Most nonlinear techniques except tSNE can not retain local structure and global structure of the data. At the same time, the clustering problem of dataset is transformed into the optimal partitioning problem of graph by spectral clustering, which reduces the impact of blast furnace operating point drift. The test based on the field history library shows its accuracy. At the same time, the data of the blast furnace condition data can be visualized, and the difference between different furnace conditions is demonstrated, which provides convenience for the research of the blast furnace condition in the future. Finally, based on historical data, the well-trained model can also be used for prediction, and it is effective and convenient for the blast furnace operator to adjust the abnormal furnace condition.

Burden distribution matrix is a key part of smooth operation for the blast furnace ironmaking process. It directly influences the burden surface of the blast furnace. Recently, many researchers establish the mathematical model from the burden distribution matrix to the burden surface. However, the research from the burden surface to the burden distribution matrix is rare. Hence, this paper establishes the generative mechanism from the burden surface to the burden distribution matrix. In this paper, our previously improved multi-layer extreme learning machine (named as EAPSO-ML-ELM) is used as the modelling algorithm of the generative mechanism. And the real blast furnace production data are used to verify the generative mechanism. Compared with the generative mechanism based on ML-ELM, the simulation results demonstrate that the generative mechanism based on EAPSO-ML-ELM has better accuracy and generalization performance.

Wireless mobile charging can be used in wireless sensor networks for energy replenishing of sensor nodes by RF energy transmission, where the mobile charging device is employed for charging multiple sensors distributed in different geographical locations. The scheduling of nodes in wireless mobile charging directly affects the energy acquisition effects and work efficiency of the nodes, which is difficult. In this paper, we will propose a novel algorithm–Adaptive Dynamic Programming (ADP) for wireless mobile charging optimization scheduling (ADP-WMCS). ADP-WMCS can jointly schedule node visit sequence and charging starting time for the visiting sensor node by optimizing the overall the energy run out times (i.e. death moment) of the sensor nodes using quadratic performance index, over the infinite horizon. Simulation results show that ADP-WMCS can effectively improve the charging performance of the sensor network.

The classification of electrocardiograms (ECG) plays an important role in the clinical diagnosis of heart disease. This paper proposes an effective system development and implementation for ECG classification based on faster regions with a convolutional neural network (Faster R-CNN) algorithm. The original one-dimensional ECG signals contain the preprocessed patient ECG signals and some ECG recordings from the MIT-BIH database in this experiment. Each ECG beat of one-dimensional ECG signals was transformed into a two-dimensional image for experimental training sets and test sets. As a result, we classified the ECG beats into five categories with an average accuracy of 99.21%. In addition, we did a comparative experiment using the one versus rest support vector machine (OVR SVM) algorithm, and the classification accuracy of the proposed Faster R-CNN was shown to be 2.59% higher.

Unlabeled samples are often readily available in our daily lives. However, valuable information contained in a large number of unlabeled samples tends to be ignored by general supervised learning models. To make full use of unlabeled samples, we propose a novel framework that combines active learning with semi-supervised learning. On one hand, we expect to label as few samples as possible while achieving guaranteed classification performance, hence it's of vital importance to design a specific active learning strategy to select only the most valuable batch of samples for expert labeling. On the other hand, the introduction of distribution information in unlabeled sample pool will bring great benefits to the model. Both labeled samples and unlabeled samples can be used for training semi-supervised classification model. In this paper, uncertainty-based active learning and manifold-based semi-supervised learning are integrated into our framework. Extreme learning machine (ELM) is adopted as our base classifier. Moreover, a novel uncertainty criterion, called Bell-Function-based uncertainty, is proposed for active learning selection for the first time. Empirical results on six public benchmark datasets show that our algorithm produces better performance in comparison with other approaches.

A bstract-Device-free localization is a new and developing technology which estimates an object's locations without requiring it to equip any devices. Channel State Information (CSI), containing more fine-grained information than Received Signal Strength Indication (RSSI), is a natural candidate for localization application and has been studied in many works. Extreme Learning Machine (ELM) is a fast and robust algorithm, but it has only one hidden layer, which limits its capacity. One of the most popular ways of improving accuracy is to use multiple different models to obtain better predictive performance. In this paper, we propose a device-free localization approach using an ensemble of ELMs in which each ELM has the same number of hidden nodes. The proposed approach models the localization task as a regression problem. First, we leverage a modified driver to collect CSI and extract phase information. The Principal Component Analysis (PCA) is then applied to reduce the dimensionality of the phase features. After that, the processed features are fed into an ensemble of ELMs to output their respective predictions. The final prediction is an average combination of them. We conducted experiments in a typical indoor environment to verify its performance, and the results demonstrated the effectiveness of our approach and of CSI.

This paper is concerned with optimal mobile sensor motion coordination for source localization and tracking, taking into account of communication constraints. It is well known that this is a constrained nonlinear optimization problem. The usual way to solve this problem is to adopt a gradient like search algorithm or control strategy to obtain a local optimal solution. In this paper, we propose a novel method, which targets at a global solution for optimal mobile sensor placement in time-difference-of-arrival (TDOA) and hybrid TDOA/angle of arrival-based far-field source localization problems. In the proposed method, first, an analytic solution to optimal mobile sensor positions for source localization and tracking is derived. Then, control laws are designed to force mobile sensors to converge to the expected optimal positions. Simulation results demonstrate our proposed method.

Source localization based on time-difference-of-arrival (TDOA) measurements from spatially separated sensors is an important problem in sensor networks. In this paper, we study the optimal geometries of a team of networked sensors for the TDOA-based localization problem, taking into account the a priori uncertainty of the source location. Two types of TDOA sensor pairing methods, namely, centralized and decentralized , are considered. Analytic solutions to the optimal sensor geometries in the ideal case are derived in both static and movable source cases. In the movable source case, sensor geometry design is investigated under an extended Kalman filter (EKF) framework. Furthermore, under sensor motion and communication constraints, we extend this paper to an optimal sensor path planning problem. Simulations demonstrate its enhancement to the source localization and tracking performance.

This paper establishes an optimal control of unknown complex-valued system. Policy iteration is used to obtain the solution of the Hamilton---Jacobi---Bellman equation. Off-policy learning allows the iterative performance index and iterative control to be obtained by completely unknown dynamics. Critic and action networks are used to get the iterative control and iterative performance index, which execute policy evaluation and policy improvement. Asymptotic stability of the closed-loop system and the convergence of the iterative performance index function are proven. By Lyapunov technique, the uniformly ultimately bounded of the weight error is proven. Simulation study demonstrates the effectiveness of the proposed optimal control method.

Collaborative target tracking is one of the most important applications of wireless sensor networks (WSNs), in which the network must rely on sensor scheduling to balance the tracking accuracy and energy consumption, due to the limited network resources for sensing, communication, and computation. With the recent development of energy acquisition technologies, the building of WSNs based on energy harvesting has become possible to overcome the limitation of battery energy in WSNs, where theoretically the lifetime of the network could be extended to infinite. However, energy-harvesting WSNs pose new technical challenges for collaborative target tracking on how to schedule sensors over the infinite horizon under the restriction on limited sensor energy harvesting capabilities. In this paper, we propose a novel adaptive dynamic programming (ADP)-based multi-sensor scheduling algorithm (ADP-MSS) for collaborative target tracking for energy-harvesting WSNs. ADP-MSS can schedule multiple sensors for each time step over an infinite horizon to achieve high tracking accuracy, based on the extended Kalman filter (EKF) for target state prediction and estimation. Theoretical analysis shows the optimality of ADP-MSS, and simulation results demonstrate its superior tracking accuracy compared with an ADP-based single-sensor scheduling scheme and a simulated-annealing based multi-sensor scheduling scheme.

Extreme learning machine (ELM) is a competitive machine learning approach for training single hidden layer feedforward neural networks with fast learning speed and good generalization performance. However, ELM also suffers from the imbalanced classification problem, which exists in many fields and degrades the performance of classifier significantly. In order to address this issue, in this paper, we propose a novel parallel one-class ELM based on Bayesian approach for imbalanced classification problem, named P-ELM. In P-ELM, the training dataset is firstly divided into k (k is the number of classes) components according to the class attribution of the samples, and then the divided training datasets will be fed into the corresponding k Kernel-based one-class ELM classifiers. Estimation of the probability density based on the output function of Kernel-based one-class ELM classifier is constructed. Thus, the class attribution of the samples can be judged directly by comparing the values of the output function of each Kernel-based one-class ELM classifier of P-ELM based on the Bayesian approach. The detailed performance comparison and discussions of P-ELM are performed with several related class imbalance learning approaches on some selected benchmark datasets including both the binary classification and the multiclass classification problems, as well as a real world application, i.e., the blast furnace status diagnosis problem. The experimental results show that the proposed P-ELM can achieve better performance.

Gas utilization ratio (GUR) is considered as an important indicator to evaluate the operation status and energy consumption of blast furnace (BF). This paper proposes a modified ensemble framework based on extreme learning machines (ELM), named ME-ELM, to establish a prediction model of GUR. Due to the randomly generated hidden layer parameters, ELM may have variations in different trials of simulations. In order to improve the negative impacts of randomness and obtain better prediction accuracy, we investigate an adaptive ensemble framework via ELMs, which can adjust the ensemble weights automatically according to the contribution of the individual ELM model to the final results. Meanwhile, an attenuation factor is added to highlight the role of new coming data to enhance the dynamic tracking capability. To verity the performance of the proposed ME-ELM, experimental comparisons are carried out with the classic ELM and ensemble ELM (E-ELM) to predict GUR using the actual production data. Simulation results demonstrate that ME-ELM is more stable and accurate than ELM and E-ELM.

Currently, class imbalance has been a challenge for classification due to its highly imbalanced instances of distinct classes. With the advantage in quantity, the majority classes can get high accuracy in classification while many instances belonging to minority classes are inclined to be classified as majority classes. In this paper, we propose a novel cost-sensitive method based on multi-layer perceptron (CMMLP) for binary classification with imbalanced data. The proposed cost matrix is used to modify the construction of loss function so as to encourage classifier to pay more attention to the accuracy of minority class by minimizing training error. In order to verify the effectiveness of CMMLP, CMMLP is applied to some benchmark datasets and Click-Through Rate (CTR) prediction datasets. Experimental results illustrate that the new cost-sensitive approach can achieve better performance for binary classification with imbalanced data than the original MLP (OMLP).

Hot metal silicon content is an important indicator of measuring the smooth operation of the blast furnace. However, the hot metal silicon content cannot be directly detected online. Hence, this paper establishes a prediction model of the hot metal silicon content based on multi-layer online sequential extreme learning machine (ML-OSELM). ML-OSELM is an online sequential version of multi-layer extreme learning machine (ML-ELM), and it can learn data one-by-one or chunk-by-chunk. It uses online sequential extreme learning machine auto-encoder (OS-ELM-AE) to perform layer-by-layer unsupervised sequential learning. The prediction model of the hot metal silicon content based on ML-OSELM is verified by using the production data of the blast furnace. Simulation results demonstrate that this prediction model can accurately and quickly predict the hot metal silicon content. Moreover, this prediction model can provide efficient decision for subsequent operation of the blast furnace.

An approximation-free adaptive PID-based sliding mode control (PIDSMC) scheme is designed for nonlinear vehicle platoon subject to asymmetric actuator saturation, capable of guaranteeing, for any initial system condition, string stability of the whole vehicular platoon. It is shown that under the proposed scheme, the whole vehicle platoon can tolerate the asymmetric actuator saturation and unmodeled dynamic nonlinearities, and the scheme also retains the main advantages of the PIDSMC technique. These advantages include robustness and capability to reduce the spacing errors and chattering and to eliminate the steady-state spacing errors. Moreover, adaptive compensation instead of approximation approach such as neuro-network and fuzzy logic approaches is adopted to attenuate the negative effects caused by asymmetric actuator saturation and unmodeled dynamic nonlinearities. Furthermore, constant time headway is used to achieve the string stability and simultaneously to increase traffic density and address the negative effect of nonzero initial spacing, velocity and acceleration errors. Compared with most existing methods, the proposed method does not linearize the system model and neither requires precise knowledge of the system model. Finally, a numerical example is proposed to show the effectiveness of the proposed scheme.

Device-Free Localization (DFL) is becoming one of the new trends in the localization field and has attracted more attentions, due to its advantage that the target to be localized does not need to be attached to any electronic device. In our previous work, we have applied extreme learning machine (ELM) to DFL combining parameterized geometrical feature extraction (PGFE) for extending the input feature of ELM. Although, the proposed PGFE-ELM for DFL is more robust and can obtain better performance, we believe that the localization accuracy of ELM-based DFL can be improved further, due to the dynamic and uncertain signal propagation environment in the monitoring area. Thus, we will propose a novel ELM based on empirical wavelet transform (EWT), named EWT-ELM, to improve the accuracy and efficiency of DFL. EWT-ELM consists of two components, i.e., the prediction model and the residual model. The prediction model is used for building the feature mapping of the input and the output through an ELM. In the residual model, the residual of the prediction model is firstly decomposed into some modes by EWT for alleviating the negative impacts of its nonlinear characteristic, and then, the decomposed modes will be modeled through the corresponding ELMs. In addition, the proposed EWT-ELM for DFL is trained in the offline training phase and performed for real-time estimation of target's location in the online localization phase. Experimental results show that the proposed EWT-ELM can improve the localization accuracy significantly comparing with the classic ELM.

With the development of energy harvesting technologies, the building of wireless sensor networks (WSN) based on energy harvesting has become possible, and helps to weaken the limitation of battery energy in WSN. The main objective of target tracking is to improve the tracking accuracy and to optimize the resource utilization, hence sensor scheduling is essential. Based on an artificial neural network energy acquisition model and extended Kalman filter (EKF) estimation for sensor data fusion, this paper will propose a novel constrained adaptive dynamic programming (ADP) algorithm for multi-sensor scheduling of energy harvesting WSN to optimize the tracking accuracy and resource utilization. Simulation results show that the proposed algorithm is feasible and efficient.