Human action recognition methods based on multimodal channels, especially the strategy combining RGB and skeleton information, have achieved remarkable success. However, the existing action recognition methods still face challenges in terms of insufficient multimodal fusion, imperfect multi-scale modeling, and inadequate preservation of spatial information. This paper proposes an innovative framework: by combining early bidirectional lateral connections with late fusion, deep interaction between modalities and complementarity at the prediction layer are achieved; the Multi-scale Spatiotemporal Attention Weighting (MSAW) mechanism is designed to adaptively fuse spatiotemporal features across different scales, selecting key frames in spatiotemporal sequences; the Pixel Sub-sampling Downsampling (PSD) module is proposed, which effectively retains spatial details during the strong downsampling process. A large number of experiments on the NTU RGB+D 60, NTU RGB+D 120, and Kinetics-400 datasets verified the effectiveness of the proposed method. The results show that our method demonstrates stronger competitiveness compared with the current state-of-the-art (SOTA) methods. Our code is available at:https://github.com/course-prog/MASW

This study focuses on cooperative tracking of multiple high speed trains, an effective way to accommodate rising passenger demand now that train to train communication is mature. A novel distributed adaptive model predictive control algorithm built on multi-point mass model is proposed. Firstly, the cooperative tracking task is recast with individual cars as the smallest nodes, yielding a distributed model predictive control strategy in which each car acts as a control agent. Subsequently, an adaptive weighting matrix is introduced to adjust the weights of multiple control input variables during the rolling optimization of model predictive control, thereby improving control accuracy and reducing tracking error. Furthermore, to accelerate computation, an adaptive prediction horizon scheme is presented that shortens or lengthens the horizon in real time according to the current tracking error. Finally, simulations with three high speed trains demonstrate that the proposed multi-point mass distributed adaptive model predictive control approach is both feasible and stable, highlighting its potential for enhancing cooperative tracking performance of multiple high-speed trains.

To address the control challenges of virtual coupling (VC) for heterogeneous permanent magnetic maglev trains with time-varying delays and uncertain disturbances, an observer-based distributed nonlinear consensus controller is proposed for a third-order nonlinear model. Specifically, a state observer is utilized to estimate the leader’s acceleration information, thereby reducing the communication burden. To compensate for disturbances, a fixed-time-convergence disturbance observer is proposed and theoretically demonstrated. In particular, the train-following interactions are implemented to avoid negative spacing errors as well as unreasonable acceleration and deceleration. Subsequently, the single-train and string stability are analyzed, and an upper bound on the allowed communications delay is provided. Finally, simulations demonstrate that the proposed controller exhibits strong robustness and anti-interference performance, enabling the VC to operate safely, comfortably, and reliably in the presence of communication delays and disturbances.

A high-speed train (HST) is a physically interconnected underactuated system consisting of both motor cars and trailer cars. During operation, all cars experience varying degrees of aerodynamic resistance, which imparts nonlinear characteristics to each car, posing significant challenges for controller design and stability analysis. Investigating the distributed tracking control problem for underactuated HSTs, where aerodynamic resistance acts on every car, remains a long-standing open problem. The challenge is further compounded when actuator faults are involved. Additionally, accurately obtaining the system parameters for a HST is difficult. To address these challenges, we propose a distributed tracking control approach that does not rely on system parameters, where each motor car uses only its own information, as well as that of the cars in front and behind. In this paper, a new Lyapunov function is innovatively established by incorporating elastic potential energy and relative kinetic energy into its construction. Based on this function, it is rigorously proved that the closed-loop tracking error system remains stable as long as at least one motor car exists, and that the velocity-tracking errors of the motor cars are guaranteed to asymptotically converge to zero. Furthermore, an innovative algorithm is proposed, which effectively reduces the cumulative position-tracking error by adjusting the desired trajectory. Compared with the existing results, the proposed method does not depend on any system parameters, and the resulting closed-loop tracking error system is guaranteed to be stable. Finally, we provide simulations on two HSTs to verify our theoretical results.

This paper presents a novel data-based adaptive control strategy based on feedback linearization to address the asymptotic tracking control problem of position and speed for underactuated high-speed train (HST). The proposed strategy accounts for basic resistances, in-train forces, multiple unknown disturbances, and unknown system parameters. The key contributions of this paper are threefold. First, the proposed strategy eliminates the reliance on exact model parameters by incorporating adaptive mechanism, which is a significant advancement over traditional feedback linearization method. Second, by combining Lyapunov stability theory with a novel output redefinition approach, the stability of the zero dynamics system for underdriven HST is rigorously demonstrated. Third, an improved equivalent control law is introduced, which not only suppresses unknown disturbances automatically but also mitigates severe chattering phenomenon. Simulation on a HST with 2 motor cars and 6 trailer cars is provided for verifying the theoretical results. Simulation results show that the proposed strategy achieves asymptotic tracking of the locomotive to the desired position and speed trajectories as well as ensures the uniformly ultimately bounded stability of the internal dynamics of all trailers.Note to Practitioners—The underactuated high-speed train (HST) system, as a practical engineering system, is characterized by multi-input multi-output, multivariable coupling, and nonlinearity. During the operation of an underactuated HST, the system is inevitably affected by basic resistances, in-train forces, multiple unknown disturbances, and time-varying system parameters. This paper aims to achieve high-precision tracking of the position and speed of underactuated HST under accounting for these factors. The proposed data-based control scheme enhances practicality by applying the feedback linearization combined with adaptive mechanism to deal with the nonlinearity of the underactuated HST. To address the challenges posed by the complex operating environments of HST, adaptive control is introduced. Additionally, an adaptive sliding mode control method is employed to enhance the robustness and fast convergence of the proposed control scheme, while effectively mitigating severe chattering. The simulation results show that the HST can achieve high-precision tracking of the desired position and speed curves in the presence of unknown system parameters and disturbances.

Process uncertainty, which is usually caused by various factors, is generally subject to unknown complex distribution. However, many existing monitoring methods are established with a single distribution, and thus they may not accurately reflect the uncertainty within process systems. In this study, a probabilistic quality- relevant monitoring (PQM-GMM) is proposed with the Gaussian mixture model to address the aforementioned issue. Different from conventional monitoring methods, the proposed method measures the process uncertainty using multiple Gaussian distributions, which can be used to approximate any unknown complex distribution. Then, the optimization problem of the proposed PQM-GMM model is solved using the expectation maximization (EM) algorithm, which includes an augmented Lagrange multiplier in the M-step for model parameter estimation. Using the obtained results, a quality-relevant monitoring model is established with three statistics. It is noted that the proposed model can also be extended to many existing methods since they share a similar structure. Besides, the detailed information such as initial value selection, missing data problem, computation complexity is discussed. The effectiveness and superiority of the proposed method are tested using a numerical simulation example and a real low-pressure heater application. In comparison with some commonly used quality-relevant methods, the proposed model can be robustly established in the presence of corrupted data, and has a better detection sensitivity for the process anomalies in both process and quality variables. Note to Practitioners—A quality-relevant monitoring method is proposed in this study with Gaussian mixture model (GMM) for detecting the abnormal conditions of industrial processes under harsh environment. Since GMM can be used to approximate any unknown complex distribution, the process uncertainty within the collected data can be meticulously measured using the proposed PQM-GMM model. Besides, the quality-independent faults and quality-related faults can also be effectively distinguished using the designed monitoring statistics.

In this study, a novel data-driven discrete-time sliding mode control (DSMC) approach is designed for an electric multiple unit (EMU) velocity tracking control system. First, the input/output (I/O) data of the EMU are equivalently modeled as a full-format dynamic linearized (FFDL) data model to facilitate the generation of a data-driven control scheme. Subsequently, a discrete terminal sliding mode function and a new hyperbolic reaching-law are introduced to simultaneously achieve fast convergence and alleviate chattering. Based on the designed sliding mode function and reaching-law, an improved discrete-time terminal sliding mode control (iDTSMC) approach is derived using the FFDL model. The proposed approach considers the error feedback, parameter estimation errors, and total uncertainties for compensation, to achieve better control performance. The key advantages of this approach include its sole utilization of input-output data from the EMU system, low controller order, robust parameter adaptability, and anti-interference capabilities. After providing the stability analysis of the proposed method, the iDTSMC scheme is compared and tested on a simulated CRH380A high-speed train experimental platform in a laboratory. The simulation results show that the velocity tracking errors of each power unit of the EMU under the proposed control scheme are within$-$0.112 km/h, 0.118 km/h, and the control forces and accelerations are within$-$51 kN, 43 kN and$-$0.952 m/s$^2$, 0.827 m/s$^2$, respectively, with stable fluctuations. Comparative experimental results demonstrate the effectiveness and superiority of the proposed strategy, which remains robust in the presence of disturbances.Note to Practitioners—This study is inspired by the problem of EMU operation control, however, it is also applicable to other systems with trajectory tracking characteristics. Existing automatic train driving control methods are usually based on a dynamics model of the train, which is easily affected by the external environment. In this study, a new data-driven control method is proposed. A discrete terminal sliding mode control function and a new hyperbolic reaching-law were designed based on a dynamic linearized data model equivalent to the EMU operation process. In contrast to similar existing works, the control scheme does not require an accurate EMU dynamic model and is easy to implement. Preliminary simulation experiments show that the method is feasible, however, it has not been incorporated into the train’s physical system, nor has it been tested in production. In future research, we plan to promote a combination of control theory and manufacturing practices.

This paper investigates decentralized tracking control problem for a single high-speed train (HST) with unknown aerodynamic resistance, motor faults and disturbances. First, a multiple point-mass model of the HST system, which reflects in-train coupling force between adjacent cars, is derived. It is assumed that the first car of the HST is a motor car, i.e. the car is capable of traction and braking. Next, under the conditions that the bias faults and the disturbances are norm-bounded, a computationally inexpensive decentralized adaptive control method with backstepping technique is developed. Utilizing the information garnered from the adaptive mechanism, the detrimental effects stemming from the aerodynamic resistance and the motor faults can be comprehensively eradicated in the motor cars. Furthermore, a linear matrix inequality technique is introduced to address the in-train coupling forces for ensuring the tracking stability of the trailer cars. It is shown that the resulting closed-loop system is not only stable, but also that the position tracking error and velocity tracking error of the motor cars can asymptotically converge to zero. Compared with the existing results, a computationally inexpensive decentralized adaptive asymptotic tracking control approach is proposed, which requires only one parameter to be updated online adaptively per motor car. Finally, simulation on a HST with 2 motor cars and 6 trailer cars is provided for verifying the theoretical results.

A novel MIMO data-driven integral predictive sliding mode control (DIPSMC) scheme is proposed based on the dynamic linearization (DL) and state observer method, intended for the automatic driving systems of multi-power unit high-speed trains (HSTs) influenced by system couplings, input constraints, and external disturbances. Initially, by introducing a nonlinear fast integral terminal sliding mode (NFITSM) surface instead of the traditional sliding mode function, facilitating rapid convergence of system errors and reducing sliding mode chattering. Additionally, a parameter update law and an adaptive extended state observer (AESO) are designed to estimate input gains and total uncertainties, respectively, addressing the issue in traditional discrete-time sliding mode control (DSMC) that requires large switching gains to handle disturbances. Subsequently, combining the rolling time domain optimization concepts in predictive control, it follows the reference trajectory of the predefined reaching law, allowing the system to explicitly handle control constraints and obtain higher tracking accuracy. This scheme concurrently accounts for and compensates the total uncertainties arising from system couplings, parameter estimation errors, and unknown disturbances. The principal advantage of this scheme is its design based entirely on a DL data model equivalent to the HST system, characterized by a low-order controller with robust chattering mitigation and disturbance rejection capabilities. Finally, comparative testing experiments of the proposed scheme are conducted on the CRH380A HST simulation platform. Experimental results indicate that under the proposed control scheme, the velocity and displacement error ranges for each power unit of the HST are within ±0.156 km/h and$\pm 1.~1$m, respectively, with control force and acceleration ranges of [−53.2 kN, 46.5 kN] and [−0.568 m/s2, 0.476 m/s2], respectively, and low chattering levels, fulfilling the efficiency and safety requirements of the trains.

Due to the excellent performance of spectral clustering (SC), it has been widely used in many fields of application. However, the high computational complexity and two successive steps have limited SC's development. In addition, the traditional SC is formulated to maximize the arithmetic mean of trace ratios which is dominated by the larger objectives and may reduce the recognition accuracy in practical applications. In this article, we propose a novel graph cut criterion to minimize the trace ratios of harmonic mean with objectives, which can avoid the worst-cluster issue without imposing any regularization or constraints. Furthermore, an efficient and effective coordinate descent (CD) method is exploited to achieve a one-step solution. Therefore, this article can simultaneously solve three main challenges in a unified framework. Extensive experiments verify that the harmonic fast one-step graph cut (HFOC) achieves superior clustering performance with relatively less time-consuming compared to the other state-of-the-art clustering methods.

Emerging communication-based control and virtually coupled train formations are critical foundations for automatic train operations to improve capacity, safety and flexibility toward intelligent rail transportation systems. High-reliability cooperative control of heavy haul train (HHT) formations encounters greater challenges due to the distinctive dynamic characteristics and uncertainties. This paper proposes a communication-based distributed global composite learning cooperative control protocol for virtually coupled heavy haul train formations that enables all HHTs to achieve autonomous dynamic steady coordination with a minimum safe interval and collision avoidance. A pneumatic braking algorithm is first presented for each HHT’s vehicles in the formation to satisfy the distinctive electro-pneumatic braking mode of HHTs and attain improved longitudinal impulses. Then, the complicated dynamic characteristics and uncertainties of HHTs are considered based on a constructed control-oriented dynamic model for HHTs and a neural network (NN) approximation. Additionally, corresponding control protocols are proposed, where the interpretability of the NN approximation and the globally uniform ultimate boundedness property hold. Finally, the control feasibility and abilities of the proposed control protocol are analyzed and demonstrated to be effective via simulation experiments.

Coupler systems are critical to the secure operation of the high-speed train. For this reason, this paper investigates the tracking control problem for the train under coupler displacement constraints. Firstly, by considering the security constraints and multiple disturbances, a multiple point-mass model of the train containing coupler displacements is established. Then, a disturbance observer is introduced to eliminate the adverse effects of complex disturbances on train operation, thereby improving the train’s anti-disturbance capability. With the help of Lyapunov stability analysis, a new upper bound of observer estimation error is developed. Next, an input compensation signal is designed to address the issues of the coupler constraints and the low control accuracy based on disturbance observer. The state deviations of the train system caused by the disturbances can be almost completely compensated by the input compensation controller. Finally, a sufficient condition is provided to ensure that the coupler constraints are not violated throughout the entire train operation. Even if the couplers are stretched or compressed under the initial condition, the compensation signal could still ensure that the couplers work safely. Simulation experiments confirm that the designed compound controller can ensure the coupler always satisfies the constraint condition and effectively suppresses external disturbances, achieving safe and high-precision tracking control of the train.

Owing to its simplicity, the traditional k - means (Lloyd heuristic) clustering method plays a vital role in a variety of machine-learning applications. Disappointingly, the Lloyd heuristic is prone to local minima. In this article, we propose k - mRSR, which converts the sum-of-squared error (SSE) (Lloyd) into a combinatorial optimization problem and incorporates a relaxed trace maximization term and an improved spectral rotation term. The main advantage of k - mRSR is that it only needs to solve the membership matrix instead of computing the cluster centers in each iteration. Furthermore, we present a nonredundant coordinate descent method that brings the discrete solution infinitely close to the scaled partition matrix. Two novel findings from the experiments are that k - mRSR can further decrease (increase) the objective function values of the k - means obtained by Lloyd (CD), while Lloyd (CD) cannot decrease (increase) the objective function obtained by k - mRSR. In addition, the results of extensive experiments on 15 datasets indicate that k - mRSR outperforms both Lloyd and CD in terms of the objective function value and outperforms other state-of-the-art methods in terms of clustering performance.

This article addresses the distributed multiple fault isolation, modeling, and the closed-loop fault estimation under asynchronous switching for high speed train (HST) with switched dynamics, which is composed of traction, coasting, and braking. First, directed-graph-quantum-learning-based multiple-agent system (MAS) classifiers are introduced to characterize the joints effects of multiple faults. Some sufficient conditions are derived under the condition that the multiple fault topology contains a directed spanning tree and cycle edge, and these conditions guarantee that the multiple fault isolation problem can be solved under randomized learning techniques. Then, single-integrator agents are employed to capture the time-varying topology of multiple fault modeling, in which edge agreement and persistence condition are used to guarantee asymptotic consensus. After that, a novel robust fault estimation design along with the switched Lyapunov function and average dwell time is proposed for the possible power actuator faults subject to asynchronous switching and electromagnetic interferences. In addition, switched estimators are designed such that the closed-loop system is asymptotically stable. A multiple fault isolation and estimation case is investigated to validate the application of this methodology.

The clustering method has been widely used in data mining, pattern recognition, and image identification. Fuzzy c-means (FCM) is a soft clustering method that introduces the concept of membership. In this method, the fuzzy membership matrix is obtained by calculating the distance between data points in the original space. However, these methods may yield suboptimal results owing to the influence of redundant features. Moreover, FCM is always sensitive to noise points and heavily subject to outliers. In this article, we propose a method called sparsity FCM clustering with principal component analysis embedding (P_SFCM). We simultaneously conduct principal component analysis and membership learning, and then add an additional weighting factor for each data point. The goal of this operation is to identify the noise or outliers. Overall, the benefit of our framework is that it retains most of the information in the subspace while improving the robustness of the noise. In this article, we employ an iterative optimization algorithm to efficiently solve our model. To verify the reliability of the proposed method, we conduct a convergence analysis, noise robustness analysis, and multicluster experiments. Furthermore, comparative experiments are conducted on both synthetic and real benchmark datasets. The experimental results show that the P_SFCM is competitive with comparable methods.

As one of the classical clustering techniques, spectral embedding boasts extensive applicability across numerous domains. Traditional spectral embedding techniques entail the mapping of graph models to low-dimensional vector spaces (indicator vectors) to facilitate hard partitioning. However, data boundaries occasionally exhibit ambiguity, thereby constraining the utility of hard partitioning. In this article, we introduce an innovative spectral embedding method, namely, joint learning of anchor graph-based fuzzy spectral embedding model and fuzzy K-means (AFSEFK). Drawing inspiration from fuzzy logic, our method employs a membership vector in lieu of the conventional indicator vector for spectral embedding, amalgamating it with fuzzy K-means to concurrently optimize membership, thereby simultaneously learning the local and global structures inherent in the data. Moreover, to enhance the quality of similarity graphs and augment clustering performance, we implement the balanced K-means-based hierarchical K-means technique to generate representative anchors. Subsequently, an anchor-based similarity graph is devised through a parameter-free neighbor assignment strategy. Comprehensive extensive experimentation with synthetic and real-world datasets substantiates the efficacy of the AFSEFK algorithm.

Zeroing neural networks (ZNNs) are effective ways to resolve time-variant problems such as plural Lyapunov equation. In general, the design convergent parameters (DCPs) of ZNN solutions influence their convergence speed. While previous constant-parameter ZNNs (CP-ZNNs) are fixed or constant, it is impractical because the parameters are time-variant in actual hardware situations. On account of this point, the variant parameter and convergent ZNNs (VPC-ZNNs) have been studied in the area sequently. Although these VPC-ZNNs outperform the CP-ZNNs, their DCPs usually keep growing over time and even become infinitely great finally. But infinitely great parameters are unacceptable for us. Meanwhile the computing resources will be wasted because of the growing the parameters even when the VPC-ZNNs become convergent. Based on these, we propose a hyperbolic tangent-type VP-ZNN (HTVP-ZNN) solution owning preassigned-time convergence to resolve time-variant plural Lyapunov equation in this article. HTVP-ZNN can adjust its DCPs, so that they gradually converge to a constant once HTVP-ZNN is convergent under preassigned time. Preassigned-time convergence of HTVP-ZNN and its upper bound are theoretically investigated. Comparative numerical experiments are provided to substantiate the favorable convergence.

In traditional graph embedding methods, graph construction is sensitive to high-dimensional data with noise and outliers, making an effective exploration of the neighborhood structure of the data difficult. Besides, with these methods, constructing graphs and reducing dimensions are disconnected and cannot be mutually optimized. To address these problems, we propose an unsupervised dimensionality reduction method based on bipartite graph, named unsupervised adaptive bipartite graph embedding (UABGE). First, the anchors are generated from the raw data by K-means or random sampling. Second, the bipartite graph, which is constructed between the samples and the anchors in the low-dimensional subspace, utilizes the adaptive allocation method to assign neighbors for each sample, so that the local structure of high-dimensional data can be captured effectively. Third, we present an objective function that combines bipartite graph construction and projection matrix learning to achieve mutual optimization between them, which can be solved with an alternating optimization algorithm. Finally, the computational complexity and the convergence of the algorithm are analyzed. Experimental results on synthetic data and publicly available datasets illustrate the effectiveness of the proposed method.

Graph embedding is a widely used method for dimensionality reduction due to its computational effectiveness. The quality of the graph and the efficiency of graph construction will directly affect the performance and the efficiency of the graph embedding methods. However, in the unsupervised graph embedding methods, the graph is not considered as an optimized graph since there is no label that can be used to construct this graph. In addition, the running of traditional graph embedding methods become very time-consuming on large-scale datasets due to the high computational cost in the step of graph construction. Aiming to solve these problems, we propose an unsupervised dimensionality reduction method based on bipartite graph, called Unsupervised Optimized Bipartite Graph Embedding (UOBGE). Representative anchors are firstly identified in the data. Then, we construct the bipartite graph between the projected samples and the projected anchors and the intrinsic graph connecting all the projected sample pairs with equal weights, which keep the local and global geometric structures of the data, respectively. Finally, the bipartite graph and the projection matrix are optimized simultaneously by introducing an alternating optimization procedure. Extensive experiments on several datasets demonstrate that the effectiveness and efficiency of the proposed method.

The following operation of multiple high-speed trains (HSTs) with high speed and great density based on the moving block and virtual coupling control system can significantly improve the performance of railway systems concerning operation safety and line capacity. However, it is difficult for HST’s conventional control approach to satisfy the service demand, including safety, energy-saving, and ride comfort, due to the significantly increasing longitudinal interaction between adjacent trains in the following operating conditions. In this paper, the characteristic model of multiple HSTs’ following operation was formulated to effectively and accurately describe train control constraints derived from the coupling relationship among adjacent trains’ operation states. Then, the cooperative model predictive control (CMPC) strategy with a multiobjective rolling optimization scheme is developed. The overall line capacity, energy consumption, and ride comfort of train platoons in the railway section are defined as optimization indexes. Furthermore, some sufficient conditions for the consensus stability of the CMPC of HST are given, based on the proposed multiple HSTs following characteristic model and cooperative optimal control strategy. Simulation experiments are carried out by using HST’s operation data collected from real-world runs. Results demonstrate the effectiveness and feasibility of the proposed modeling and control method.

To solve a general time-varying Sylvester equation, a novel integral recurrent neural network (IRNN) is designed and analyzed. This kind of recurrent neural networks is based on an error-integral design equation and does not need training in advance. The IRNN can achieve global convergence performance and strong robustness if odd-monotonically increasing activation functions [i.e., the linear, bipolar-sigmoid, power, or sigmoid-power activation functions (SP-AFs)] are applied. Specifically, if linear or bipolar-sigmoid activation functions are applied, the IRNN possess exponential convergence performance. The IRNN has finite-time convergence property by using power activation function. To obtain faster convergence performance and finite-time convergence property, an SP-AF is designed. Furthermore, by using the discretization method, the discrete IRNN model and its convergence analysis are also presented. Practical application to robot manipulator and computer simulation results with using different activation functions and design parameters have verified the effectiveness, stability, and reliability of the proposed IRNN.

A high-speed electric multiple unit (HSEMU) is a power-distributed train consisting of multiple trailers and motor cars, which is becoming a very handy and powerful transportation tool for our daily life. It is critical to control HSEMU synchronously for improving system performances, but challenging to be achieved due to environmental uncertainties, coupling, and nonlinearity. This paper aims to develop advanced modeling and control techniques to achieve the synchronous tracking control of each car body with harmonic motion. To characterize the dynamics of HSEMU, a multi-adaptive neurofuzzy inference systems (ANFIS) model is built through learning sample data collected from real runs. Then, a model predictive controller is designed to implement the synchronous tracking control of HSEMU on the basis of the multi-ANFIS model. A stability analysis of the closed-loop control system is given. Simulation experiments with comparisons on real HSEMU running data are carried out. Results demonstrate that our proposed modeling and control techniques significantly improve the operational performance of HSEMU in terms of safety, punctuality, comfort, and parking accuracy.

Efficient and safe following operation of a high-speed train (HST) under the moving block system (MBS) has been the trend for high-speed railway (HSR) transportation. However, due to the complex operating environment and changeable operation state of HST during the following operation process, it becomes inapplicable for conventional train control approaches to satisfy HST's multi-objective operation demand including safety, punctuality, energy efficiency, and ride comfort. To resolve this problem, in this paper, practical models for the real-world characteristics of HSR's line parameters and HST's following control, as well as the multi-objective optimization model with some novel metrics, are established for achieving the optimal following control of HST under MBS. Besides, the multi-objective particle swarm optimization algorithm is modified with the preference information of control sensitivity and energy efficiency, so as to efficiently obtain the optimal following control strategy based on the real-time data of HST like velocity, location, and speed restriction. Furthermore, the convergence performance of our proposed method is demonstrated by comparative tests which also help select the proper algorithm parameters. Finally, the efficiency and feasibility of the proposed framework are illustrated by showing some experimental results.

For constructing a span-lateral inhibition neural network (S-LINN) with optimal architecture and parameters for actual application, a self-organizing optimization approach is proposed in this paper to tune the architecture and parameters simultaneously. This self-organization pruning algorithm is to build a modified significant index function to evaluate the significance of hidden neurons. A preprocessing training of the initial network with assumed redundant hidden neurons will be allowed in the tuning process. A subsequent learning after the self-organization pruning process is also implemented to optimize the parameters of pruned network. The proposed self-organizing approach has been tested on two benchmark problems in the areas of breast cancer dataset diagnosis and nonlinear dynamic system identification problem. Simulation results demonstrate that the proposed method has good exploration and exploitation capabilities in terms of searching the optimal structure and parameters for S-LINN. The proposed method also can be used to optimal standard neural network.

Multiobjective operation optimization for high-speed trains (HSTs) is hardly implemented by manual operation due to the increasing operational complexity and environmental uncertainties. In this paper, a T-S fuzzy bilinear model is established based on the nonlinear dynamics of the HST. A new adaptive predictive control approach based on the T-S fuzzy bilinear model of HST control is proposed with consideration of security, punctuality, and energy efficiency. In view of the model's adaptability and the approach's real-time performance, a lazy learning algorithm is used to adjust the parameters of the model and controller online while the model prediction error exceeds a given threshold. Simulation results based on the real HST running data show that the proposed approach contributes a significant improvement in HSTs' tracking precision and energy efficiency.