Junpeng Li

Dynamic multi-swarm particle swarm optimizer with cooperative learning strategy

Graphical abstractDisplay Omitted HighlightsA new cooperative learning strategy is hybridized with DMS-PSO.Information can be exchanged among sub-swarms before the regrouping process.Experimental results show that DMS-PSO-CLS has a superior performance. In this article, the dynamic multi-swarm particle swarm optimizer (DMS-PSO) and a new cooperative learning strategy (CLS) are hybridized to obtain DMS-PSO-CLS. DMS-PSO is a recently developed multi-swarm optimization algorithm and has strong exploration ability for the use of a novel randomly regrouping schedule. However, the frequently regrouping operation of DMS-PSO results in the deficiency of the exploitation ability. In order to achieve a good balance between the exploration and exploitation abilities, the cooperative learning strategy is hybridized to DMS-PSO, which makes information be used more effectively to generate better quality solutions. In the proposed strategy, for each sub-swarm, each dimension of the two worst particles learns from the better particle of two randomly selected sub-swarms using tournament selection strategy, so that particles can have more excellent exemplars to learn and can find the global optimum more easily. Experiments are conducted on some well-known benchmarks and the results show that DMS-PSO-CLS has a superior performance in comparison with DMS-PSO and several other popular PSO variants.

An improved krill herd algorithm

A deficiency of KH is analyzed.Why KH cannot achieve the excellent balance between exploration and exploitation in optimization processing is explained.To overcome the defect, an improved KH with linear decreasing step (KHLD) is proposed. Krill herd (KH) inspired by the herding behavior of the krill individuals is a new swarm intelligent algorithm which is proved to perform better than other swarm intelligent algorithms. However, there are some weak points yet. In this paper, we analyze a deficiency of KH which cannot achieve the excellent balance between exploration and exploitation in optimization processing and proposed an improved KH-krill herd with linear decreasing step (KHLD). Twenty benchmark functions are used to verify the effectiveness of these improvements and it is illustrated that, in most cases, the performance of KHLD is superior to the standard KH.

Stochastic gradient with changing forgetting factor-based parameter identification for Wiener systems

Abstract The parameter estimation problem is considered for a class Wiener systems. First, the effect of the forgetting factor on the stochastic gradient algorithm is analyzed. Then, a Wiener system stochastic gradient with a changing forgetting factor algorithm is presented which makes full use of the forgetting factor. Finally, an example is provided to test and verify the effectiveness of the proposed algorithms.

Finite-time output-feedback synchronization control for bilateral teleoperation system via neural networks

The finite-time control problem is considered for bilateral teleoperation system via output feedback approach. A new observer is designed for the velocity estimation and the resulting velocity error system is proved to be semi-globally stable. The observer based output feedback finite-time controller is developed by employing a novel nonsingular fast integral terminal sliding mode. The closed-loop system is proved to be stable based on Lyapunov stability theory. It is shown that the master-slave synchronization error converges to zero in finite time. The merit of the proposed method is that the designed controller only uses the position information which renders that the master-slave synchronization error reaches zero in the prescribed time. Simulation and experiment are performed and the results demonstrate the effectiveness of the proposed method.

A fast training algorithm for extreme learning machine based on matrix decomposition

Extreme Learning Machine (ELM), a competitive machine learning technique for single-hidden-layer feedforward neural networks (SLFNNs), has proven to be efficient and effective algorithm for regression and classification problems. However, traditional ELM involves a large number of hidden nodes for complex real world regression and classification problems which increasing the computation burden. In this paper, a decomposition based fast ELM (DFELM) algorithm is proposed to effectively reduce the computational burden for large number of hidden nodes condition. Compared with ELM algorithm, DFELM algorithm has faster training time with a large number of hidden nodes maintaining the same accuracy performance. Experiment on three regression problems, six classification problems and a complex blast furnace modeling problem are carried out to verify the performance of DFELM algorithm. Moreover, the decomposition method can be extended to other modified ELM algorithms to further reduce the training time.

Fixed-time Coordination Control for Bilateral Telerobotics System with Asymmetric Time-varying Delays

Fixed-time coordination in dynamical systems means system trajectories converge to the desired trajectories in determined time which is independent of the system initial states. In this paper, a novel fixed-time coordination control approach for nonlinear telerobotics system with asymmetric time-varying delays is proposed to provide faster convergence rate and higher convergence precision. The neural networks (NNs) and the parameter adaptive method are combined to approximate the uncertain model of the teleoperator, the upper bound of the NNs estimation errors and the external disturbances. Then the corresponding adaptive NNs fixed-time controller is designed without using the derivatives of the time-varying delays. Dynamic surface control (DSC) is employed to avoid the singularity. Moreover, considering the nonpassive human operator and remote environment insert forces, the stability criterion for the closed-loop system is also developed. Then by choosing proper Lyapunov functions, the master-slave coordination errors converging into a deterministic domain in fixed-time with the new controller is proved in the presence of the exogenous forces from human operator and remote environment. Furthermore, the exact convergence time is presented only with the designed parameters. Some comparisons are conducted in simulation to show the superior performance of the proposed control approach. Finally, experimental results are also given to demonstrate the effectiveness of the new control method.

Robust adaptive uniform exact tracking control for uncertain Euler–Lagrange system

ABSTRACT This paper offers a solution to the robust adaptive uniform exact tracking control for uncertain nonlinear Euler–Lagrange (EL) system. An adaptive finite-time tracking control algorithm is designed by proposing a novel nonsingular integral terminal sliding-mode surface. Moreover, a new adaptive parameter tuning law is also developed by making good use of the system tracking errors and the adaptive parameter estimation errors. Thus, both the trajectory tracking and the parameter estimation can be achieved in a guaranteed time adjusted arbitrarily based on practical demands, simultaneously. Additionally, the control result for the EL system proposed in this paper can be extended to high-order nonlinear systems easily. Finally, a test-bed 2-DOF robot arm is set-up to demonstrate the performance of the new control algorithm.

Silicon content prediction and industrial analysis on blast furnace using support vector regression combined with clustering algorithms

Silicon content prediction is quite significant for supervising the state of blast furnace and is usually selected as the indicator to represent the thermal state. In practical industry, the fluctuation exists in the operation of blast furnace all the time. What’s worse, it is inaccurate to build the predictive model with many outliers. To solve these problems, this paper has developed a model to predict the silicon content using support vector regression (SVR) combined with clustering algorithms, including hard C-means (HCM) clustering and fuzzy C-means (FCM) clustering. Through data processing, the data points are clustered based on the similarity, and then different SVR models are established. In order to make full use of FCM, a new method using multiple SVRs and FCM based on membership degree (MFCM-SVRs) is proposed where the membership degree is applied to eliminate the outliers. Simulation results verify that the multiple SVRs based on HCM (HCM-SVRs) and MFCM-SVRs possess superiority in terms of accuracy and speed, which makes the method serve better for practical production.

Security analysis and improvement of a user-friendly remote authentication protocol

Remote user authentication is very important in a distributed computer environment. Recently Wu and Chieu devised a user-friendly remote authentication protocol using smart card. This article shows that their protocol is vulnerable to the forged login attack. Then a simple improvement is suggested to eliminate the vulnerability.

Stochastic gradient based iterative identification algorithm for a class of dual-rate Wiener systems

Parameter estimation problem is considered for a class of dual-rate Wiener systems whose input-output data are measured by two different sampling rate. Firstly, a polynomial transformation technique is used to derive a mathematical model for such dual-rate Wiener systems. Then, directly based on the dual-rate sampled data, a dual-rate Wiener systems stochastic gradient algorithm (DRW-SG) is presented. In order to improve the algorithm convergence rate, a dual-rate Wiener systems stochastic gradient algorithm with a forgetting factor algorithm (DRW-FF-SG) is presented. For making full use of the forgetting factor, a dual-rate Wiener systems stochastic gradient algorithm with an increasing forgetting factor algorithm (DRW-IFF-SG) is presented which performs excellently. Finally, an example is provided to test and illustrate the proposed algorithms.