Yixin Yin

Hamiltonian-Driven Adaptive Dynamic Programming for Continuous Nonlinear Dynamical Systems

This paper presents a Hamiltonian-driven framework of adaptive dynamic programming (ADP) for continuous time nonlinear systems, which consists of evaluation of an admissible control, comparison between two different admissible policies with respect to the corresponding the performance function, and the performance improvement of an admissible control. It is showed that the Hamiltonian can serve as the temporal difference for continuous-time systems. In the Hamiltonian-driven ADP, the critic network is trained to output the value gradient. Then, the inner product between the critic and the system dynamics produces the value derivative. Under some conditions, the minimization of the Hamiltonian functional is equivalent to the value function approximation. An iterative algorithm starting from an arbitrary admissible control is presented for the optimal control approximation with its convergence proof. The implementation is accomplished by a neural network approximation. Two simulation studies demonstrate the effectiveness of Hamiltonian-driven ADP.

An improved PSO for path planning of mobile robots and its parameters discussion

An improved path planning algorithm for mobile robot in known environment is presented in this paper. Firstly, the environment representation method was introduced and a path connecting the start node and the goal node was coded as a particle. Then, a particular “active region” for particles was investigated according to the location of obstacles. The initial particle population was generated within this region and particles flied in the “active region” to search for the optimum path. Simulation studies in simple and complex environment proved that the effectiveness of proposed algorithm. Finally, the parameters discussion is also carried out in order to demonstrate the influence of parameters such as population size, particle dimension, and maximum iteration number.

An Improved ELM Algorithm Based on PCA Technique

This paper proposes a modified ELM algorithm named P-ELM subject to how to select the number of hidden nodes and how to get rid of the multicollinear problem in calculation based on PCA technique. By reducing the dimension of hidden layer output matrix(H) without loss of any information through PCA theory, the proposed P-ELM algorithm can not only ensure the full column rank of newly generated hidden layer output matrix(H′), but also reduce the number of hidden nodes resulting the improvement in training speed. In order to verify the effectiveness of P-ELM algorithm, the comparative simulations are performed. The simulation results illustrate the better generalization performance and the stability of the proposed P-ELM algorithm.

A Novel Improved ELM Algorithm for a Real Industrial Application

It is well known that the feedforward neural networks meet numbers of difficulties in the applications because of its slow learning speed. The extreme learning machine (ELM) is a new single hidden layer feedforward neural network method aiming at improving the training speed. Nowadays ELM algorithm has received wide application with its good generalization performance under fast learning speed. However, there are still several problems needed to be solved in ELM. In this paper, a new improved ELM algorithm named R-ELM is proposed to handle the multicollinear problem appearing in calculation of the ELM algorithm. The proposed algorithm is employed in bearing fault detection using stator current monitoring. Simulative results show that R-ELM algorithm has better stability and generalization performance compared with the original ELM and the other neural network methods.

A Specialized Particle Swarm Optimization for global path planning of mobile robots

A specialized global path planning algorithm for mobile robot based on Guaranteed Convergence Particle Swarm Optimization (GCPSO) is proposed. An environmental map was set up and a path connecting the start node and the goal node was coded as a particle. Then, a particular “active region” for particles was mapped out according to the location of obstacles. The initial particle population was generated within this region and particles flied in the “active region” to search for the optimum path. In the search process, both acceleration coefficients and inertia weight of particle swarm optimization algorithm are self-adaptively adjusted and invalid particles are replaced by global optima or local optima in the adjacent area. The simulation studies in both simple environment and complicated environment are carried out and the simulation results show that the proposed algorithm has advantages such as faster search speed and higher search quality.

On control design and tuning for first order plus time delay plants with significant uncertainties

A novel active disturbance rejection control (ADRC) solution and a particular tuning method are presented for a class of time delay system (TDS) with uncertainty. First, the complicated process dynamics is modeled as a simple first order plus large time delay (FOPTD) plant, with the difference between the actual dynamics and its model treated as disturbances to be rejected. Then the reduced order linear extended state observer (RLESO) with input delay is proposed to estimate the time delay state and disturbance. It is shown how the time delay could be eliminated from the characteristic equation of the closed-loop system by manipulations of controller parameters. Secondly, the one parameter tuning (OPT) technique is developed where all controller parameters are made function of a single coefficient. In comparison with optimal proportional-integral-derivative (PID) controller and twice optimum controller (TOC), the simulation results show that the proposed method not only has better accuracy and faster response, but also ensures better robustness and adaptability against uncertain model parameters and external disturbances, especially for the plant with very large time delays.

Path planning based on fuzzy logic algorithm for mobile robots in static environment

Firstly, the static fuzzy controller is designed and comparative simulation studies with the artificial potential field method show that it has better performance such as shorter path length, less computing time and local minimum and unreachable problems are avoided. Specific obstacle avoidance strategy is proposed for the partially unknown environment and simulation results illustrate its effectiveness. Moreover, improved control rules are developed respectively for u-slot and maze environments by modifying and optimizing some rules of the universal fuzzy controller.

Adaptive Control of Delayed Teleoperation Systems with Parameter Convergence

It is well known that parameter convergence in adaptive control can bring about an improvement of system performance, including accurate online identification, exponential tracking, and robust adaptation without parameter drift. However, strong persistent-excitation (PE) or sufficient-excitement (SE) conditions should be satisfied to guarantee parameter convergence in the classical adaptive control. This paper proposes a novel adaptive control to guarantee parameter convergence without PE and SE conditions for nonlinear teleoperation systems with dynamic uncertainties and time-varying communication delays. The stability criterion of the closed-loop teleoperation system is given in terms of linear matrix inequalities. The effectiveness of this approach is illustrated by simulation studies, where both master and slave are assumed to be two-link manipulators with full nonlinear system dynamics.

Prediction of the hot metal silicon content in blast furnace based on extreme learning machine

Silicon content in hot metal is an important indicator for the thermal condition inside the blast furnace in the iron-making process. The operators often refer the silicon content and its change trend for the guidance of next production. In this paper, we establish the neural network model for the prediction of silicon content in hot metal based on extreme learning machine (ELM) algorithm. Considering the imbalanced operating data, weighted ELM (W-ELM) algorithm is employed to make prediction for the change trend of silicon content. The outliers hidden in the real production data often tend to undermine the accuracy of prediction model. First, an outlier detection method based on W-ELM model is proposed from a statistical view. Then we modified the ordinary ELM and W-ELM algorithms in order to reduce the interference of outliers, and proposed two enhanced ELM frameworks respectively for regression and classification applications. In the simulation part, the real operating data is employed to verify the better performance of the proposed algorithm.

Leader–Follower Output Synchronization of Linear Heterogeneous Systems With Active Leader Using Reinforcement Learning

This paper develops optimal control protocols for the distributed output synchronization problem of leader–follower multiagent systems with an active leader. Agents are assumed to be heterogeneous with different dynamics and dimensions. The desired trajectory is assumed to be preplanned and is generated by the leader. Other follower agents autonomously synchronize to the leader by interacting with each other using a communication network. The leader is assumed to be active in the sense that it has a nonzero control input so that it can act independently and update its control to keep the followers away from possible danger. A distributed observer is first designed to estimate the leader’s state and generate the reference signal for each follower. Then, the output synchronization of leader–follower systems with an active leader is formulated as a distributed optimal tracking problem, and inhomogeneous algebraic Riccati equations (AREs) are derived to solve it. The resulting distributed optimal control protocols not only minimize the steady-state error but also optimize the transient response of the agents. An off-policy reinforcement learning algorithm is developed to solve the inhomogeneous AREs online in real time and without requiring any knowledge of the agents’ dynamics. Finally, two simulation examples are conducted to illustrate the effectiveness of the proposed algorithm.