A Comparison of Particle Swarm Optimization and Genetic Algorithm Based on Multi-objective Approach for Optimal Composit
This paper proposes an intelligent tuning methods of linear and nonlinear parameters for composite nonlinear feedback (CNF) control using multi objective particle swarm optimization (MOPSO) and multi objective genetic algorithm (MOGA). The main advantage
- PDF / 679,325 Bytes
- 11 Pages / 439.37 x 666.142 pts Page_size
- 74 Downloads / 153 Views
Abstract. This paper proposes an intelligent tuning methods of linear and nonlinear parameters for composite nonlinear feedback (CNF) control using multi objective particle swarm optimization (MOPSO) and multi objective genetic algorithm (MOGA). The main advantage of the methods lies in its efficient fitness/objective evaluation approach of the algorithms such that it can be computed rapidly to obtain an optimal CNF with good system response. In order to yield an efficient technique for fitness evaluation, it is achieved by utilizing a multi objective approach, thus avoiding the use of single objective approach to evaluate the fitness. MATLAB simulations are used to test the effectiveness of the proposed techniques. Nonlinear vehicle model is constructed to validate the controller performance. The model is also simplified to a linear model for designing the CNF. The superiority of the proposed methods over the manual tuning method are improved with 98 percent reduction in error.
1 Introduction Vehicle stability system is crucial to be control precisely especially in a severe cornering maneuver to avoid oversteer or understeer situations. Hence, active front steering system (AFS) has been widely investigated by many researchers for vehicle yaw rate tracking control to achieve a good system response. The implementation of composite nonlinear feedback (CNF) controller for active front steering system (AFS) is significant owing to its benefit mainly in improving the transient performance. In CNF, the optimal tuning parameters namely linear feedback gain F and nonlinear gain parameters ( and c) are desirable to obtain a good system response. The CNF controller was formerly designed by [1]. Linear feedback gain can be designed by ensuring the closed loop system has a small damping ratio in order to achieve a fast output response. From the previous works, the techniques that have been applied are pole placement [2], H2 and H1 [3], LQR method [4] and many more. In the CNF nonlinear part, the designer is required to tune the nonlinear gain parameters. As in [5], they selected the nonlinear gain parameters by using the classical root locus © Springer Science+Business Media Singapore 2016 L. Zhang et al. (Eds.): AsiaSim 2016/SCS AutumnSim 2016, Part I, CCIS 643, pp. 652–662, 2016. DOI: 10.1007/978-981-10-2663-8_67
A Comparison of Particle Swarm Optimization and Genetic Algorithm
653
theory. According to [6], the value of c can be fixed by setting the steady state system with a desired damping ratio that has been initially chosen. The other implementation to optimize the CNF parameters is by using Hooke Jeeves method for controlling the speed of DC motor [6]. Despite of the other tuning methods addressed in the previous works, there appear to be an absence in utilizing an artificial intelligent technique to optimize the parameters. It is a powerful method to save the computational time and ease the complexity in designing the CNF controller. Thus, this paper proposes multi-objective particle swarm optimization (MOPSO) and multi-obj
Data Loading...
