Model Reduction Based On Improved Particle Swarm Optimization

With the rapid development of modern society and the increasing complication of industrial processes, the scale of controlled object appeared in the practical application is also growing, which leads to the increasing of the system model order, and causes the increasing of the computational difficulty and controlling costs. Therefore, model reduction theory has been a hot research field, although many domestic and foreign researchers have made a lot of contribution in this field in recent years, namely they reduce the high-order object which is difficult to be controlled in the actual process. Using a simple low-order model to replace the actual higher-order object can reduce the difficulty of designing a controller and improve the effectiveness and precision of control. However in the basis as well as we can go further to explore the area.On the basis of previous work, this paper does further research on model order reduction combined with large systems of higher order model and the improved particle swarm optimization algorithm. The main contents of the study include the following two aspects: on the model reduction problem, in this paper, a number of high order system models are selected as the typical low order model respectively. It is beneficial to the implementation of the project and ensures the stability of the system. On the improved particle swarm optimization algorithm problem, this paper proposed a model reduction method based on adaptive immune binary particle swarm optimization algorithm(AIBPSO) by studying the artificial immune and particle swarm algorithm, at the same time, the error parameters of the reduced-order system and the original system are analyzed and the simulation output curves of the two systems in the step response are analyzed. On the one hand, the paper compared the general model reduction methods(such as Padé approximation) and particle swarm optimization algorithm based on model reduction methods in both time domain and frequency domain. On the other hand, the model order reduction method based on the basic particle swarm optimization algorithm and the AIBPSO based model reduction method are compared from the fitness value curve under different performance comparison. The results show that the proposed method based on the AIBPSO model has better effect on the higher order system model.