Neural Network Intelligent Controller Based On Particle Swarm Optimization

In order to obtain higher efficiency and better performance in complex nonlinear systems, neural network and particle swarm optimization algorithms are introduced. Because of the nonlinear characteristic of neural network and strong optimization ability of particle swarm algorithm, they have become one of the most important methods to solve complex control problems.To avoid the difficulties in tuning the parameters of the traditional PID controller, the paper studies the principle of BP neural network self-tuning PID controller. For the static functionality of BP neural network, the difficulties in choosing and other shortcomings, the paper introduces a new neural network-PID neural network. From an example of non-linear single-variable simulation, it is demonstrated that a faster convergence and a smaller systematic error are obtained and the anti-jamming performance is better at the same time.Because of the time-consuming of the PSO algorithm training, weak capability of global search, probability of falling into local optimum, this paper presents an improved particle swarm optimization algorithm (MPSO). After testing the improved particle swarm algorithm through standard test bench, the simulation results show that MPSO algorithm has higher search accuracy and faster speed of convergence.In order to control complex nonlinear multivariable coupling system, the paper selects MPIDNN network. Simulation results show that MPIDNN can achieve decoupling control and has a good anti-jamming performance. To avoid the random preset of the weight value, the paper proposes a PID neural network based on MPSO. To begin with, we utilize the particle swarm algorithm to optimize the initial weight value of PID neural network. Next, we successfully implement the control strategy of a nonlinear coupling system using the improved PID neural network, and subsequently compare the results with the method based on the BP algorithm and PSO algorithm optimizing MPIDNN parameters. The comparison indicates that the improved PID neural network based on MPSO outperforms the original PID neural network in the time consumption of achieving control strategy and the speed of response. Therefore, MPIDNN based on MPSO has a better control effect.