Application of Particle Swarm Algorithm to Optimization of PID Neural Network

The original weight values of PM neural network are usually determined randomly and tend to sink into the local optimization in their learning process. To overcome the deficiency of PIP neural network, the paper employs the particle swarm algorithm to optimize the PID neural network. To begin with, the particle swarm algorithm is used to acquire the optimized weight values of PID neural network. Next, by using the optimized weight values, we can optimize the PM neural network. Additionally, the performance of the improved PID neural network is assessed using a nonlinear coupling system. The simulation shows that the improved PID neural network effectively relieves the deficiency of the original PID neural network and has some obvious advantages in the calculation accuracy and convergence speed over the original PID neural network.