Improving neural control with a PID to have a better response

Neural Networks (NN) can be used with plants as system controllers for the most tracking control systems. One of neural controls is the model reference control in which one NN is used as the identified plant, the other as the Neural Controller (NC) based on the identifier. The response under neural control is not good enough in general, such as poor tracking behavior shown in steady-state error, and poor response under variation of plant´s parameters. With a PID control in parallel with the NC, all drawbacks mentioned above can be alleviated accordingly. By using particle swarm intelligence, parameters of PID can be easily found.