A single-neuron PID adaptive multicontroller scheme based on RBFNN

In order to improve the control performance of the multicontroller proposed by Guo and Jutan (Canadian Journal of Chemical Engineering, 79, 817-22, 2001), a single-neuron PID multicontroller scheme based on a radial base function neural network (RBFNN) is proposed in this paper. This scheme has four controllers, specifically a set-point controller, two load controllers and a proportional controller. These controllers may be designed independently to achieve good control performance for both set-point tracking and load rejection. In particular, the set-point controller and the load controller have been chosen as single-neuron PID controllers. The model parameters and the parameters of the two single-neuron PID controller are updated in real time. For simplicity, the feedforward controller can be chosen as a unity gain proportional controller. It guarantees physical realizability and provides complete compensation for measurable disturbance. The simulation results show that the single-neuron PID adaptive multicontroller scheme based on RBFNN is very effective and the controller is of relatively strong robustness.