Implementation of neural network based controller using Verilog

Abstract

A simple and robust real time controller that works very well for linear systems with optimal gain tunings is the PID controller. But, PID controllers do not work properly if plant dynamics are changing fast or when the plant is highly nonlinear. However, many of the industries still rely on it. Hence, in most of the plants an auxiliary controller coexists to help the primary PID controller to work better by compensating for uncertainties present during control operation. Neural network has proven to be a good candidate as this auxiliary nonlinear controller. Neural network can effectively compensate for unknown uncertainties and also act as an intelligent control. The success of the neural network as an auxiliary controller has been reported in practical applications such as motion control system, signal processing and controlling robot manipulators. Nowadays, parallel programmable logic devices, such as the field programmable gate array (FPGA), have become powerful hardware options, offering low cost, high execution speed, reconfigurability and parallelism. This work intends to exploit the current available resources in commercial FPGAs to implement servo control for hard disk drive system. Simulation and experimental results included in this paper show the viability of exploiting the parallelism and modularity of a Virtex 6 FPGA to implement a high sampling rate neural network-RBF based controller. This control system platform will allow fast prototyping of new control concepts and evaluation of non-linear control.