Research Of PID Intelligent Controller Based On FPGA

The process of industrial production is sometimes nonlinear and uncertain, so it is difficult to establish accurate mathematic module, which is hard to achieve perfect controlling effect within using routine PID controller. As an important branch of intelligent control, neural network (NN) has favorable nonlinear mapping performance and higher parallel information processing capability, and it has been the most engaging theory and method for nonlinear system modeling, recognition and control. Meanwhile, the neuron has very strong ability in information synthesis, study memory, self-study, and adaptation, so, it can deal with some processes that are difficult to describe with the model or rule. The neuron combined with PID to be applied to the actual control, makes you can adjust PID parameters online, makes the system has strong anti-interference ability, adaptive capacity and better robustness.Now, the main research of artificial neural network is the research of neural network theory, neural network application and neural network implementation. These three areas are interdependent and mutually reinforcing relationship. This dissertation emphasizes particularly on the neural network implementation of technical research, innovatively uses embedded system development technology of FPGA to achieve the research and design of single neuron PID intelligent controller, and packages it into a dedicated IP core for other control system.At first, single neuron PID controller's principle and PID algorithm are introduced. And secondly, the single neuron PID controller is designed in MATLAB, which based on the given object model in the closed control system to take the perfect system output. Then, the implementation of single neuron intelligent control algorithm based on FPGA is investigated. The VHDL hierarchal design of the controller is carried out, simulated by QuartusII6.1. Two simulation results show that the performance of single neuron intelligent controller based on FPGA is better than the single neuron PID intelligent controller designed by MATLAB.The design mainly includes the weight modified module, error module, weight generating and output module. And each module of the design is optimized, so that the design of this article not only uses less hardware resources, but also has fast speed. Simultaneously, it also improves the control performance of traditional controllers.