Research Of Neural Network Intelligent Controller Based On FPGA

Intelligent control theory is a very complex new subject, that is one of the frontiers of automation. The development of intelligent control has opened up a new avenue to solve the control of the uncertainty and the complex nonlinear systems. With the high-speed development of the microprocessor technology and EDA (Electronic Design Automation) technology a new PLC gradually emerge in the current field of industrial automation.The new PLC is based on the EDA technologies and more and more people of all ages in a variety of motion control field. That FPGA (Field Programmable Gate Array) application technology is a research focus currently. Therefore, the study of neural network intelligent algorithms based on FPGA is of great practical significance. This article first introduces the characteristics of neural networks, elements and advantages of FPGA technology and on this basis, developed a FPGA-based neural network PID controller, which is a FPGA-based embedded control system.This paper introduces a structure of the new PLC's composition based on ARM + FPGA structure and then focuses on a intelligent control algorithm module in the new PLC, namely the use of FPGA's Nios II soft core processor to achieve the artificial neural network intelligent control algorithm so that PLC is able to call in the intelligent control module in the run and achieves the intelligent of PLC controller.The design uses a reconfigurable FPGA technology and starts with the three layers error back propagation neural network as a typical model. In the design according as mature BP algorithm formula and using design way of text file input, logic functions is correct by validating. The method provides a reliable basis for neural network hardware implementation.The hardware mainly uses FPGA chip of the Alter company's Cyclone series as the core processor. the external circuit includes the heating and testing devices, A/D converter module and voltage amplifier.Software mainly includes BP neural network PID control algorithm, A/D acquisition module and the display output module. Thereinto the BP neural network PID control algorithm is embedded in the FPGA. The development tool includes integrated development platform Quartus II, Nios II and SOPC Builder.The article focuses on the following content: the error back-propagation of neural network(BP);the realization of neural network PID intelligent controller;input module's realization; the design and realization of the voltage signal puts the doltish;nios II develoment tools'use means; the serial asynchronous communication in FPGA, including double-byte serial asynchronous communication transceiver's realization, the baud rate generator, data receiver, data generator; the realization of temperature regulation and control with the temperature control device, including the PID parameter tuning and the neural network PID control algorithm's simulation.Finally the system's software and hardware are debugged.Furthermore the experiment test is made based on this. The simulation result indicates that this system performance is stable,the movement is reliable and the system has certain reference value.