The Research And Application Of Neural Network Control In FPGA

Since psychologist McCulloch and mathematician Pitts cooperated and put forwarded the math model of neuron form, that is MP model, it has been more than 60 years for the research of artificial neural network. Now, the research of artificial neural network is divided into three aspects mainly: the research of neural network theory, the research of neural network application and the research of neural network implementation. The relation among the three aspects is as following: the research of theory is the base of the whole research of neural network, which provide gist and direction for the application and implementation of neural network; the application of neural network bring forward new problem for the research of neural network and put forward new demand on the implementation technology of neural network, which also drive the development of neural network theory and realization; the implementation of neural network is to use all kinds of technology, implement all types of neural network model and its computation in reason and effectively, which set up a bridge between the neural network theory and neural network application.This paper emphasizes particularly on the research of neural network's implementation technology. After introduce and compare all the implementation scenario using software and hardware in existence, the way of using FPGA to implement neural network PID controller is expounded in emphasis. At last, the FPGA implementation of neural network PID controller in DVD servo system is introduced and the MATLAB simulation is presented.All the programs in this paper is programmed by Verilog HDL, simulated by Modelsim 5.6d, synthesized by Symplify 7.1 and implemented in Virtex â…¡ of Xilinx.