Design Of Four-Switch Buck-Boost Power Supply Based On BP Neural Network PID Control

The PID controller is independent of the mathematical model of the controlled object,which has characteristics of stability,easy implementation and strong antiinterference ability,etc.It is the most widely used and mature controller in the power conversion system.However,as the controlled object becomes more and more complicated,the traditional PID control is difficult to achieve the ideal control effect for the four-switch buck-boost converter with strong nonlinearity.BP neural network has the ability to approximate any nonlinear functions,which provides an method for improving the performance of power conversion system based on traditional PID control.This thesis analyzed the shortages of traditional PID control based on the multimode characteristics of four-switch buck-boost power supply,studied the principle of neural network according to its application in industrial process control.And then,a parameter adaptive PID controller is designed which can satisfy the multi-mode operation of the four-switch buck-boost power supply.On this basis,a four-switch buck-boost power supply prototype with DSP as the core is designed.Thus,the main research contents are described as follows:1.Through the use of the fact that the forward voltage drop of FET is almost proportional to the current,and the combination of advantages of asynchronous four tube boost-boost converter,this thesis studies the working principle of four-switch boost-boost converter.Meanwhile,according to the design requirements of the prototype,the basic composition and principle of digital DC-DC converter are studied,and the existing problems of digital control are analyzed.2.In view of the limitations and advantages of the PID controller for the control of non-linear systems,the combination method of neural network and PID control is analyzed,the deficiency of traditional BP neural network PID control is studied,and its optimization method is studied from two aspects of structure and learning algorithm.Based on these,a mathematical model of the converter is established by using the switching element averaging method,the control compensation design is completed with the mathematical model,and the training samples are extracted from experimental debugging.The network training and control system simulation are completed in the Matlab/Simulink environment,and the BP neural network parameter adaptive PID controller for the four-switch buck-boost power supply is designed.3.The design of the prototype of four-switch buck-boost power supply includes hardware and software.The hardware is mainly the peripheral circuit of the DSP,which provides the basis for the normal work of the software.The software includes interrupt service program,dual duty control algorithm implementation,etc.It can be used to complete the acquisition and storage of the controlled quantity,the calculation of the controlled quantity according to the control strategy,and other tasks of updating the output of the new control quantity.Through the study of neural network PID control and four-switch buck-boost converter,the prototype of four-switch buck-boost power supply is designed on the basis of DSP.The test environment is built to test the prototype for verifying the correctness of the dual duty control algorithm.In addition,the power supply load transient,modal switching,efficiency and other power performance aspects are tested one by one.Among them,the maximum efficiency of the prototype is 96% when measured at full load,and the efficiency can be maintained above 90% when the load is more than 2A,which proves the feasibility of the control scheme in the thesis.