基于模糊神经网络的逆变点焊电源恒电流控制设计及仿真

推导了逆变点焊过程控制模型,并构建了逆变点焊模糊神经网络恒电流控制系统结构.根据该模型采用先正弦后恒定输入的方法对模糊神经网络(FNN)进行分段离线学习,提高网络的泛化能力和自适应能力.在线控制时,利用训练后的网络仅做正向模糊计算,输出逆变桥开关管占空比改变量的方法保证逆变器恒电流输出.最后使用MATLAB高级语言编程,完成了整个系统的仿真实验.仿真结果表明:分段训练后的FNN使用该方法可以实现逆变点焊电源的恒电流控制.

Design and simulating of inverter spot-welding power supply with constant current based on fuzzy neural network

A controlling model of inverter spot-welding process and a fuzzy neural network configuration about inverter spot-welding with constant current control were built in this paper.This fuzzy neural network was trained by off-line method to enhance the generalization and self-adaption ability with the way that sinusoidal trace input firstly,and then constant input.A method that just does some ositive-going calculations to input modifications of the duty cycle was utilized to achieve constant current output of the inverter,when this trained network was used in on-fine control.In the end.the emulator experimentation of the whole system was finished with MATLAB. The simulation results show the constant-current of inverter spot-welding power supply can be actualized by this method.