逆变点焊电源模糊神经网络自适应控制模型研究

Study of Fuzzy Neural Network Self-Adaptive Control Model for Spot Welding Inverter

研究了逆变点焊电源恒流自适应控制的模糊神经网络模型,设计了模糊神经网络结构。利用BP算法,采用先正弦函数输入后恒定输入的方法对网络进行了分段训练,并使用MATLAB语言,对系统进行了自适应控制和比例因子影响的仿真分析。结果表明,逆变点焊电源恒流自适应控制模糊神经网络,能够实现在线调整隶属函数参数,控制系统可快速感知外来干扰和过程变化,平均控制相对误差小于2.08%;比例因子的选取,对系统有很大影响,不同的比例因子在与训练好的网络结合进行控制时,系统控制效果不同。

In this paper, a fuzzy neural network model for the constant current self-adaptive control system of a spot welding inverter is studied, and the fuzzy neural network architecture is built. The network is trained step by step by using back propagation learning algorithm, the first input is a sinusoidal function and the later one is a constant. The self adaptive control of the system and the effect of the scale factors on the system are simulated and analyzed by using the MATLAB. The results show that the constant current self-adaptive fuzzy neural network control system of the spot welding inverter can adjust the membership function parameters on-line in accordance with the disturbances and process variations, the average control error is less than 2.08%. The scale factors have a great influence on the control system, and the trained network working with different factors has different control effect.