基于双隐层L-M算法的BP神经网络机械加工误差预测控制模型

摘要

Analyzed error transmission and formation, and constructed double hidden L-M algorithm BP neural network predictive control model according to multiple input and output. The model is used to predict and control total amount of feed,first and second amount of feed by means of the rigidity of process system,workpiece hardness,before and after processing the radial error. Experimental and simulation shows that the model can guide the production, optimize the machining process, improve products quality. Finally, software system of error predictive control with LABVIEW and MATLAB realize predictive control visualization.%分析了误差的来源和传递方式,针对机械加工过程高度非线性、多输入和多输出的特点,构造了双隐层L-M算法BP神经网络误差预测控制模型.根据工艺系统刚度、工件硬度、加工前、后径向误差来预测控制刀具径向总进刀量、第一、第二次刀具径向进刀量,实验和仿真结果表明该模型能指导生产、优化加工工艺和提高产品质量.最后,采用LAB-VIEW软件和MATLAB软件编制了误差预测控制系统,实现了预测控制的可视化.