恒切削力加工过程的神经网络预测控制

Neural Network Predictive Control of Constant Force in Cutting Process

切削过程恒力控制对于提高生产率、保证加工精度具有重要作用。文章以车削加工过程为对象,研究恒切削力加工过程中当切削深度发生突变时,如何减少系统输出切削力超调的问题。将预测控制策略与神经网络理论相结合,在神经网络学习时,使切削进给在切削深度发生突变前提前发生相应变化,提出了恒力切削过程的神经网络预测控制算法。仿真结果表明,与传统自适应神经网络控制相比,加工过程的神经网络预测控制能够有效的解决在背吃刀量发生突变时,加工系统输出的切削力过大的问题,显示出比加工过程传统神经网络自适应控制更好的综合性能。

Constant force control in cutting process is of importance for improving productivity and ensu- ring precision. The turning machining process is studied in this paper. The focus is on the reduction of o- vershot of cutting force output when sudden variants in the depth of cut occur in machining process of constant force control. The predictive control strategy and neural network are combined. The cutting fed- erate is changed relevantly before the sudden cutting depth variation in neural network learning. And the neural network predictive control algorithm in machining process of constant force control is proposed. The result of computer simulation shows that the problem of excessive cutting force output with sudden variants in back cutting depth can be effectively solved. Compared with self-adaptive neural network con- trol, the neural network predictive control behaves with better comprehensive nronertie~