基于信息优化的加工过程神经网络控制

为了提高加工过程生产率和保证加工精度,以加工过程的恒切削力控制作为研究对象,将信息论原理和神经网络智能控制理论应用于加工过程控制,以信息熵作为加工过程智能控制系统的性能测度能统一各级性能指标,将神经网络作为加工过程控制输入和系统输出的信息传输通道,确定了神经网络基于信息优化的目标函数,推导出了信息优化的三层BP神经网络学习算法,提出了恒力切削过程中基于信息优化的神经网络控制系统框架.通过加工过程的仿真实例证明,与传统自适应神经网络控制方法相比,基于信息优化的神经网络控制方法收敛精确,速度快,振荡小,系统超调量小,具有较好的综合性能.研究结果为信息理论应用于加工过程控制提供了有效途径.

Information-optimization based neural network of force control in cutting process

In order to improve productivity and ensure precision of machining process,information principle and neural network were applied in control of machining process with the constant force control as study object.The neural network was used as information-transmission channel for inputs and outputs of machining process control with information entropy as unified performance indexes at various levels of the intelligent control system.The objective function of neural network based on information-optimization was confirmed and learning algorithm of the three-layer BP neural network was derived.The information-optimization based neural network control system of constant force in machining process was proposed.Simulation instance of machining process proves that,compared with self-adaptive neural network control,the proposed control is of faster and more accurate convergence,less vibration and less overshoot,and behaves with better comprehensive properties.The research result provides effective way for information theory to he applied in control process of machining.