电火花线切割机床可靠性精确分配方法研究

Researching on Wire Electrical Discharge Machining Reliability Accurate Allocation Method

针对电火花线切割机床可靠性分配中的关键问题和难点,提出了一种基于神经网络、采用逆向思维进行可靠性精确分配的新方法。根据可靠性预计和可靠性分配为互逆过程的思想,利用误差反向传播的改进算法,以电火花线切割机床平均无故障间隔时间和对应的系统可靠度为神经网络输入,以各子系统可靠度相互比值为网络输出,应用实验数据对神经网络进行训练。借用神经网络中权值和阈值的含义反映了各个子系统可靠度变化对系统可靠度的影响,从而确定子系统可靠性分配权重,实现对电火花线切割机床可靠性进行精确分配。

For solving the key problems and puzzles in Wire Electrical Discharge Machining （WEDM） reliability allocation, a reliability allocation method based on the neural network and converse thinking is proposed. Based on the thinking of reliability prediction and allocation is reciprocal processing, and using the improved error backward propagation, the WEDM Mean Time between Failures （MTBF） and the corresponding reliability are taken as the input of neural network; the mutual proportion of every subsystem＇s reliability is taken as the output of neural network. The neural network is trained by sample data. The influence of system reliability under sub-system reliability changing is reflected by the meanings of the neural network weight and threshold. Then the reliability allocation weight of every sub-system can be achieved, the WEDM reliability can he allocated accurately.