Abstract:

It can be an important vehicle to investigate surface roughness of the parts caused by the electrochemical processing and working current, handling backlash, electrolyte temperature, machining time, electrolyte ratio etc. These relevant processing parameters presented a non-linear relationship according to the parts roughness characteristics. Due to this relevance mechanism was still unexplored in the current electrochemical machining method, this paper investigated the relationship of previous processing parameters and the roughness characteristics in a mathematics way. Accordingly, a mathematical model was set up for expected BPNN training study. Through the experimental  data acquisition and training operations, it shows that in this way a minor prediction error is produced. In this case, utilizing the orthogonal analysis method, the surface characteristics with large changement amplitude can be realized with the optimized processing parameters.

Key words: electrochemical machining, surface roughness, neural network(NN), prediction

摘要：

电化学加工的表面粗糙度与加工电流、加工间隙、电解液温度、加工时间、电解液配比等工艺参数密切相关，而这些工艺参数与工件表面粗糙度之间为复杂的非线性关系，建立其关联一直是电化学加工中的难题。以BP神经网络为基本工具，建立了加工参数与表面粗糙度之间关系的数学模型，利用实验数据训练网络，结果表明可实现较小的预测误差;应用正交法分析实验数据，实现了可使表面粗糙度参数变化幅度较大的加工参数的优化配置。

关键词: 电化学加工, 表面粗糙度, 神经网络, 预测