基于概率神经网络的控制图模式识别

Control Chart Pattern Recognition Based on PNN

近年来,人工神经网络被广泛应用于复杂过程质量异常的监控中。文献表明人工神经网络方法存在结构选择困难的问题,其解决主要通过研究人员的经验,耗时多且识别率低。本文提出使用概率神经网络来识别六类典型控制图模式,以改进神经网络识别器的设计效率。研究和仿真试验结果表明,概率神经网络不仅拓扑结构设计简单,而且识别率高。

In the recent years, artificial neural networks （ANN） have been widely used to monitor and control the process abnormal pattern occurred especially in the complex manufacturing quality control. But literatures show that this kind of work is primarily addressed according to the developer＇s personal experiences which would cost too much time with lower recognition accuracy. This paper proposes to use probability neural network （PNN） to recognize the six typical kinds of control chart patterns of the process to improve the design efficiency of the NN pattern recognizer. Research and numerical simulation result shows that PNN has not only the feature of simpler topology structure but also the higher pattern recognition accuracy.