基于概率神经网络的TE过程故障诊断

摘要

Probabilistic neural network (PNN) - an important deformation of radial basis network, its learning speed is very suitable for fault detection problems.But when the network input data is too large, the network will be complex to calculate, and calculation speed will be very slow.This paper uses the principal component analysis (PCA) to reduce the dimensions of the process data, and then treated data as input, so the calculation speed of the network will be enhanced.Finally, the method is used for Tennessee - Eastman process fault diagnosis.The test results show that the method is effective, and easy to achieve.%概率神经网络(PNN)-径向基网络的重要变形,它的学习速发快,很适合于故障检测问题,但是当网络输入样本过大时,网络的计算就会很复杂,计算速度就会很缓慢.本文提出用主元分析(PCA)对过程数据进行降维,然后将处理过的数据作为网络输入,这样使网络的计算速度得到了提高.最后将提出的方法用于田纳西伊斯曼过程(Tennessee-Eastman Process,TE过程)的故障诊断中,测试结果表明该方法行之有效,易于工程实现.