摘要
文章介绍了一种基于神经网络系统(PNN)的传感器故障诊断改进方法,用于对建筑电气故障进行预测分类,该方法采用贝叶斯分类决策理论,以高斯函数作为激励函数。通过仿真系统模型试验,该方法的诊断准确率为90%,在故障识别方面具有良好的诊断效果。
This paper introduces an improved method of sensor fault diagnosis based on the system of probabilistic neural net-work (PNN) ,which is used to predict and classify the building electrical faults and in which the Bayesian classification decision the-ory is adopted,with Gaussian function as the incentive function. Through simulation system model test,the diagnostic accuracy ofthis method is 90%,which has a good diagnosis effect in fault identification.
出处
《科技创新与应用》
2018年第8期11-13,共3页
Technology Innovation and Application
关键词
概率神经网络(PNN)
故障诊断
建筑电气
probabilistic neural network(PNN)
fault diagnosis
building electricity