基于概率神经网络的中央空调故障诊断研究

建立中央空调的静态物理模型，通过抽取特征向量残差，建立中央空调故障状态和特征向量之间的映射关系。通过人工引入故障，对某大楼中央空调运行的现场测量，测量结果经处理后输入概率神经网络，经运算后对中央空调进行故障检测与诊断。结果表明，该方法简单可行，运算时间短，能够实现对中央空调的故障检测与诊断。

Investigation of central air-conditioning fault diagnosis based on probabilistic neural network

The physical model about central air conditioning was established under the static conditions. The mapping relationship between fault states and eigenvector about central air conditioning was founded through extraction the residual of eigenvector. The fault detection and diagnosis were implemented on the basis of field experiment test for a real building air conditioning, which was artificially introduced into series faults one by one. The results indicate the probabilistic neural network is practical, rapid and simple for central air conditioning fault detection and diagnosis.