摘要
详细阐述了小波神经网络(WNN)的原理、结构,并对传统的BP算法进行了改进。以空调系统传感器故障检测问题为目标,提出了基于WNN的故障诊断方法。通过采集天津博物馆中的传感器数据,对训练好的WNN进行了传感器故障诊断能力的验证,对温度传感器的1℃偏差故障、0.05℃/s速率漂移故障、完全故障、与不同方差下的精度等级下降故障进行了仿真,结果表明:这种方法对传感器故障具有很好的诊断效果。
The principle and structure of wavelet neural network (WNN) is described. Focusing on airconditioning system' s sensor fault diagnosis, a new method based on WNN is proposed. Using the actual running data from Tianjin museum building control system, the trained WNN is proved efficient enough to detect the bias of 1℃, drift of 0.05 ℃/s , complete faults and various precision degree decrease of sensors in the air-conditioning system.
出处
《传感器与微系统》
CSCD
北大核心
2007年第4期54-57,共4页
Transducer and Microsystem Technologies
关键词
空调系统
故障诊断
小波神经网络
BP算法
air-conditioning system
fault diagnosis
wavelet neural network (WNN)
BP algorithm