Research Of Sensor Fault Diagnosis Methods In Air-conditioning Unit Based On Wavelet Neural Networks

Sensor fault is one of the typical faults in Air-conditioning system. The fault sensor's signal will make the control system to do a wrong decision-making, thus causing the wasting of energy and descend of indoor air quality. Therefore, the accuracy and reliability of sensor is very important for control and monitoring of Air-conditioning system. The research of sensor fault diagnosis is very necessary. This paper mainly focuses on Air-conditioning system's sensor fault diagnosis to carry on the research. The main results include:1. Reading great deal of documents of domestic and international, and summarized the existent methods of Air-conditioning system's sensor fault diagnosis based on it. The advantages and disadvantages of those methods are also expatiated. Aiming at the shortages of the diagnosis method using neural network, the method of fault diagnosis based on wavelet neural network is presented.2. Deeply analyzing the advantages of wavelet neural network by comparing with forward neural network, deducing the wavelet neural network's training algorithm and making improvement for traditional BP algorithm from two aspects.3. Making research on the type of Air-conditioning system sensor, and mainly discussing the fault of precision grade decrease.4. Collecting the actual running data from Tianjin museum building control system, analyzing and pre-disposing the data.5. Designing the algorithm programs of wavelet neural network and BP network by using MATLAB, validating the conclusion that WNN network is better than BP network from both training speed and network performance sides by lots of simulation.6. Validating the trained wavelet neural network for sensor fault diagnosis ability. The simulation testified that fault diagnosis system based on wavelet neural network had better fault diagnosis effect.In the paper, combining wavelet theory and neural network and using it to diagnose sensor fault in intelligent building's Air-conditioning units, and testifying this network has satisfying fault diagnosis ability. The subject is very meaningful to perfect sensor fault diagnosis methods in Air-conditioning system.