Study On Fault Diagnosis Expert System Of Centrifugal Chillers Based On Neural Network

Building energy consumption in the proportion of social energy is graduallyrising, while the air-conditioning system energy consumption is an important part ofbuilding energy consumption, timely and effective troubleshooting air conditioningsystem can not only maintain a comfortable indoor environment and ensure normalproduction, but also reduce the energy consumption of the device. As the core of theair conditioning system, chillers energy consumption accounts for about55%of thewhole system energy consumption, the result of chillers running with fault are worsethan any other equipments, what’s worse, only the experts of the equipmentmanufacturers could remove the fault, and this lead trouble clearing cost much time.So, intelligent monitoring and fault diagnosis of air conditioner has a high researchvalue.Taking centrifugal chillers for example, combines the respective merits of neuralnetworks and expert systems in fault diagnosis, Fault Diagnosis Expert System ofCentrifugal Chillers Based on BP neural network was proposed. Choosing7typicalfaults of centrifugal chillers as the researching object, and determine the parameterscharacterizing fault symptoms. As the initial weights of BP network was randomlyselected, so the network is easy to get into a local optimum and to spend a lot of time,while genetic algorithm has strong global search ability, using this ability to pick theinitial weights of BP network and making experiment, the result show that the BPnetwork can astringe faster when using genetic algorithm. Finally, using MATLABand VB for developing a user-friendly and easy operate intelligent fault diagnosissystem of centrifugal chillers.