Research And Realization Of The Online Fault Intelligent Diagnosis Method In Terms Of The Central Air-conditioning In Intelligent Community

Central air-conditioning is one of the important facilities in community. Because of it consists of numerous equipments and complex system, it is very difficult for repair staff to find the fault and repair in a short time if there is malfunctions. And that will cause a great impact on the community residents’ life and work. The traditional off-line monitoring can be very time-consuming, thus it is difficult to realize the real-time management of the central air-conditioning. Accordingly, in terms of online monitoring of central air-conditioning, it has extremely vital significance for improving the repair efficiency of equipments, reducing the affect on residents and extending the life of equipments.The intelligent community is a platform which gathers facilities, systems and functions of the community based on the internet of things and internet. In this paper, the central air-conditioning in an intelligent community is studied. With the help of internet of things and sensors, the parameter data that react to the running state of the central air-conditioning is collected in real time. Afterwards, the remote real-time monitoring and intelligent malfunction diagnosis of central air-conditioning can be realized on internet. The research contents are as follows:(1) The existing general fault diagnosis methods at home and abroad are summarized. Based on it, this study employs the neural network as the fault diagnosis method concerning the drawbacks that the central air-conditioning system is complex, fault relation is not clear, working conditions is poor, etc..(2) In this paper, the BP neural network model and the SOM neural network model are established, so as to meet the requirements of speed accuracy and reliability in terms of fault warning of central air-conditioning. Moreover, the simulation tests are conducted and the results of fault diagnosis are analyzed as well.(3) The BP neural network is easily trapped into local optimum and needs a large number of samples, meanwhile the SOM neural network have sparse classification and low accuracy. In order to conquer these shortcomings, the SOM-BP tandem neural network model is proposed in which the output of SOM neural network can be forwarded as the input of the BP neural network. By training and testing of the tandem network and analyzing the training error and testing accuracy, the conclusions can be obtained that the improved algorithm can accelerate the convergence and improve the accuracy of fault diagnosis.(4) Based on the above model and algorithm, this paper develops the intelligent on-line fault diagnosis system for the central air-conditioning in intelligent community, and this system is embedded with the SOM-BP tandem neural network fault diagnosis algorithm. Then this system is tested in practical application environment which achieved satisfactory results.