Faults Detection And Diagnosis For Soft And Hard Faults Of Air Handling Unit(fame)

The rapid development of society and improved living standards make Air Conditioning system(Heating,Ventilation and Air Conditioning,HVAC)playing an important role in many fields.But often occurred various types of air conditioning system failure,especially the progressive soft fault will not be able to timely find,not only will cause the increase of energy consumption,reducing the service life of equipment,can also lead to the decrease of air quality.It is necessary to develop some various fault detection and diagnosis strategy in air conditioning systems to allow for different types of the fault can be a good test,and identify the fault type,so as to take measures to reduce the influence of the fault,in order to improve the operation efficiency of the system,reduce the operating cost and energy consumption.This paper use principal component analysis based on signal processing detetecting the soft and hard faults of air handling unit and analysis the detection effects of different fault degree ?different fault and different modeling approach.And found that fault detection effect is not only related to variable selection of data sets,and the extent of the fault and the influence of fault on the system.At the sanme time,principal component analysis will not be able to timely and effective to detect the soft fault occurs with a certain delay.As a result of principal component analysis is the method of linearization and the model is fixed,thus this fault detection strategy is not suitable for nonlinear time-varying air conditioning system,it is necessary to make an improvement.Therefore,based on the method of kernel principal component analysis principal component analysis and adaptive moving window make improvements respectively,and verify the effectiveness of the two kinds of fault detection strategy.Results showed that the kernel principal component analysis can effectively improve the effect of fault detection and soft fault is detected in a timely manner.Principal component analysis and moving window while fault detection result is bad,but a low false alarm probability,reduce the air conditioning system in the process of monitoring the unnecessary trouble.Principal component analysis and its improved method can effectively detect the fault occurred,but do not recognize the occurrence of fault locationand and can't provide effective guidance for the follow-up maintenance.Therefore this paper,by using genetic algorithm to optimize neural network approach for fault identification,improved the accuracy of diagnosis.And the fault detection and diagnosis strategy is applied to the Visual Studio software,to monitor the running status of air conditioning.