The Simulation Research Of Subway Station Chiller Fanlt Detection Based On KICA-SVDD

With convenient,safety,high speed and large capacity,the subway is widely used in modern society.Ventilation and air conditioning system can provide a good environment for subway station.It provides comfort for passengers.It is an indispensable part of the subway station.Chiller is one of the most important equipments in ventilation and air conditioning system.It produces the cold source for the station.Because it runs under the variable operating conditions,it is very prone to fault.If the fault can not be ruled out in time,it is bound to bring discomfort to passengers,increase the energy consumption of the system and shorten the life of the equipment.Therefore,it becomes the focus of research about how to detect the fault effectively.In this paper,the water chiller of subway station is studied.I research on some typical fault of chiller.Aiming at the problem of nonlinear,non Gauss property and outliers in data acquisition,I horoughly study on kernel independent component analysis algorithm.(1)The fault detection method of chiller based on KICA is studiedThrough the analysis of the data of American Society of Heating,Refrigerating and Air-Conditioning Engineers,it is found that there is nonlinear and non Gauss between the variables of chiller.It is a very prominent performance for these features using KICA.Therefore,in this paper,I firstly use KICA to process the data,and then construct statistics.The experimental results show that the fault detection model based on KICA is better than ICA and PCA.It can improve the accuracy of fault detection.(2)A fault detection model based on outlier detection is studiedThe experimental data comes from the real scene.It is inevitable about abnormal data.Therefore,in order to solve the problem of outliers in data,this paper presents a method based on outlier detection.It is compared with the traditional methods based on distance.This method does not need to set the threshold artificially.It can increase the reliability of the algorithm.Through the analysis of the data processing,the data is smooth,more aggregated,and the negative entropy value is larger by this method.It is found that the algorithm is more robust and the accuracy of fault detection is improved.(3)Research on the fault detection method based on KICA-SVDD with particle swarm optimizationKernel independent component analysis assumes that the process variables obey the non Gauss distribution,while the traditional Mahalanobis distance cannot satisfy the hypothesis.The blindness of parameter selection is existenced in the process of realization.To solve the above problems,this paper proposes a fault detection method based on KICA-SVDD with particle swarm optimization.The method maps the data into a high dimensional feature space using nuclear techniques.Establish an optimal classification hyperplane in the feature space.The plane maximum the distance of normal data and fault data,minimize the internal normal data and fault data dispersion to solve the problem of parameter selection.Subsequently,Support Vector Data Description is introduced to construct the R statistic,which overcomes the shortcomings of traditional statistics.The method is validated by ASHRAE 1043-RP data,and the results show that the method is superior to the traditional method,which improves the accuracy of fault detection.(4)Analysis and Research on the field experimenThe research of this subject is applied to the training platform of a university in Beijing to prove the effectiveness of the method.The results show that the proposed method can significantly improve the fault detection rate,compared with the traditional methods,especially for some small fault.Through the application of this method,it is helpful for early fault detection.This method can prevent the abnormal shutdown,prolong the service life and reduce the energy consumption of the equipment.