Application Of Online Fault Diagnosis Of Refrigeration System Based On Adaptive PCA

With the rapid increase in the development of science and technology and industrial automation level, the central air-conditioning applications is increasing, resulting in the internal structure of air conditioning and refrigeration systems are increasingly complex, and the losses can not be underestimated when faults take place. How to establish a viable and efficient refrigeration system operating condition monitoring system has become the research focus areas.This paper first introduces the theory and research status and research results of fault diagnosis technology. Then the status of the cooling system and refrigeration principle in detail is analyzed, and a physical model of the refrigeration system is established. By analyzing common faults of cooling systems, the paper sets up a corresponding fault feature table, and the appropriate experimental monitoring points are selected. Given the features of correlation between the refrigeration system to monitor variables is relatively large and the number of monitoring variables is large, the fault diagnosis method based on principal component analysis is of the applicability in the cooling system fault diagnosis.Because the conventional PCA model is based on historical data for a longer time window to establish the main element model, which is equivalent to obtain the main element model data in this time period, the ability of the model describing the original system is not strong enough when parameters change slowly before a fault taking place which causes misdiagnosis. In response to this limitation, we use online adaptive fault diagnosis method based on PCA. The law makes training data into smaller time window and updates corresponding covariance matrix within the time window in a real-time fashion, ultimately based on the covariance matrix obtained using steepest gradient descent to calculate the main elements, and the establishment of the main element model to calculate the control threshold of fault detection. Adaptive capacity of this method is strong, the speed is fast, the main element model established is of relatively high precision, and capable of detecting diagnosis of soft fault system. It makes the foundation of putting adaptive PCA method into the use of refrigeration system fault diagnosis.Finally, this paper uses the online adaptive fault diagnosis method based on PCA fault diagnosis refrigeration system, through the acquisition of experimental monitoring points to several groups of fault data and conventional PCA fault detection method of comparative experiments, found that PCA-based online self fault diagnosis method can be successfully adapted to detect the timing and location of the fault. Experimental results are very good, can obviously diagnose the soft fault in cooling systems which can not be easily found.