Research Of The Fault Diagnosis Of Air Compressor Based On Principal Component Analysis

This paper purposes to find an efficient and feasible fault diagnosis method which can apply to air compressor. The paper uses the air compressor’s principle of work, physical structure, fault type and fault principle, summarizes the main fault characteristics and the fault diagnosis system’s realized requirements of air compressor.The paper focuses on how to solve the high related problem between too many fault detection variables and detection variables of the air compressor in process of finding the suitable fault diagnosis algorithm. In order to solve the problem, the paper uses the Principal Component Analysis technology (PCA) as the test data preprocessing algorithm. The algorithm can find out dates’the main changed direction and trend in high dimensional space, thus can pick up characteristic vectors including most of the original dates information to replace the high dimension and high correlation original dates through analysis the sample dates’distribution in high dimensional space. Based on PCA as the test data preprocessing algorithm, this paper brings up combining PCA and Radical Basis Function (RBF) neural network fault diagnosis method of air compressor and combining PCA and D-S evidence theory fault diagnosis method of air compressor.Air compressor fault diagnosis method based on Principal Components Analysis (PCA) and Radical Basis Function (RBF) neural network, deal with the collected data from a large, high correlation original data set by establishing the main element model of the state on operation of the air compressor. Simplify the original data by using the feature extraction methods. And then, make use of the simplified sample data to train the RBF neural network. Finally, achieve the air compressor fault classification by the trained RBF identify network. The method can give full play to the advantages of PCA technology in data dimensionality reduction and in addition to correlation, which greatly simplified the complex test data. At the same time, the PCA dimension reduction greatly simplified the computing process for the training and recognition of the RBF network, which can improve the speed of the training and recognition of neural network. Meanwhile, the reduce of the data dimension, which the neural network processing to, not only avoid the potential risk of breakdown due to the data dimension,which the RBF network is dealing with in the training process, is too high, but also improve the resolution of the neural network.Air compressor fault diagnosis method based on Principal Components Analysis and D-S evidence theory is a fault diagnosis method based on a thought of information fusion. The method observed by different states on operation (different evidences) of the Air compressor. Judge the state on operation of the Air compressor by analyzing the characteristic information under the every evidence of the testing data. Merge judgment results under the every evidence into a comprehensive result based on D-S combination rules. Thus, achieve the ending judgment about states on operation of the Air compressor. This method can analysis the information of the testing data more comprehensive. Achieve the high precision fault isolation and discrimination. In addition, this method also has characteristics of faster processing speed and strong anti-interference ability.