Research Of Fault Diagnosis For Rotating Machinery Based On PCA And Ant Colony Algorithm

Rotating machinery has been widely applied in modern industry. It will bring hugeeconomic loss once these equipments have faults. There are more effects to be considered,because of its structure becomes more and more complex, and the intelligent degree is higherthan before with the rapid development of computer technology. It is a problem to realize theaccurate efficient fault diagnosis. We need to extract the effective information to removeredundant information from lots of information. This article is precisely based on thisconsideration, using principal component analysis to extract the feature information, andusing ant colony algorithm to get efficient clustering diagnosis of equipment. The mainly jobis as follows:(1) Summarized typical failure mechanism of rotor system which is the key part of therotating machinery, discussed the methods of rotating machinery signal measurement, faultfeature extraction and fault pattern recognition.(2) The basic principle of principal component analysis is elaborated, the theory forrotating mechanical fault feature extraction based on PCA is also introduced, severalcommonly used methods of principal component selection are summarized. The feature isextracted by the method for feature extraction of rotating machinery based on the principalcomponent analysis through example test. The paper presented an idea about adaptiveprincipal component selection, fault cluster accuracy factor is defined.(3) The basic principle of ant colony algorithm optimization is introduced. Ant colonyalgorithm is improved by its highlight characteristic in solving traveling salesman problem.The method of clustering for rotating machinery fault diagnosis based on ant colonyalgorithm is founded through its strong searching ability when solving TSP problem. Theorthogonal experiment for initial parameters in the method of ant colony algorithm forrotating machinery fault diagnosis with multiple factors in multiple levels is designed. Theexample analysis proved that the method of clustering based on ant colony algorithm forrotating machinery is validity and reliability.(4) The model of fault diagnosis for rotating machinery based on the principalcomponent analysis and ant colony algorithm is founded. The model of feature extraction forrotating machine based on Gauss kernel function used in principal component analysis isfounded, and the principle of the adaptive principal component selection is introduced in this paper. The model of fault diagnosis for rotating machinery based on the principal componentanalysis and ant colony algorithm with the adaptive principal component selection is alsofounded. It is proved that the method of fault diagnosis for rotating machinery based onprincipal component analysis and ant colony algorithm is accuracy and efficiency.