The Application Of Artificial Immune System In Rotating Machinery Fault Diagnosis

With the continuous progress of modern technology, rotating machinery is developing towards the direction of high-speed, large-scale and complication. Once the rotating machinery failure occurs, it will cause huge economic losses and safety issues. However, at the present stage, the fault diagnosis methods of rotating machinery have more or less problems. Artificial immune system has the advantage of it can establish multi-layers diagnosis mechanism,need a little prior knowledge, and only need a partial list of abnormal samples. Based on these, artificial immune system based rotating machinery diagnosis method has a broad development space.This paper studies the rotating machinery fault diagnosis based on artificial immune system, establishes a rotating machinary fault diagnosis system. This paper includes three parts: data processing, fault detector and two-level fault diagnosis.Data processing section uses amplitude domain analysis to extract dimension parameter and dimensionless parameters, and these parameters are processed using normalized method and linear dimension reduction to structure normal feature space and abnormal feature space.Fault detector section contrastes the real-value negative selection algorithm and V-detector negative selection algorithm; improves clonal selection algorithm to optimize the detector generation process.Fault diagnosis section againsts that the negative selection algorithm can only recognize normal and abnormal, rather than diagnosis the fault type, proposes a new two-level diagnosis method. This method constructes multiple detector sets, uses multi-detector sets fusion diagnosis method to diagnose fault type. Finally, detector sets are updating on-line through the diagnostic result of fusion matrix.Finally, a fault diagnosis system is constructed based on MATLAB GUI.Experimental results show that the improved clonal selection algorithm can optimize the negative selection algorithm’s detector generation process. The proposed fault diagnosis system can effectively identify the fault type of rotating machinery. Diagnosis results can reach a good accuracy and have a certain application value.