An intelligent diagnostic method based on optimizing B-cell pool clonal selection classification algorithm

Abstract

The trend of intellectualization and complication of mechanical equipment makes the demand for intelligent diagnostic methods more and more intense in industry. In view of the difficulty of obtaining mechanical fault samples and the requirement of clear and reliable diagnosis results, intelligent diagnosis methods need to adapt to the learning of small samples and have the interpretability of white box model. In this paper, inspired by biological immunity, an intelligent fault diagnosis method was proposed-optimizing b-cell pool clonal selection classification algorithm (OBPCSCA). The OBPCSCA provides a method to construct unique B-cell pools corresponding to specific antigen pools, and uses greedy strategy to generate memory B-cell pools. The experimental comparison with AIRS and AICSL on four UCI benchmark data sets shows that the OBPCSCA has a better balance between the number of memory cells and the accuracy of classification. In particular, compared with AIRS, the OBPCSCA can greatly reduce the number of memory B-cells on the premise of ensuring high classification accuracy. In comparison with the top general classifiers, the OBPCSCA has certain competitiveness in these four data sets. Finally, the algorithm was applied to the bearing data set of Case Western Reserve University for fault diagnosis, and the results showed effectiveness of the algorithm.