Clonal selection programming for rotational machine fault classification and diagnosis

The automatic control of technical systems requires increasingly advanced fault diagnosis to improve system reliability and safety. In this paper, a clonal selection programming (CSP)-based fault detection method is introduced. The CSP is inspired by genetic programming (GP) and immune programming (IP). The proposed method has been verified with electrical faults and mechanical faults operating at different rotating speeds. Machine vibration signals are translated into four feature vectors and encoded according to the structure of antibody. Then the extracted features are processed of a CSP-based classifier. Clone classifier uses a powerful search strategy that can get a near-optimal solution in a large search space. The experimental result indicates that the CSP based method can improve the performance significantly and very robust, which indicates that the method is extremely useful for practical industrial applications.