Extracting and Optimizing Sound Features in Mechanical Fault Diagnosis Using Genetic Programming
Wang Feng,Qu Liangsheng
(School of Mechanical Engineering,Xi'an Jiaotong University,Xi'an 710049,China)

Abstract: A fault detection method is introduced, which uses compound features optimized by genetic programming based on single features. Some single features can be combined to form a compound feature. A compound feature obtained from sound signals is used to diagnose faults of rolling bearings, and its effectiveness is verified in practice. Based on the method, a new method is presented, which is improved by the information fusion technique. The features from sound signals and vibration signals are combined, and a new compound feature can be obtained by genetic programming. This feature can be used to detect faults of rolling bearings with higher efficiency and reliability.
Keywords: fault diagnosis;genetic programming;information fusion;pattern recognition