Fault diagnosis model based on Gaussian support vector classifier machine

摘要

In view of the bad diagnosing capability of standard support vector classifier machine (SVC) for fault diagnosis pattern series with Gaussian noises, Gaussian function is used as loss function of SVC and a new SVC based on Gaussian loss function technique, by name g-SVC, is proposed. To seek the optimal parameter combination of g-SVC, particle swarm optimization (PSO) is proposed. And then, a intelligent fault diagnosing method based on g-SVC and PSO is put forward. The results of its application to car assembly line diagnosis indicate that the diagnosing method is effective and feasible.