A fault diagnosis method based on optimized RVM and information entropy for quadruped robot

Abstract

An relevance vector machine (RVM) method is proposed to diagnose the fault of the quadruped robot's hydraulic systems, which is based on information entropy (IE) and cuckoo search algorithm of Gaussian disturbances (GCS). Firstly, information entropy is utilized to preprocess the hydraulic system's raw data, to remove the redundant information and to reduce the data dimension; subsequently, GCS algorithm is utilized to optimize the kernel parameter of RVM; lastly, the RVM multiple classifiers is set up. The vitality of the Bird's Nest Changes is increased by adding gaussian disturbances to Cuckoo search algorithm, which is based on the simulation of cuckoo's parasitic breeding strategy. The experimental results show that, compared with other fault diagnosis methods, the proposed method can reduce training time and increase fault classification accuracy.