Abstract: In view of the shortcomings of the existing methods of fault diagnosis for underground cables with low accuracy and large error, a method of cable fault diagnosis based on HHT transform and FOA_LSSVM is proposed. According to the underground cable fault signal, through the HHT transform to extract the feature component of underground cable fault signals, the characteristic components and underground cable fault types were extracted as the input and output of FOA_LSSVM, to identify the fault types of underground cables. Taking 150 sets of underground cable fault data as the test object, the experimental results show that FOA_LSSVM has higher accuracy than GA_LSSVM, PSO_LSSVM and DE_LSSVM, and is more suitable for the diagnosis and identification of underground cable fault.

Key words: fruit fly optimization algorithm, least square support vector machine, cable fault, HHT transform