Abstract: Because spiking neural networks(SNNs)could convey both temporal and spatial information at the same time, and contain features that were more attractive than those of traditional neural networks(NNs), SNNs were more suitable for analyzing the dynamic time-series signals. A novel fault diagnosis method based on SNNs was proposed to distinguish manipulators' collision and obstruction failure states from the normal working state,as manipulators approaching the grasping position. The architecture of the SNNs for fault diagnosis was discussed, and the results for SNNs fault diagnosis methods with different SNNs' topologic structures and parameters were compared. Experimental results showed that the proposed fault diagnosis method based on SNNs was effective and helpful for manipulators' fault diagnosis, which was also important for manufacture industries' smooth and safe running.

Key words: fault diagnosis, manipulators, Spiking neural networks, time-series signal, STDP learning