Fault diagnosis algorithm of industrial wireless sensor networks based on immune danger theory

Since fault detection algorithm can only be used to acquire fault occurrence time and fault location, the information related to failure reason is still very limited, thus failing to provide sufficient support for making reasonable failure recovery schemes. Therefore, fault diagnosis algorithm is required to identify the failure types. As the fault features of industrial wireless sensor networks are quite complicated, complete fault-feature knowledge base and feasible diagnosis algorithm are quite necessary during the diagnosis process. Due to this reason, we proposed a fault diagnosis algorithm based on immune danger theory. In this algorithm, through antigen classification process, the fault-feature samples can be classified according to fault types. Through antibody training process, the fault-feature knowledge base with good fault-description ability can be built. When the fault samples enter into diagnosis process, the fault types can be detected according to KNN. The experiment results have shown that the time length required for diagnosis in our algorithm can be shorten significantly while maintaining high accuracy. Its performance could meet the low-delay demand of industrial scenarios perfectly.