Fault Diagnosis for the Electric Starting System of Self-Propelled Artillery Based on Information Fusion Technology

Abstract

Fault diagnosis based on multi-sensor information fusion technology processes multi-source information and data of the monitoring system in various manners such as detection, parallel and related processing, estimation, comprehensive treatment and so on so as to maximize the use of system knowledge and the information provided by the available detectable quantity of the system in fault diagnosis. Compared with the single sensor, multi-sensor information fusion enjoys obvious advantages in reducing information uncertainty, improving information accuracy obtained by the system and advancing system reliability and fault tolerance capability. As the accuracy of traditional fault diagnosis method is not high, considering the characteristics of faults in the electric starting system of self-propelled gun, a method of fault diagnosis is presented here based on network information fusion technology. The diagnostic process is divided into two level diagnosis, that is subsystem and system level. System adopts BP neural network in fault mode classification, while at system level D-S evidence theory is used in the process of synthetic decision evaluation on the entire system malfunction, ensuring accurate and fast fault diagnosis, which greatly shorten the corrective maintenance time.