Study On Automatic Loading Control System Fault Diagnosis Based On Particle Swarm-neural Network

Automatic loading control system is the key subsystem of the artillery automatic loadingsystem, whose performance is good or bad will directly affect the power of artillery and itssurvival ability, and automatic loading control system is one of the high fault incidencesubsystems on artillery, so how to prevent its fault will play a crucial role for the overalloperational effectiveness of the artillery. As military equipment system structure is morecomplicated, the use of the traditional fault diagnosis method has been difficult to meet to bethe accuracy requirement of automatic loading control system fault diagnosis. In order toguarantee that automatic loading device can accomplish its prospective function within theprescriptive time and complex environment, this paper uses the chaotic particle swarmoptimizing neural diagnosis network method to conduct the experimental research for thefault diagnosis research of the automatic loading control system of a certain type artillery.First, the development of automatic loading system at home and abroad and the majorsignificance of the fault diagnosis are introduced. The composition and the movementsequence of automatic loading system are primarily analyzed, and hardware composition andsoftware design module of automatic loading control system are also studied. According toworking principle of automatic loading control system, statistical analysis on the main failuremodes of automatic loading system is done.Second, the feasibility of a combination of particle swarm optimization algorithm andneural network and its basic principle are studied, and weights and thresholds of the BP neuralnetwork is optimized using particle swarm optimization algorithm. There are somecharacteristics for particle swarm optimization algorithm, such as simplicity, easy realization,not too many adjustment parameters, etc, but the diversity of population is gradually reduced in the latter part of the optimization process, leading to some phenomena happening, forexample stagnation precocious, so chaos optimization strategy is introduced to the particleswarm optimization algorithm in this paper, which not only enriches the diversity of thepopulation, but also further improves the performance of the hybrid algorithm.Finally, the neural network fault diagnosis system is constructed. On the MATLABplatform, the BP neural network is separately trained and tested with the BP algorithm, thePSO-BP algorithm and the CPSO-BP algorithm, and results of the three algorithms arecompared and analyzed, the final conclusions is that CPSO-BP algorithm is an more idealfault diagnosis algorithm for the automatic loading control system, which has a higherdiagnostic efficiency and accuracy.