Research On Fault Prediction Of Electronic Equipment In Missile System Based On FPGA

Missile weapon plays a very important role in modern war,and it is an important guarantee for every country to protect its people and land.In the era of peace,missile maintenance is particularly important.The traditional way of missile maintenance is very inefficient and expensive,so it is imperative to find a new way of missile maintenance.Fault prediction is a new way of missile maintenance.In recent years,neural network has shown its unique algorithm advantages in the field of prediction.Applying neural network to missile fault prediction will greatly reduce the manpower and material resources caused by missile maintenance.Neural network can flexibly use mathematical means to realize machine learning with its idea of parallel operation,but when computer software runs neural network,it is serial operation,which can not fully play the advantages of neural network,so this paper proposes a hardware means to realize the prediction of neural network to electronic equipment fault.Firstly,this paper analyzes the failure mode and failure mechanism of various components,determines the type and number of components involved in the prediction,and defines the fault injection method and fault simulation method.Then,three typical circuit structures of missile weapon are designed to inject faults into the components to be predicted.Through the collection of the voltage values of the key points in these three circuits,a complete fault database is established,and each fault corresponds to a set of voltage values.The established fault database is used to train and verify the probabilistic neural network,constantly adjust the algorithm statements and algorithm parameters,realize the probabilistic neural network fault prediction through MATLAB,compare the prediction results with the real results,verify the accuracy of the program code and the feasibility of the algorithm,and then realize the probabilistic neural network fault prediction in FPGA through vivado The output uses the four LED lights on the FPGA board as the indicator lights.It is proved by experiments that the predicted output results of the probabilistic neural network algorithm on the FPGA are completely consistent with the actual situation,that is to say,it verifies the research of the missile system electronic equipment fault prediction on the FPGA.Finally,two optimization methods are proposed,which can reduce the prediction time and the prediction risk.Then,the data is input into the optimized program,and the simulation results are compared with the original test results to verify the feasibility and rationality of the optimization.