Radar Fault Diagnosis Technology Based On Darkroom Testing

Radar is widely used in military and other fields,and radar fault diagnosis technology is a key technology to ensure the reliability and safety of radar systems.Due to the limitations of space size,small vehicle-mounted,airborne,and missile-borne radars are designed with integration and unification as the main focus,which greatly increases the difficulty of fault diagnosis.To address the above challenges,this paper focuses on radar fault diagnosis technology and explores the following two technical aspects and methods:on the one hand,using the testability D matrix-related methods to establish radar system testability models,and obtaining fault diagnosis trees through test sequence optimization algorithms? on the other hand,analyzing and predicting fault types of radar anechoic chamber test MTD data through machine learning algorithms.The main research content of this paper is as follows:(1)The testability model of radar system is constructed and a Rollout algorithm based on sub-module testability modeling is proposed for the computational explosion problem of Rollout algorithm.In this paper,the testability block diagram of radar system and the testability D-matrix acquisition technique are used to construct the testability D-matrix of radar system in modules and merge each module D-matrix by D-matrix merging algorithm to generate the testability D-matrix of radar system considering the complex structured characteristics of radar system.In order to solve the problem of computational explosion of Rollout algorithm on complex systems,SM-Rollout algorithm is proposed based on the above sub-module modeling idea,which obtains the fault diagnosis tree of the whole system by generating fault diagnosis trees for each module separately and stitching the module diagnosis trees,which effectively reduces the running time of Rollout algorithm.(2)A test sequence optimization algorithm based on heuristic particle swarm and information entropy is proposed.The traditional particle swarm algorithm cannot directly obtain the fault diagnosis tree and the optimal desired test cost for the test sequence optimization problem.In this paper,we redefine the mapping of particle position,velocity,etc.in particle swarm optimization algorithm and test sequence optimization problem by combining heuristic particle swarm optimization algorithm and information entropy algorithm,and use the fault diagnosis tree obtained by information entropy algorithm as the objective function of optimization of particle swarm algorithm.An example is given to illustrate the operation flow of the algorithm,and the correctness of the algorithm is verified by the radar system testability D matrix and stochastic simulation experiments.Finally,the main control parameter selection problem is analyzed by stochastic simulation experiments.(3)A radar fault diagnosis strategy based on Res Net network is proposed.To address the problem that there is no publicly available radar fault diagnosis dataset on the network,the radar fault diagnosis dataset is constructed by collecting radar darkroom test data through the radar fault acquisition system? and the dataset is divided and data enhanced.Radar fault diagnosis models are established on the basis of three structures,Res Net18,Res Net34 and Res Net50,and the models are trained under optimizer SGDM,Adam and RMSProp respectively,and compared with common fault diagnosis methods based on pattern recognition,and the results show that the radar fault diagnosis method based on Res Net18 has the optimal performance...