Design And Implementation Of Intelligent Fault Diagnosis Method For A Military Radar

With the application of high and new technology in radar, modern military radar has become a complex electronic system involving machinery, electronics, control,computer and ultrashort wave technology. The traditional fault diagnosis and exclusion technology have defects of low efficiency and poor reliability, thus affect the daily trainings and military exercises, or even result in irreparable consequences of military actions. Aiming at the compact structure, various signals, varied background noise and poor electromagnetic conditions of a certain type of military radar, an intelligent fault diagnosis method based on fault tree analysis and RUSBoost algorithm is designed and researched.1. The structure and working principle of a certain type of military radar was researched, common faults and possible reasons of the radar were researched according to its structure characteristics and working conditions.2. Fault trees of transmitted pulse fault of launching system and display device fault of receiving system of the radar were established based on fault tree analysis,then qualitative analysis and quantitative analysis of the fault trees were researched. The key factors of fault phenomenon were obtained by qualitative analysis, and influence level of a bottom event to top event was analyzed by quantitative analysis.3. The complexity, diversity and imbalance characteristics of the military radar signals were analyzed, a classification method for the radar signals based on RUSBoost was proposed, and a tris-indexs evaluation system was established.Classification and diagnosis experiments based on ANN, SVM and RUSBoost were executed respectively. The results indicated that, RUSBoost algorithm does well in precision, recall and F1-measure, which shows excellent diagnosis performance.4. An intelligent fault diagnosis system based on fault tree analysis and RUSBoost was designed, and cyclical fault diagnosis experiments for the radar system were carried out. The results indicated that, the diagnosis system hasadvantages of high diagnosis accuracy, extremely low misdiagnosis rate and high efficiency, and it also can count maintenance records and historical faults, which provides reliable basis for performance recovery and evaluation of the military radar.