Research On Techniques Of Fault Diagnosis For Certain Mine Fuze And Design Of Testing System

The effective maintenance of weapon systems is an important part of our national defense. Among them, fault detection and location for the weapons systems is of great strategic significance. In the field of military equipment, weapons systems become more and more complex. Inspect ion and maintenance of traditional artificial means have failed to meet the requirements of modern supporting systems to protect the weapon sysems. Now Military Automatic Test System (ATS) has become the necessary mean to ensure reliable operation. However, the maintenance of the army's weapon systems and is in a relatively low level, and the development of testing equipment is slower than that of weapons systems. Just the existing testing equipment and quality of technical supporting personnel are difficult to meet the demand for military training and combat. Therefore, there is an urgent need to develop an effective and automatic fault detection and location system with lower maintenance requirements for quality of technical supporting personnel. And it will be of great military significance and play a positive role in promoting the army's automated testing level and support equipment system performance.This paper combines the national defense research projects and focuses on the key technologies of automatic test system--analog circuit fault diagnosis technologies, and designs a prototype model for automatic fault testing than can be adapted to a particular type of mine fuze.Based on the BP neural network powerful capability of association,memorizing, fault-tolerance,robust and nonlinear mapping, the paper presents a smart analog circuit fault diagnosis method--artificial neural network method. Diagnosis results show that the method is applicable not only to hard fault diagnosis,but also to soft fault diagnosis, and the diagnosis is better than the traditional fault dictionary method. To enhance the generalization ability of neural network for the fault models,the paper studies the impact of the noise study samples on the generalization ability of the neural network. Meanwhile, in order to reduce the ambiguity between the various fault modes, and based on wavelet transform optimization techniques, the paper presents a loose wavelet--neural network analog circuit fault diagnosis method. It uses wavelet transform as a pre-processor for the neural network to effectively extract the features of the fault modes and then submit them to the neural network. And this can greatly improve the accuracy of fault diagnosis. Meanwhile, the wavelet function and feature extraction method are stuied to see the impact on the fault diagnosis results. Combined with the fault example, the paper proposes a new simple and effective fault feature extraction method. The paper also gives a brief introduction to the compact wavelet neural network.Finally, this paper presents a detailed design of automatic test system hardware,...