基于BAM和FTA在车长周视指挥镜系统故障诊断专家系统的应用研究

摘要

文中对车长周视指挥镜系统的故障进行了深入了解与研究，提出了一种基于BAM神经网络和FTA相结合的诊断方法，利用故障树分析法建立各个分系统的故障树，使用最小割集对系统进行分解，生成BAM神经网络的训练样本。设计出一种权值矩阵的方法，克服了BAM神经网络数据样本冗余的问题，提高了BAM神经网络的训练效率以及收敛速度。应用该方法对某型坦克的车长周视指挥镜系统进行故障诊断，结果表明：该系统能够对故障进行准确定位，提高了故障诊断效率。%Failures for tank commander panoramic system was understood and studied on this article, a new fault diagnosis technology based on BAM neural network theory and FTA, established fault tree by using fault tree analysis theory. Then the minimal cut set method was used to decompose the system and obtain learning sample. A method of generating the weighted matrix was proposed, it can reduce learning sample of redundancy, improve the training efficiency and speed up convergency. At last the issue presented an example to illustrate the application, the result showed that the new method could carry on the accurate positioning of the fault and it was effective to solve these issues.