分布式故障诊断专家系统在运载火箭发射决策中的应用研究

Research and Application of Distributed Fault Diagnosis Expert System in Launching Decision-making

根据新一代运载火箭地面测发控系统的一体化设计要求,将分布式故障诊断专家系统应用到运载火箭发射决策中,建立了运载火箭分布式故障诊断专家系统的总体结构。根据故障信息的流动方向将诊断过程分为征兆获取级、分系统诊断级以及全局诊断和决策级3级。征兆获取级通过特征提取和神经网络模块实现复杂测试数据的征兆自动获取。分系统诊断级和全局系统诊断级的诊断推理均建立在"三表制"知识表达方法上,但根据各自任务的不同,分系统诊断级采用征兆驱动的正向推理,并结合深度推理算法实现规则的遍历问题;全局系统诊断和决策级采用面向故障的反向分级推理,并提出了故障的分级策略。

Based on the integrated design requirement of ground test launch and control system for new generation launch vehicle, distributed fault diagnosis expert system is constructed. The diagnostic process is divided into three levels, i. e, symptom acquisition level, subsystem diagnosis level, global diagnosis and decision-making level. The symptom acquisition level is designed by character extraction and neural networks. At the other two levels, reverse inference and deep inference facing the fault models are used to accomplished the diagnostic tasks. This research offers the key technique to launching decision - making of launch vehicle.