基于多信号流模型的诊断策略动态生成

Dynamic Generation Method for Malfunction Diagnosis Strategy Tree Based on Multi-signal Flow Model

摘要:

针对静态故障树诊断序列长、人工可干预度差和不支持多现象并行推理问题,提出了基于D-矩阵和Rollout信息启发搜索算法的故障诊断策略动态生成方法,介绍了基于TEAMS多信号流模型的D-矩阵数据获取方法和Rollout信息启发搜索算法的DLL独立封装技术、输入输出数据规范和动态交互控制方法;基于此开发的交互式智能诊断系统(IIDS)软件平台经实际现场测试诊断验证,对诊断现场出现的多种客观条件,如系统工作模式或测试资源变化、用户干预和多故障现象并发推理等具有很好的适应性,故障覆盖率和隔离率指标高,对提高装备故障诊断的效率、准确性和灵活性具有显著作用。

Abstract:

To solve the applying problems of long diagnostic sequence, poor artificial intervention and no support to parallel diagnosis of multiple fault phenomena about the static generation method for the malfunction diagnosis strategy tree, with considering the field testing and diagnosis of ship equipments,on the basis of the TEAMS multi-signal flow model data structure, the study of crack method to D-matrix and the standalone DLL files, input/output specification and interactive control method of the search algorithm intrigued by the ROLLOUT information, the dynamic generation method for the malfunction diagnosis strategy tree is designed in the paper. The generation method for the D-matrix data based on the TEAMS multi-signal flow model and the interative control method for search algorithm intrigued by the ROLLOUT information is introduced in the study. Based on the dynamic diagnosis strategy of the malfunction tree and the distributed test diagnosis methods, a platform of interactive intelligent diagnosis system ( IIDS) is developed. The system with high adaptability to the objective conditions for the diagnosis of the scene, such as system operating mode or test resource changes, user intervention and parallel diagnosis of multiple fault phenomena, it’s proved to have good effect in improving efficiency, accuracy, flexibility of equipment fault diagnosis.