论文部分内容阅读
提出基于故障树和神经网络模型的诊断方法，提出面向故障树的基于框架和广义规则的知识表示方法及相应的确定性和可能性推理策略，对于可能性推理的结果，通过基于神经网络模型的学习诊断来进一步确定其状态．在Ｗｉｎｄｏｗｓ环境下，用ＢｏｒｌａｎｄＣ＋＋实现了一个原型系统，通过对“实践４号”卫星能源系统故障模拟实验台的诊断验证了系统的有效性．
A diagnosis method based on fault tree and neural network model is put forward. Based on the fault tree, a knowledge representation method based on framework and generalized rules is proposed, and corresponding deterministic and probable reasoning strategies are proposed. For the reasoning results, Learn the diagnosis to further confirm its status. In the Windows environment, a prototype system was implemented with Borland C ++. The effectiveness of the system was verified through the diagnosis of “Practice 4” satellite energy system fault simulation test bed.