论文部分内容阅读
通过从故障仿真数据中提取常见故障的效应参数症状 ,建立发动机的集合覆盖多故障诊断模型。根据当前发动机已有测量参数的症状与假设故障的相应参数效应症状的对比结果 ,对各假设故障赋以相应的奖惩值 ,然后综合诸参数的奖惩值得到相应假设故障集的合理性指标值 ,基于该指标值利用遗传算法进行启发式的假设故障产生与测试。最后 ,基于所提取的效应参数症状进行了仿真故障数据样本的多故障诊断分析 ,并与经参数优选后的多故障诊断结果进行了比较。仿真结果表明 ,基于参数偏离方向提取的效应症状具有较高的故障分辨率 ,经过参数优选后能进一步提高故障的分辨率与误分故障样本的集中程度。
By extracting the symptom of the effect parameters of common faults from the fault simulation data, a set of engines is established to cover the multi-fault diagnosis model. According to the comparison between the symptom of the existing engine measurement parameters and the corresponding parameter effect symptom of the hypothetical fault, the corresponding rewards and punishments are assigned to the hypothetical faults, then the rewards and punishments of the parameters are combined to get the reasonable index value of the corresponding hypothetical fault set, Based on the index value, genetic algorithm is used to generate and test the heuristic hypothetical fault. Finally, based on the extracted symptom parameters, the multi-fault diagnosis analysis of the simulated fault data samples is performed and compared with the multi-fault diagnosis results after the parameter optimization. The simulation results show that the effect symptom extracted based on the deviating direction of the parameter has a higher fault resolution, and after the parameter is optimized, the resolution of the fault and the concentration of fault samples can be further improved.