基于粗糙集和支持向量数据描述的发动机视情维修研究

Research of Engine Condition-based Maintenance based on Rough Set and Support Vector Data Description

每个热力参数都可从某一角度反映航空发动机的性能状态,在分析各个性能参数变化特点基础上,选取能够最大程度反映发动机性能变化的参数组合。运用统计粗糙集理论属性重要性判定方法,确定各个参数的权值,从而进一步对所选参数组合进行信息融合,能够更加准确的判断发动机性能状况。提出了利用支持向量数据描述的发动机视情维修阈值确定方法。利用非目标数据即性能下降到限制值附近时刻的数据,来提高视情维修的准确率。仿真实验结果表明,该方法能够准确确定视情维修时机,对发动机视情维修时机选择有一定的参考价值。

Every thermodynamic parameter may be using in reflecting engine condition, on basis of analyzing change character of every performance parameter, selecting parameter combination that expresses change of performance furthest. The uniformity of rough sets and statistical methods was employed to find out the significance of each attribute, the weight coefficients could be determined. Further selected parameter combination is fused, therefore engine condition is judged accurately. The method based on support vector data de-scription for valve determination of condition-based maintenance is presented. By adopting outlier object sample, the accuracy of condi-tion-based maintenance is enhanced. The result of simulation experiment shows this method can determine time of condition-based ma-intenance accurately. It has reference value for time selecting of condition-based maintenance.