基于遗传优化混合模型发动机稳态数据融合

Aero-Engine Steady-State Data Fusion Based on Genetic Optimized Mixture Model

中文关键词:  航空发动机  稳态数据  高斯混合模型  遗传优化

英文关键词:aero-engine  steady-state data  Gaussian mixture model  genetic optimization

中文摘要:

      航空发动机稳态数据受工作环境、测量方式、控制规律等多方面因素影响，其试验数据常存在异常值，会对稳态性能参数计算结果造成影响。基于数据来源的正态性假设，采用高斯混合模型对数据进行筛选分类。在模型求解算法方面，采用遗传优化算法克服期望极大(EM)算法的局部收敛性，使计算结果具有一定的全局最优性，并复合应用EM算法弥补遗传算法收敛速度慢的不足。使用赤池信息准则或贝叶斯信息准则(AIC/BIC)作为数据筛选结果评定指标，并验证该方法的可行性。在数据融合方面，提出基于最优模型参数相似理论进行数据融合的方法，融合后数值接近真实稳态数据值，可作为该稳态数据片段的特征值。

英文摘要:

      The steady state data of aero-engine is affected by the working environment,measurement method and control law.The test data often have outliers,which affects the calculation results of steady-state performance parameters.Based on the assumption of normality of data sources,Gaussian mixture model is used to screen and classify the data.In terms of model solving algorithm,genetic optimization algorithm is used to overcome the local convergence of exceptation maximization (EM) algorithm,so that the calculation results have a certain global optimality,and EM algorithm is combined to compensate for the slow convergence speed of genetic algorithm.Akaike information criterion or Bayesian information criterion (AIC/BIC) is used as the evaluation index of data screening results,and the feasibility of this method is verified.In the aspect of data fusion,a data fusion method based on the optimal model parameter similarity theory is proposed.The fused value is close to the real steady-state data value and can be used as the eigenvalue of the steady-state data segment.