基于支持过程向量机的航空发动机排气温度预测  

Aeroengine exhaust gas temperature prediction using support process vector machine

作　　者：于广滨[1,2] 丁刚[1] 姚威[1] 黄龙[2] 

机构地区：[1]哈尔滨工业大学机电工程学院,黑龙江哈尔滨150001 [2]哈尔滨理工大学机械动力工程学院,黑龙江哈尔滨150080

出　　处：《电机与控制学报》2013年第8期30-36,共7页Electric Machines and Control

基　　金：哈尔滨市科技攻关项目(2011AA1BG059);黑龙江省国际合作项目(WB10A104)

摘　　要：针对航空发动机气路性能衰退主要是由时间累积效应造成的这一问题,为反映航空发动机气路性能参数时间序列中实际存在的时间累积效应,以预测航空发动机气路性能衰退规律,本文从泛函分析的角度出发,提出了一种支持过程向量机模型。并建立了基于支持过程向量机的时间序列预测模型,且以Logistic混沌时间序列预测为例验证了该预测模型的有效性。在此基础上建立了基于支持过程向量机的航空发动机排气温度预测模型,并采用遗传算法进行模型参数的优化选择。通过航空发动机排气温度预测实际应用案例对提出的模型进行了验证,实验结果表明:支持过程向量机预测结果的平均相对误差为2.81%,优于传统支持向量机的预测结果。The gas path performance deterioration off the aeroengine is mainly influenced by the time cumulative effect in the practical engineering. In order to predict the gas path performance deterioration, a support process vector （SPVM） model is proposed from the point view of functional analysis, and a time series prediction model based on the SPVM is developed, an its efficience is verifiedied by the Logistic chaotic time series prediction. Then, an EGT prediction model is developed based on the SPVM, and the genetic algorithm is employed to optimize the parameter selection of the prediction model. The proposed EGT prediction model is validated by the real EGT datum from some airline company, the relative error of the prediction results of the SPVM is 2.81%. According to the results, the SPVM has a higher precision than the model based on the traditional support process vector model.

关 键 词：航空发动机排气温度 泛函逼近 过程支持向量机 时间序列预测 遗传算法 

分 类 号：TM46[电气工程—电器]