基于RBF神经网络的燃气轮机监测参数非线性时间序列预测  

Monitoring parameters nonlinear-time-series prognostic based on RBF artificial neural network

作　　者：张乔斌[1] 王文华[1] 贺星[1] ZHANG Qiao-bin;WANG Wen-hua;HE Xing(College of Power Engineering,Naval Univ.of Engineering,Wuhan 430033,China)

机构地区：[1]海军工程大学动力工程学院,武汉430033

出　　处：《海军工程大学学报》2018年第2期96-99,共4页Journal of Naval University of Engineering

摘　　要：针对常规方法对燃气轮机性能参数趋势分析缺乏外推能力的缺陷,基于某型燃气轮机低压涡轮出口温度的运行数据建立了非线性时间序列数学模型,并采用RBF径向基函数神经网络预测了燃气轮机重要监控参数的变化趋势。结果表明:RBF神经网络预测精度高,可为燃气轮机关键监测参数的预测提供一种新的方法。Since the traditional methods has the limitation in prognosticating the trend of gas turbine′s key monitoring parameters,an X-type gas turbine′s nonlinear time series mathematic model is established based on the data of the low-pressure turbine output temperature.By RBF artificial neural network,the trend is prognosticated.The results show that the RBF-basedmonitoring parameters prognostic method is more robust.Research results show that this new method is feasible in gas turbine′s key monitoring parameters prognostics.

关 键 词：燃气轮机 监测参数 RBF神经网络 

分 类 号：TK47[动力工程及工程热物理—动力机械及工程]