基于RB F神经网络的燃气轮机监测参数非线性时间序列预测

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 estab-lished based on the data of the low-pressure turbine output temperature.By RBF artificial neural net-work,the trend is prognosticated.The results show that the RBF-basedmonitoring parameters prog-nostic method is more robust.Research results show that this new method is feasible in gas turbine′s key monitoring parameters prognostics.