数据驱动的燃气轮机联合循环机组退化分析与预测  

Data-driven Degradation Analysis and Prediction of Gas Turbine Combined Cycle

作　　者：曹启威 陈时熠[1] 向文国[1] CAO Qiwei;CHEN Shiyi;XIANG Wenguo(School of Energy and Environment,Southeast University,Nanjing 210096,Jiangsu Province,China)

机构地区：[1]东南大学能源与环境学院,江苏省南京市210096

出　　处：《中国电机工程学报》2025年第6期2243-2250,I0017,共9页PROCEEDINGS OF THE CHINESE SOCIETY FOR ELECTRICAL ENGINEERING

基　　金：国家科技重大专项(2017-I-0001-0001)。

摘　　要：燃气轮机联合循环机组在长期运行后,性能会出现退化,需要及时进行维护。但由于机组退化程度不能直接测量,因此,该文基于改进的深度前向神经网络开发一种数据驱动的退化建模方法,将能够表达重型燃气轮机联合循环系统状态的关键测点数据融合为健康指数以表征联合循环机组的健康状态。通过这种方法分析联合循环机组及其部件性能变化的趋势。结果表明,机组的性能退化主要来自于其蒸汽系统的退化,这与电厂大修总结报告的描述相吻合,证明该方法的有效性。为满足维修的实际需要,采用基于时间序列的长短期记忆网络建立高精度的预测模型,预测部件及机组的健康指数的均方根误差均小于0.03,可以为联合循环机组维修决策提供一定依据。After long-term operation,gas turbine combined cycle units will degrade and require timely maintenance.Since the degree of unit degradation cannot be directly measured,a data-driven degradation modeling method is developed based on an improved deep forward neural network,which integrates the key data that can express the state of the heavy-duty gas turbine combined cycle system into a health index to characterize the health status of the combined cycle unit.Through this method,the performance trends of the combined cycle unit and its components are analyzed.It is assessed that the deterioration of this unit is mainly due to the degradation of its steam system.This is consistent with the description of the power plant overhaul summary report,which proves the effectiveness of this method.In order to meet the actual needs of maintenance,a high-precision prediction model is established through a long short-term memory network based on time series.The root mean square errors of the predicted health indexes of the unit and its components are less than 0.03,which can provide a reliable basis for the maintenance decisions of the combined cycle units.

关 键 词：燃气轮机联合循环 健康指数 神经网络 退化建模 预后分析 

分 类 号：TM611[电气工程—电力系统及自动化]