基于ISMA-LSTM的燃气轮机剩余寿命预测  

Remaining Life Prediction of Gas Turbine Based on ISMA-LSTM

作　　者：张文海 黄伟 张剑飞 ZHANG Wen-hai;HUANG Weil;ZHANG Jian-fei(School of Automation Engineering,Shanghai Electric Power University,Shanghai 200090,China;Huaneng International Power Co.,Ltd.Yuhuan Power Plant,Taizhou Zhejiang 317604,China)

机构地区：[1]上海电力大学自动化工程学院,上海200090 [2]华能国际电力股份有限公司玉环电厂,浙江台州317604

出　　处：《计算机仿真》2025年第2期128-133,共6页Computer Simulation

基　　金：国家自然科学基金项目(52006131);国家电网公司华东分部科技项目(H2021111)。

摘　　要：剩余寿命预测是设备健康管理技术中的关键问题之一。针对于传统燃气轮机寿命预测模型参数不易选择、预测精度低等问题,提出了一种基于改进的黏菌算法(ISMA)和长短期记忆神经网络的燃气轮机剩余寿命预测模型。模型引入了可调节反馈因子,能加快算法的收敛速度。同时,结合人工蜂群搜索机制,增强了黏菌个体的全局搜索能力,更好地实现了LSTM网络的超参数预测。与CNN-LSTM网络模型相比,ISMA-LSTM模型在进行燃气轮机剩余寿命预测的均方根误差和评分下降了18.8%和31.8%,取得了更好的预测结果。Remaining life prediction is one of the key issues in equipment health management technology.Aiming at the problems of difficult selection of parameters and low prediction accuracy of traditional gas turbine life prediction model,a gas turbine remaining life prediction model based on improved slime mold algorithm(ISMA)and long short-term memory neural network is proposed.The model introduces an adjustable feedback factor to speed up the convergence of the algorithm.At the same time,combined with the artificial bee colony search mechanism,the global search ability of slime mold individuals is enhanced,and the hyperparameter prediction of LSTM network is better realized.Compared with the CNN-LSTM network model,the ISMA-LSTM model reduces the rms error and score by 18.8%and 31.8%when predicting the remaining life of the gas turbine,achieving better prediction results.

关 键 词：燃气轮机 剩余寿命预测 黏菌算法 长短期记忆网络 人工蜂群搜索机制 

分 类 号：TM743[电气工程—电力系统及自动化]