融合改进PSO与TCN的核电冷却水泵故障诊断模型

打开文本图片集

关键词：冷却水泵故障诊断；时间卷积网络；粒子群算法；收敛速度；故障诊断精度DOI：10.15938/j. jhust.2025.04.003中图分类号：TP391.4 文献标志码：A 文章编号：1007-2683（2025）04-0018-12

Fault Diagnosis Model for Nuclear Power Cooling Water Pump by Integrating Improved PSO and TCN

PENG Junjie’，LI Yang²， HOU Yimin³ （1.Taishan Branch，China Nuclear Industry Maintenance Co.，Ltd.，Jiangmen 529228，China; 2.School of Mechanical Engineering，Northeast Electric Power University，Jilin 132O12，China; 3.School of Automation Engineering，Northeast Electric Power University，Jilin 132O12，China）

Abstract：A nuclearpower cooling water pump fault diagnosis model（AMsPSO-TCN）isproposed，which integrates improved PSOandTCN，toaddresstheisuesof variableperatingconditionsanddificultyinfault diagnosisofnuclearpowercoingwater pumps.This modelutilizesanAdaptive Multi Swarm Particle Swarm Optimization Agorithm（AMsPSO）tooptimizeandadjust the thresholdandbiasoftheTemporalConvolutionalNetwork（TCN），andusestheoptimizedTCNforfaultdiagnosisofcoolingwater pumps.Intheexperiment，theAMsPSOalgorithmwasfirstcomparedwith existing improvedPSOalgorithmsforbenchmark function optimization.TheresultsshowthattheAMsPSOalgorithmhasbeteroptimizationaccuracyandfasterconvergencespeed.Thenusing the measureddataofacoling waterpumpinanuclearpowerplantasexperimentaldata，the AMsPSO-TCNmodelimproved he fault recognition accuracy by more than 8. 1% compared to existing models，reaching 98.3% .Therefore，the AMsPSO-TCN model is an effective fault diagnosis model.

Keywords：coling waterpumpfault diagnosis；temporalconvolution network ；particle swarm algorithm；convergencespeed；fault diagnosis accuracy

0 引言

随着核能技术的发展，核电在发电领域比重逐渐增加。（剩余14056字）