时序卷积网络在转子热固耦合应力预测及寿命评估中的应用

打开文本图片集

关键词：时序卷积网络；压缩机转子；热应力；寿命评估

中图分类号：TK14文献标志码：A

DOI：10.7652/xjtuxb202510008 文章编号：0253-987X（2025）10-0087-09

Application of Temporal Convolutional Network in Rotor Thermo-Mechanical Coupling Stress Prediction and Life Assessment

LI Weiye1， FENG Jianxin1，WEN Siguo1， ZHANG Xiaolong²，YUAN Qi’，LI Pu1 （1. School of Energy and Power Engineering，Xi'an Jiaotong University，Xi'an 71oo49，China; 2.Xi'an Shaangu Power Co.，Ltd.，Xi'an 710075，China）

Abstract：To address the challenge of rapidly predicting transient thermal stress during the startup of turbomachinery rotors，which is difficult due to high computational costs，a compressor rotor surface temperature field and stress field prediction method based on temporal convolutional network（TCN） is proposed. The finite element method is utilized to compute the temperature field，stress field，and service life of the compressor rotor under cold startup conditions. A TCN model is then employed for temperature field and stress field prediction and life assessment of the rotor，and the results are compared with those obtained from three other neural network models ： long short-term memory （LSTM），gated recurrent unit （GRU），and Transformer. Simulation results demonstrate that under cold startup conditions， the TCN model exhibits optimal performance in predicting transient thermal stress of the rotor. Compared to the Transformer，LSTM，and GRU models，the coeficient of determination for thermal stress predictions improves by 0.03% ， 0.60% ，and 0.36% ，respectively，while the coeficient of determination for equivalent stress predictions under combined loading increases by 0.10% ， 0.48% ，and 0.02% .Moreover，the computational eficiency of the TCN model is significantly improved compared to traditional finite element thermo-mechanical coupled analysis，with a computation time only 0.25% that of the finite element method. This proposed method enhances prediction accuracy and provides technical support for the rapid prediction of transient thermal stresses and life assessment of turbomachinery rotors.

Keywords： temporal convolutional network; compressor rotor; thermal stress; lifespan assessment

转子作为汽轮机、燃气轮机、压缩机等旋转机械的关键部件，在工作过程中承受温度以及包括离心力、热应力、气动应力在内的机械载荷，不同启动工况下的温度和应力时变耦合关系复杂，很有可能导致转子热疲劳失效。（剩余13400字）