基于IWOA-BP算法的超声冲击重轨残余应力预测

Prediction of residual stress of ultrasonic impact heavy rail based on IWOA-BP algorithm

重轨在轧制、矫直等工艺过程中都会产生残余应力,这对重轨的力学性能有很大的影响。超声冲击处理是改善重轨表面残余应力的有效方法。采用3种不同的冲击头和5种工艺参数对重轨表面进行超声冲击处理,并建立有限元仿真模型模拟冲击过程。以实验与仿真数据为基础,建立IWOA-BP算法预测模型,与单层BP、深层BP和传统WOA-BP模型相比,IWOA-BP算法预测表面残余应力和次表层最大残余压应力的精度更高,其平均绝对误差MAE分别为0.008和0.012,决定系数R2分别为0.999 88和0.997 22,表明该模型能够有效地预测重轨超声冲击处理后的残余应力。

Residual stress would be produced in the process of rolling,straightening and treatment of heavy rail,which had a great influence on the mechanical properties of heavy rail.Ultrasonic impact treatment is an effective method to improve surface residual stress of heavy rail.Three different kinds of impact heads and five process parameters were designed for ultrasonic impact treatment on the surface of heavy rail,and the finite element simulation model was established to simulate the impact process.Based on experimental and simulation data,an IWOA-BP algorithm prediction model was established.Compared with single-layer BP,deep BP and traditional WOA-BP model,the IWOA-BP algorithm has higher accuracy in predicting surface residual stress and subsurface maximum residual compressive stress,its MAE values are 0.008 and 0.012,and the R?values are 0.99988 and 0.99722,respectively,indicating that the model can well predict the residual stress after ultrasonic impact treatment of heavy rail.