Abstract: During hot rolling,the cladding layer of remanufactured work rolls is susceptible to complex performance degradation,affecting the quality of the rolled products.For real-time degradation monitoring and life prediction of these rolls,an online pulsed eddy current monitoring methodology and system were developed.Using the Variational Mode Decomposition—Hilbert (VMD-Hilbert) method,the increment of the marginal spectral energy sum within the 0~5000Hz frequency range was extracted as the roll degradation characteristic.A state degradation model accounting for individual variability was then formulated.Through the integration of the Expectation-Maximization algorithm (EM) and Kalman smoothing filter,the adaptive estimation of unknown parameters within the model was achieved.The Analytic Hierarchy Process (AHP) was employed to fuse historical and real-time monitoring data,facilitating real-time residual life prediction for the remanufactured work rolls.Empirical evaluations indicated the capability of the proposed method to forecast the residual lifespan of these rolls,suggesting its potential application in related fields.

Key words: residual life prediction, remanufacturing, work roll, pulsed eddy current, Wiener process

摘要： 再制造工作辊在热轧工作过程中熔覆层易发生复杂的退化行为,而其性能退化会影响轧制产品的质量。为实时监测轧辊退化情况并预测其剩余寿命,构建一套在线脉冲涡流监测方法和系统,采用变分模态分解-希尔伯特变换(VMD-Hilbert)方法提取0~5000 Hz频率边际谱能量和的增量作为轧辊退化特征。基于此,构造考虑个体差异的状态退化空间模型,结合最大期望(EM)算法和Kalman平滑滤波算法联合对模型中未知参数进行自适应参数估计,并利用层次分析法融合监测的历史数据与实时数据,以实现对再制造工作辊实时剩余寿命预测。实验结果表明,所提方法能够准确预测再制造工作辊剩余寿命,具有一定的应用前景。

关键词: 剩余寿命预测, 再制造, 工作辊, 脉冲涡流, Wiener过程