基于数据分解及因果推理的设备可靠性预测模型

Reliability prediction model of equipment based on data decomposition and causal inference

为解决设备的可靠性数据受多种因素影响而同时具有线性特征和非线性特征的问题,提出1种集合经验模态分解法辅助的设备可靠性预测组合模型,该模型结合差分整合移动平均自回归模型和因果推理模型。首先,对原始数据采用集合经验模态分解法,得到固有模态函数分量和余项;其次,将模态函数分量输入差分整合移动平均自回归模型得到线性分量,进而将线性分量和原始数据作差,得到非线性分量;最后,基于该非线性分量,提出因果分析模型,实现对设备可靠性的有效预测。研究结果表明:与流行的可靠性预测模型相比,组合模型分别在平均绝对误差和均方根误差指标上降低0.015 9和0.026 5,进一步证明本文所提方法的正确性和有效性。研究结果可为工业生产中提升设备可靠性预测提供新思路。

In order to solve the problem that the reliability data of equipment had both linear and nonlinear characteristics due to the influence of multiple factors,a combined model of equipment reliability prediction assisted by the ensemble empirical mode decomposition method was proposed,which combined the autoregressive integrated moving average model(ARIMA)and causal analysis model.Firstly,the ensemble empirical mode decomposition(EEMD)method was applied to the original data to obtain the intrinsic mode function components and residual items.Secondly,the mode function components were input into ARIMA to obtain the linear components,thus the linear components and original data were used as difference to obtain the nonlinear component.Finally,a causal analysis model was proposed on the basis of this nonlinear component to effectively predict the equipment reliability.The results showed that in comparison to the popular reliability prediction models,the combination model reduces the mean absolute error and root mean squared error indicators by 0.0159 and 0.0265,respectively,demonstrating the correctness and efficacy of the proposed method.The results can provide the new idea for improving the equipment reliability prediction in industrial production.