基于多阶段退化建模的谐波减速器实时可靠性评估与寿命预测

Real-Time Reliability Evaluation and Life Prediction of Harmonic Reducer Based on Multistage Degradation Modeling

谐波减速器是工业机器人的关键核心部件,其可靠性的实时评估和剩余寿命预测对于提升工业机器人的工作性能和健康监管具有重要意义.作为一种复杂的高精度机械部件,其退化过程表现出明显的多阶段性特点.因此,针对谐波减速器的退化特性,提出基于Gamma过程的多阶段退化模型对谐波减速器性能退化过程进行精确描述.首先,根据谐波减速器退化趋势的变化,进行多阶段退化过程的划分;然后采用历史性能指标数据,基于最大皮尔逊相关系数准则估计模型先验分布的超参数,建立基于Gamma过程的多阶段退化模型.在此基础上,针对在役设备的实际运行特点和工程现场中性能指标数据获取困难的问题,提出采用振动特征来建立高斯过程回归模型,对性能指标值进行精准预测以实现对退化模型后验分布参数的实时更新.最后在此基础上对谐波减速器进行实时可靠性评估和剩余寿命预测.通过对谐波减速器可靠性实验数据的分析表明,所提出的方法能够实现可靠性的实时评估,并且与单一阶段退化模型相比,该方法对剩余寿命的预测精度更高.

A harmonic reducer is a key component in an industrial robot,and its real-time reliability evaluation and residual life prediction are important for improving the working performance and health supervision of the industrial robot.As a complex and high-precision mechanical component,its degradation process shows an obvious multistage characteristic.Therefore,considering the degradation characteristic of the harmonic reducer,this study proposes a multistage degradation model based on a Gamma process to accurately describe the performance degradation process of the harmonic reducer.First,the multistage degradation process is divided according to the change of the degradation trend of the harmonic reducer.Using the historical performance index data,the super parameters of the prior distribution of the model are then estimated based on the maximum Pearson correlation coefficient criterion to establish a multistage degradation model based on a Gamma process.In addition,aiming at the actual operating characteristics of the equipment in service and the difficulty in obtaining the performance index data in the engineering site,a Gaussian process regression prediction method based on vibration characteristics is proposed to predict the performance index value.On this basis,the posterior distribution parameters are updated in real-time.Finally,the real-time reliability evaluation and the residual life prediction are carried out.The analysis of the reliability test data of the harmonic reducer shows that the proposed method can evaluate the real-time reliability of the equipment.Compared with the single-process model,this method has higher accuracy in residual life prediction.