基于递归特征的关键设备状态评估方法研究

An Algorithm for Key Equipment Condition Monitoring Based on Recurrence Characteristics

设备保持安全和稳定的运行,一直是企业生产管理的重要任务,具有非常重要的经济和社会意义。如何对设备的运行状态进行有效评估,能够快速识别设备运行时的异常状态,是学术界长期关注的重要研究方向。本文基于设备运行过程中的振动信号数据,结合了关联递归图法和多元统计质量控制方法的优势,创新性的提出基于关联递归特征的设备状态评估算法,并利用穷举法对算法参数的训练过程进行优化。通过对滚动轴承滚动体表面损伤状态评估案例分析,表明该算法可以有效识别滚动轴承的异常状态。

Due to the importance of equipment condition monitoring in manufacturing systems and other areas,it is a very important research topic to develop equipment condition monitoring methods to identify the abnormal status timely.This paper will develop a novel equipment condition monitoring algorithm based on vibration signals.The continuous-scale recurrence characteristic and multi-variate statistical quality control theory have been used to build the algorithm.First,the continuous-scale recurrence plots of the vibration signals are derived by using the recurrence plot(RP)method.Five types of continuousscale recurrence are extracted to quantify the vibration signals’characteristic.Then,the multi-variate T2 control chart is used to monitor these features.Due to the assumption that all the data should follow a normal distribution,T2 bootstrap control chart is proposed to estimate the control chart parameters.A real case study of rolling element bearing working status monitoring demonstrates that the proposed method achieves a very good performance.