计算机集成制造系统 ›› 2022, Vol. 28 ›› Issue (3): 700-708.DOI: 10.13196/j.cims.2022.03.005

• • 上一篇    下一篇

加工监控数据准确性量化评价及优化研究

唐思游,胡小锋+,刘颖超   

  1. 上海交通大学机械与动力工程学院
  • 出版日期:2022-03-31 发布日期:2022-03-28
  • 基金资助:
    上海市科学技术委员会资助项目(19511105302)。

Accuracy quantitative assessment and optimization of processing monitoring data

  • Online:2022-03-31 Published:2022-03-28
  • Supported by:
    Project supported by the Science and Technology Commission of Shanghai Municipality,China(No.19511105302).

摘要: 数据准确性对于数据分析结果的有效性具有重要影响。针对实际加工过程监控数据理论真值未知,其数据准确性难以量化评价的问题,提出一种基于测量不确定度和局部敏感哈希相结合的数据准确性评价及优化方法。首先建立测量不确定度法将数据的不确定度转化为准确度,并通过层次分析法和变异系数法赋权计算综合评价结果。然后,采用欧氏距离下局部敏感哈希及k邻近算法对数据质量进行优化。最后,利用汽轮机转子轮槽加工监控数据进行验证,并与指数平滑去噪、基于距离的离群点处理、最小协方差估计与孤立森林方法进行对比分析,检验所提方法的有效性及数据质量量化评价与优化的重要性。

关键词: 加工过程, 监控数据, 数据准确性, 量化评价与优化, 测量不确定度, 汽轮机转子轮槽

Abstract: Data accuracy deeply influences validity of data analysis.Caused by unavailable theoretical true value of process monitoring data,data accuracy is difficult to quantitatively evaluate.A data accuracy assessment and optimization method combined measurement uncertainty and locally sensitive hashing was proposed.Measurement uncertainty method was applied to convert uncertainty to accuracy.To weigh and calculate comprehensive evaluation,the analytic hierarchy process and coefficient of variation were used.Data quality was optimized by locally sensitive hashing under Euclidean distance and k-nearest neighbor algorithm.To verify the proposed method,monitoring data of turbine rotor groove was used.Exponential smoothing,distance-based outlier detection,minimum covariance estimation and isolation forest were utilized to verify the effectiveness of measurement uncertainty method and importance of data quantitative evaluation and optimization.

Key words: processing, monitoring data, data accuracy, quantitative assessment and optimization, measurement uncertainty, turbine rotor groove

中图分类号: