Abstract: Aiming at the energy efficiency research problem of machining process under the mixed-flow production mode,a method of energy efficiency analysis and state monitoring based on recurrence analysis and image processing technology was proposed.The machining power data of different workpieces was collected through off-line experiments.Then the time-domain characteristics of each workpiece's power under different states were computed and the clustering centers were found by clustering analysis on the power samples.After that,the reference power database of the workpieces was established based on the clustering centers.In mixed-flow production mode,the similarity between the real-time input power and the reference power of all kinds of workpiece was evaluated by recurrence analysis and image matching algorithm,and then the type and state of the processed workpiece were identified.Based on the result of workpiece state recognition,the energy efficiency for processing the workpiece was calculated.Based on the cross recurrence plot of the real-time machining power and its corresponding reference power of each machining process,the process state was monitored according to the recurrence quantification analysis.The effectiveness of the method was verified with a case study.It proved that the proposed method could be applied to analyze the energy efficiency of workpiece and identify the anomaly during the machining process in the mixed-flow production mode,and the identification accuracy was about 98.33%.

Key words: machining, energy efficiency, state monitoring, recurrence analysis, image processing, mixed flow production mode

摘要： 针对混流生产模式下机械加工过程的能量效率研究问题,提出一种基于递归分析和图像处理技术的能量效率分析与状态监测方法。通过对各类工件分别进行多次加工试验,采集加工过程功率数据,应用聚类算法分析功率的时域特征并找出聚类中心,基于此建立各工件加工过程的参考功率数据库;在混流加工过程中,通过递归分析和图像匹配算法判断实时加工功率曲线与各类工件参考功率的相似程度,进而识别出实时加工工件类型和工件状态;基于工件状态识别的结果,计算得到工件加工过程的能量利用率;最后基于各工序的实时加工功率及其对应的参考功率交叉递归图,应用递归定量分析法实现对工序状态的监测。经过实验验证了方法的有效性,将该方法应用于混流加工模式下工件的能量效率水平分析和异常工序状态的在线识别,识别精度可达98.33%。

关键词: 机械加工, 能量效率, 状态监测, 递归分析, 图像处理, 混流生产模式