Gridding modeling and extraction of spatiotemporal correlation characteristics are conducted using unsupervised clustering algorithms to achieve continuity labeling results for support pressure data and to identify areas with pressure agglomeration at the working face.Based on the continuity labeling results,the label with the highest classification value in the same cycle is adopted as the label for this cy-cle,thereby periodic distribution curve of the incoming pressure at the working face can be obtained,which enables the fully programmed solution of the characteristic parameters of the periodic incoming pressure.Simultaneously measuring both the dynamic pressure coefficient and the number of continuous incoming pressure cycles enables the classification of various levels of incoming pressure strengths.Com-pared with the periodical incoming pressure characteristics obtained by manual observation in the field,the method yielded incoming pressure discrimination results with 90.91％accuracy and a 100％success rate,validating the reliability of the results.In terms of incoming pressure classification,the relatively strong on-site pressures are accurately recognized,while the strong pressures are classified in depth,thus the scientific and accurate classification of incoming pressures is realized.
关键词
支架状态/时空特征/连续性分类/来压判别/来压分类
Key words
support status/spatiotemporal characteristics/continuity classification/incoming pressure discrimination/incoming pressure classification
维普
万方数据