Font Size: a A A

The Research On Automatic Extraction Of CAM Template Based On Random Forest Clustering

Posted on:2018-03-06Degree:MasterType:Thesis
Country:ChinaCandidate:B X GongFull Text:PDF
GTID:2371330566951112Subject:Materials Processing Engineering
Abstract/Summary:PDF Full Text Request
CAM template has the ability to improve the efficiency of NC process and standardization level,which plays a significant role to enhance the research and development strength of enterprises.Thus,how to extract the CAM template file with high accuracy,wide coverage and strong applicability becomes a key issue that is needed to be solved in NC programming.Based on the real cases,combined traditional NC programming with machine learning,this paper put forward an algorithm based on random forest as a new method to extract the CAM template automatically,making template files were no longer restricted by the knowledge level and experience.In this thesis,the numerical models of NC process characteristics were constructed as data bases for research by making full use of real instances.The distance similarity model and random forest similarity model were studied.By conducting multiple sets of comparative tests,the results showed that the latter was more effective ad easy to cluster when performing the internal structure of the NC process dataset.The random forest clustering algorithm was proposed and implemented by integrating clustering analysis with random forest.And the number of each NC operation unit constituting the CAM template file was intelligently matched by silhouette coefficient,which optimizing the template-extraction solution.In contrast with typical K-Medoids,the feasibility and accuracy of the random forest clustering algorithm were verified,and template files extracted by the random forest clustering presented more diversity and applicability.Taking experienced template as the inspection criterion,the accuracy rate was more than 40%,and the average rate of NC operation methods reached 59.5%.A CAM template automatic extraction system was designed and implemented on the UG NX8.5 platform.The random forest clustering algorithm and typical K-Medoids algorithm were provided for users to provide decision support for the use of the extracted CAM templates.The extracted template files were more normative and readability,with favorable engineering application value.
Keywords/Search Tags:CAM template, random forest algorithm, cluster analysis, similarity model
PDF Full Text Request
Related items