Multi-Information Direction Online Identification Based on RF-ELM Single Crystal Diamond Circular Arc Tool Indexing and Grinding

In view of the high hardness, wear resistance, anisotropy and other characteristics of single-crystal diamond cutters, in order to improve the grinding efficiency and grinding quality, this paper takes single-crystal diamond circular cutting tools as the research object, and proposes an online identification method based on CEEMD-RF-ELM single-crystal diamond circular cutting tools. The tool vibration signal and the acoustic emission signal in the grinding process are taken as characteristic signals, the signal noise reduction is carried out by CEEMD method, the feature signals are analyzed by time domain, frequency domain and time-frequency domain methods, and the feature signals are screened by the Relief-f algorithm to determine the feature parameters strongly related to the grinding direction of the tool. With the input feature set of ELM neural network optimized by RF, the online recognition model of tool grinding direction with multi-information RF-ELM is established, and then the grinding direction on each arc of tool grinding process is judged. The experimental results show that the recognition accuracy of RF-ELM model after multi-information fusion reaches 93.18%. The extreme learning machine training network optimized by random forest algorithm has significant advantages in the online recognition accuracy of tools, and can effectively identify the indexing and grinding direction of circular arc tools. In this paper, a new method for online orientation recognition of single-crystal diamond tool indexing and grinding process is proposed.