A non-destructive surface burn detection method for ferrous metals based on acoustic emission and ensemble empirical mode decomposition: from laser simulation to grinding process
zhensheng yang,haixi wu,zhonghua yu,youfang huang
DOI: https://doi.org/10.1088/0957-0233/25/3/035602
IF: 2.398
2014-01-01
Measurement Science and Technology
Abstract:Grinding is usually done in the final finishing of a component. As a result, the surface quality of finished products, e. g., surface roughness, hardness and residual stress, are affected by the grinding procedure. However, the lack of methods for monitoring of grinding makes it difficult to control the quality of the process. This paper focuses on the monitoring approaches for the surface burn phenomenon in grinding. A non-destructive burn detection method based on acoustic emission (AE) and ensemble empirical mode decomposition (EEMD) was proposed for this purpose. To precisely extract the AE features caused by phase transformation during burn formation, artificial burn was produced to mimic grinding burn by means of laser irradiation, since laser-induced burn involves less mechanical and electrical noise. The burn formation process was monitored by an AE sensor. The frequency band ranging from 150 to 400 kHz was believed to be related to surface burn formation in the laser irradiation process. The burn-sensitive frequency band was further used to instruct feature extraction during the grinding process based on EEMD. Linear classification results evidenced a distinct margin between samples with and without surface burn. This work provides a practical means for grinding burn detection.