Tool Condition Monitoring of Single-Point Dresser Using Acoustic Emission and Neural Networks Models
Data de publicação2014-03-01
Direito de acesso
MetadadosExibir registro completo
Identification and online monitoring of the dresser wear are necessary to guarantee a desired wheel surface and improve the effectiveness of grinding process to a satisfactory level. However, tool wear is a complex phenomenon occurring in several and different ways in cutting processes, and there is a lack of analytical models that can represent the tool condition. On the other hand, neural networks are considered as a good approach to resolve the absence of an analytical or empirical model. This paper describes a method to characterize the dresser wear condition from acoustic emission (AE) signal. To achieve this, some neural network models are proposed. Initially, a study on the frequency content of the raw AE signal was carried out to determine features that correlate the signal and dresser wear. The features of the signal were obtained from the root mean square and ratio of power statistics at nine frequency bands selected from AE spectra. Combinations of two frequency bands were evaluated as inputs to eight neural networks models, which have been compared with their classification ability. It could be verified that the combination of the frequency bands of 28-33 and 42-50 kHz best characterized the dresser wear condition. Some of the models produced very good results and can therefore ensure the ground part will be within project specifications.