WEAR MONITORING OF SINGLE-POINT DRESSER IN DRY DRESSING OPERATION BASED ON NEURAL MODELS Pedro O. Junior, Rubens V. Souza, Fábio I. Ferreira, Cesar H. Martins, Paulo R. Aguiar, Eduardo C. Bianchi. FEB - Faculty of Engineering, Bauru Campus, Department of Electrical and Mechanical Engineering, UNESP - São Paulo State University Av. Luiz Edmundo Carrijo Coube, 14-01, Vargem Limpa, CEP 17033-360, Bauru, Sao Paulo, Brazil pedrojunior5@aedu.com, rvcgps@hotmail.com, fab.kf@hotmail.com, ds_cesar@hotmail.com, aguiarpr@feb.unesp.br, bianchi@feb.unesp.br ABSTRACT The monitoring of different machining processes has been studied for years, however many processes still do not have a final solution for their controls. The dressing, as it is of great importance in the finishing of workpieces produced through the grinding, is an operation whose monitoring becomes necessary. In order to make the dressing automation and, in this case, the process of dresser exchange, there is a need for efficient and low- cost monitoring. The vibration sensor has great potential, but it is still little used for this purpose. In this work the vibration sensor and neural models were used to classify the wear of dressing tools for three different conditions. Dry dressing tests and data acquisition were performed in a surface-grinding machine. The raw signals were further filtered in different frequency bands. Then, two statistics were computed, which served as inputs to the neural models. The results were quite satisfactory for some models. KEY WORDS Grinding process, tool condition monitoring, single-point dresser, vibration sensor, and artificial neural networks. 1. Introduction According to Wegener et al., [1], the slogan '' Grinding is dressing '' is kept in the community of grinding process. Dressing indicates, besides all other parameters of the process, the importance of the technology of grinding wheels conditioning in the manufacturing results. The process of the grinding wheel conditioning, which consists of dressing and cleaning, determines the rate of material removal, the forces in the grinding, the quality of the surface and the properties of the materials in the subsurface zone. Grinding is a complex manufacturing process influenced by many factors, such as workpiece, machine, grinding wheel and process configuration. The grinding wheel, which is the cutting tool, is the only factor that differentiates the grinding process from other machining processes. It is known that the topography and the conditions in which the grinding wheel is prepared deeply influence on the performance of the grinding, which is evidenced by the cutting forces, consumed energy, temperature in the cutting zone, and often, in the finish of the wokpiece [2]. In addition, the surface of the grinding wheel plays an important role in the roughness of the pieces [3]. According to Marinescu et al., [4], dressing is the surface conditioning process of the grinding wheel aiming its remodeling when it lost its original shape due to wear. It is the joint operation of profiling and edging of the conventional grinding wheels, aiming to restore the efficiency of the tool cutting. On the other hand, according to Aguiar et al., [5], the use of worn dressers can provide less sharpness to the tool, causing an increase in cutting forces and faster loss of edges of grains. The control of wheel operation and the elimination of undesirable conditions can be detected by an online monitoring system. Therefore, the development of a monitoring system and the process control, in real time, is of fundamental importance. The dressing operation and many other precision manufacturing processes require small forces and little power consumption. In this way, conventional power and force sensors are not suitable for monitoring these processes. Acoustic emission (AE) sensors have been used, as in the research of Martins et al. [6], due to its nature based on highly sensitive piezoelectric elements. In addition to AE, other sensors of the piezoelectric type are currently used to monitor the dressing operation, such as the vibration sensor, for example, which are increasingly present in works that seek to improve its quality and process productivity [7]. The vibration signals are presented as an alternative for the monitoring of manufacturing processes when AE is not used. According to Zeng & Forssberg [8], the grinding process emits strong vibration signals that occur in various forms of acoustic pressure and mechanical vibrations. The measured vibration signal can be studied using methods of time domain analysis and spectral estimation, whose variations can be related to the characteristics of the grinding process [8], [9]. According to Hassui et al., [10] the Root Mean Square (RMS) of the Proceedings of the IASTED International Conference February 20 - 21, 2017 Innsbruck, Austria Modelling, Identification and Control (MIC 2017) DOI: 10.2316/P.2017.848-054 178