Computational Intelligence Applied to the Automatic Monitoring of Dressing Operations in an Industrial CNC Machine
In manufacturing, grinding is the process that shapes very hard work pieces with a high degree of dimensional accuracy and surface finish. The efficiency of the grinding process is regarded as a very important issue in the modern and competitive metal-mec
- PDF / 537,194 Bytes
- 20 Pages / 439.2 x 666 pts Page_size
- 30 Downloads / 183 Views
2
3
Department of Electrical Engineering, Federal University of Cear´ a – CT, Caixa Postal 6001 – CEP - 60450-760, Fortaleza, Cear´ a, Brazil Department of Computer Science, University of S˜ ao Paulo, Caixa Postal 668,13560-970, S˜ ao Carlos, SP, Brazil Department of Production Engineering, University of S˜ao Paulo, Av. Trabalhador S˜ aocarlense, 400, 13566-590, S˜ ao Carlos, SP, Brazil
1 Introduction In manufacturing, grinding is the process that shapes very hard work pieces with a high degree of dimensional accuracy and surface finish. The efficiency of the grinding process is regarded as a very important issue in the modern and competitive metal-mechanic industry, since it usually represents the major portion of processing costs [1–8]. The grinding process is strongly dependent on the topography surface of the grinding wheel, an expendable wheel that carries an abrasive compound on its periphery, since it is the cutting tool in grinding operations [1–3, 9–11]. The process responsible for preparing the topography surface of the grinding wheel is named dressing [2, 9, 12]. This process removes the current layer of abrasive, leading to a fresh and sharp surface. Thus, the dressing process has an important effect on the efficiency of the grinding process, because the quality of the cutting tool directly affects the quality of the final product [2]. This research targets a common problem in industrial grinding operations. Every time a grinding wheel is changed, or after an automatic balancing procedure, it is necessary to perform dressing operations in order to get a concentric working surface. The number of dressing strokes is normally determined by trial and error. If an excessive number of dressing strokes is executed, the grinding wheel’s life-cycle is reduced and that implies an economic loss. If an insufficient number of dressing strokes is executed, the quality of the grinding process will be compromised and it also leads to an economic loss caused by work pieces. The solution proposed in this paper is to develop an automatic monitoring dressing system that provides support to execute the just-right A.P. de Souza Braga et al.: Computational Intelligence Applied to the Automatic Monitoring of Dressing Operations in an Industrial CNC Machine, Studies in Computational Intelligence (SCI) 116, 249–268 (2008) c Springer-Verlag Berlin Heidelberg 2008 www.springerlink.com
250
A.P. de Souza Braga et al.
number of dressing strokes. As a result, it guarantees the high quality of final products, its efficient use of grinding wheel’s life-cycle and eventually a lower cost [1, 3, 9, 10]. An automatic monitoring dressing system is often regarded as a nontrivial issue [2, 3, 6, 9, 12]. The particular nature of the grinding wheel, which has its abrasive grains randomly spaced on the work surface, makes it more difficult than a pure mathematics problem. A model-based approach for the monitoring problem has to include many functional parameters that could not be easily measured [4]. Thus, we propose to monitor dressing operations throug
Data Loading...
