Predicting Surface Roughness in Grinding using Neural Networks

摘要: High rates of manufactured items have been machined by grinding at some stage their production process, or processed machines whose precision is a direct result abrasive operations. However, even being the process most used in industry for obtaining high level surface quality, it remains as one difficult and least understood processes (Wang et al., 2005). That maybe has origin mistaken faith extremely complex to be due large number cutting edges irregular geometry, speed, very small depth cut which varies from grain grain. In addition, according (Haussi & Diniz, 2003), indicated when workpiece demands good surface, dimensional geometrical quality. Thus, usually last steps machining operations chain. When reaches this point, aggregated value, makes possible rejection expensive. Monitoring mandatory optimization control. Acoustic emission (AE) become an increasingly popular monitoring technique. The sensors are inexpensive, easy mount, analog signal processing comparatively simple, but techniques extracting reliable information signals still lacking (Hundt 1997; Aguiar 2002). Electrical power also largely researches. can monitored either electric current motor product between voltage signals, gives electrical consumed motor. estimate force easily obtained if model available (Aguiar Some researchers shown acoustic combined provide significant results phenomena 2002; Dotto 2006; Kwak Ha, 2004; 2006). Neural network attracted special interest research owing its functions learning, interpolation, pattern recognition, classification. Various examples applications into engineering field reported 2005; Wang 2001). According 2005), roughness important factors assessing determining quality part. practical, predicting controlling O pe n A cc es s D ab e w .ite ch on lin e. co m