A Novel Intelligent Monitor for Manufacturing Processes

Abstract

The development of condition monitoring in continuous-time mass production is a traditional but critical topic owing to its role in increasing cost-effectiveness. A new method for intelligent manufacturing processes monitor was presented using the newly developed learning approach, Support Vector Regression, and related noise density model for efficient and effective modeling. A new similarity measurement method was introduced resulting in a real-time system developed. The system was applied for certain manufacturing process condition monitoring. The experimental results showed that the proposed method is substantially more effective than others. In addition, the new monitor requires only normal condition samples for implementation, an attractive feature for shop floor applications. The novel system ensures the automatic and intelligent operation, and it has potential applications on the other manufacturing processes.