DESIGN AND IMPLEMENTATION OF AGENT BASED SPC MONITORING SYSTEM

Although the benefits of statistical process control (SPC) have been already verified in the manufacturing industries, it still highly depends on manual diagnosis and analysis which consumes considerable time and effort of the engineers. This paper tries to overcome the limitations through automation of SPC monitoring activities through intelligent software agents. The requirements of automation are derived from main SPC monitoring functions, and two types of specified agents were designed. The simple agent models are for the basic functionalities such as data gathering, pattern recognition, and diagnosis of the process status. The composite agent models which consist of simple or other composite agents represent SPC actions such as multistage data analysis or continuous/dynamic monitoring activities. Adoption of the agent concept provides the design phase which enables to add a new agent type (complement of a SPC function) and modify the properties of agents in operations (dynamic reconfiguration). A prototype agent system which connected to an external database with sample manufacturing process data was implemented, and it shows that the specified functions of the proposed agents are successfully validated.