An Intelligent Multiagent Approach for Selection of Grinding Conditions

The advantages of a multi-agent approach are presented for the selection of grinding conditions. The agents consist of case based reasoning, neural network reasoning and rule based reasoning. Case based reasoning is employed as the main problem-solving agent to select combinations of the grinding wheel and values of control parameters. Rule based reasoning is employed where relevant data are not available in the case base. A neural network is employed to select a grinding wheel if required. The operator makes the final decision about the wheel or the values of control parameters. The multi-agent approach combines the strengths of the different agents employed, to generate hybrid solutions and overcomes the limitations of any single approach. A blackboard method was used as the means of integrating the multi-agent system. The system works as expected and demonstrates the potential of using artificial intelligence for selection of grinding conditions, as well as the capability to develop a powerful database by learning from experience.