Agents Modeling Experience Applied To Control Of Semi-Continuous Production Process

The lack of proper analytical models of some production processes prevents us from obtaining proper values of process parameters by simply computing optimal values. Possible solutions of control problems in such areas of industrial processes can be found using certain methods from the domain of artificial intelligence: neural networks, fuzzy logic, expert systems, or evolutionary algorithms. Presented in this work, a solution to such a control problem is an alternative approach that combines control of the industrial process with learning based on production results. By formulating the main assumptions of the proposed methodology, decision processes of a human operator using his experience are taken into consideration. The researched model of using and gathering experience of human beings is designed with the contribution of agent technology. The presented solution of the control problem coincides with case-based reasoning (CBR) methodology.