Expert system for control purpose based on CLIPS

The use of AI techniques in complex process control of industrial environments, introduces some problems. The first is related to which kind of tools can be used and their match to system requirements. The next one is how to incorporate new features if needed and how to integrate them. The last one is related to the final implementation. Expert systems available in public domain source code seems a good solution in order to reach the advantages of an ES developed specially for the application without spending time developing low level ES features. An ES in source code allows us to select some parts of an ES kernel useful for the particular application. Actually we can find commercial ES in source code like CLIPS. We describe the problem from the point of view of developing ES with CLIPS. This tool is a complete ES language written in C code, it uses the RETE fast pattern matching algorithm to implement forward chaining inference engine. The availability of the source code allows us to cut the parts not used in a particular application and recompile it in a specific machine for control purposes. Therefore we can link the modified code with real time run-time libraries, obtaining applications dealing with real time constraints. Based on this tool, an intelligent controller has been developed which uses fuzzy logic for uncertainty handling