On the development of industrial-grade intelligent supervisory systems for power plant operation

Development of comprehensive intelligent systems to achieve total power plant autonomy at once is a formidable task, for which no guarantee of success in performance, time and budget exists. Instead, it is proposed to develop small-scale intelligent systems, tailored to specific supervisory applications, which could be eventually integrated into larger systems to incrementally achieve plant autonomy. This paper presents a low-cost general-purpose platform to develop such industrial-grade intelligent supervisory systems. Major issues solved include selection of: a) the generic functions to be performed by the system to achieve the desired autonomy, b) the paradigm to merge the generic functions to attain coherent behavior in the intelligent system domain, c) the AI and CI software tools best suited to program the generic functions as software agents, and d) the software platform to integrate the intelligent system software agents, As an example, an intelligent supervisory system for a gas turbogenerator is also presented