A MULTI-AGENT SYSTEM-BASED REFERENCE GOVERNOR FOR MULTIOBJECTIVE POWER PLANT OPERATION

A large-scale power plant requires optimal set-points, namely references, for optimal operation. In a 600 MW oil-fired drum-type boiler power unit, the set-points considered are for the main steam pressure and reheater/superheater steam temperatures. The set-points should be mapped with varying unit load demand and satisfy the conflicting operation requirements of the power plant. The conflicting requirements are tackled by multiobjective optimization problem to produce the optimal mapping. In general, the set-points obtained by using a fixed nonlinear function cannot provide optimal power plant operation. This paper presents a methodology, Multi-Agent System-based Reference Governor (MAS-RG), to realize the optimal mapping by searching for the best solution to the multiobjective optimization problem. In searching for the optimal set-points, a heuristic optimization tool, Particle Swarm Optimization (PSO), is utilized to solve the multiobjective optimization problem. Moreover, the approach provides the means to specify optimal set-points for controllers under a diversity of operating scenarios on-line.