Predictive Control Based on an Auto-regressive Neuro-fuzzy Model Applied to the Steam Generator Startup Process at a Fossil Power Plant

摘要

This paper presents an application of artificial intelligence techniques to the improvement of the operation of a thermoelectric unit. The capacity for empirical learning gained from artificial intelligence systems was utilized in the development of the strategy. A neuro-fuzzy model for the steam generator startup process is obtained from experimental data. Ultimately, the neuro- fuzzy model is combined with a predictive control algorithm to produce a control strategy for the heating stage of the steam generator. This provides the operators at the fossil power plant with the necessary information to efficiently accomplish the heating process. The information gained from the control strategy is not directly applied to an automatic control scheme; it is presented to the operator who then decides on its application. Therefore, in this way the information is used to develop a strategy that takes into consideration the personal capacity and the working routine of the operator. The simulation tests that were carried out demonstrated the feasibility and the beneficial results that can be obtained from the application of any of the three variants of predictive control proposed in this paper.