Real-time Optimal Load Dispatching of Power Plant Based on Data Mining Method

Abstract A solution to real-time optimal load dispatching problem of power plant is introduced in this paper. Based on analysis of amounts of operating data, it is proved that the characteristics curve of unit coal consumption rate (CCR) against load are constructed by discrete points and it can't be deduced by linear regression. At the default condition of accurate thermodynamic calculation, the factors impacting unit CCR for power supply are studied at different operating conditions. Data mining method is employed to predict CCR for power supply at different operating points and environmental condition based on history and real-time data. Thus, discrete mathematical model between load and CCR is build up for optimal load dispatching (OLD) among the units. Global optimal solution can be approached by dynamic programming method for the reason of the low dimension (number of units in one plant) of this issue now. The method proposed in this paper is applied in a power plant of Datang group and it is proved to be practical and effective.