Data mining based plant-level load dispatching strategy for the coal-fired power plant coal-saving: A case study

One of the most urgent problems facing coal-fired power plants in China today is the coal-waste because several units in one plant experience a partial rated output situation at the same time, which may increase the coal consumption of the power plant. Here we proposed a new hybrid methodology for plant-level load dispatching to minimize coal consumption for the coal-fired power plant. The proposed strategy consists of two parts: A part was to establish the coal consumption model to find the relationship between the coal rate, ambient temperature and load based on PSO- SVM; the other one based on GA to find the load-dispatching optimal solution to minimize the average coal rate of the whole power plant. Additionally, grey correlation analysis was employed to pick up the most relevant input variables to reduce the complexity of the regression model. This work is based on continuously measured supervision information system data from an actual coal-fired power plant with three different types of capacity units in China. Results showed that the proposed strategy performances better than the normal AGC to single unit for power plant energy saving. It is of significant for coal-fired power plant to reduce the coal consumption.