Power Generation Dispatching for Environmental Protection Based on Recursive Neural Network and Ant Colony Optimization Algorithm

We propose a new regression model to fit power generation and emission data (SO2, NOx, and soot) by using recurrent neural network (RNN). On the basis of the regression model and ant colony optimization (ACO), we design a real-time power generation dispatching algorithm for reducing total pollutant emissions under the premise of completing real-time power generation and achieving energy saving and emission reduction. We evaluate our proposal by using the real electricity data of Anhui Electric Power. Experimental results show the effectiveness of our method.