During the formulation and adjustment of the annual power generation plan of cogeneration units

the local power plants frequently require adjusting the plan to meet the heating requirements. To solve this problem

this paper proposes a method for predicting the electric load. On the basis of the local thermal load

this method uses hypothesis test method to extract the thermal load of the power plant that needs to regulate the power generation plan

and forecast the thermal load by the improved Elman neural network method

then determines the generating capacity of the power plant via the grey neural network method. This method is of vital significance to reasonably determine the power generation plan and control the purchase cost according to the heating demand put forward by the power plant.