It is difficult to accurately determine the heating capacity of local thermal power plants and the generation capacity of cogeneration units under the principle of "determine power by heat". To solve this problem

a calculation method of the generation capacity of cogeneration units based on improved neural network and energy conservation method is proposed. Firstly

the optimal number of neurons in hidden layer of BP neural network(BPNN) is determined according to empirical formula and trial method. Then

self-adaptive genetic algorithm(AGA) is used to optimize the initial weights and thresholds of the BPNN

thus to enhance the global search ability of the network

and build up the AGA-BPNN heat supply prediction model to achieve accurate prediction of heat supply. Finally

the expression of the maximum and minimum generation capacity of cogeneration units based on the principle of "determine power by heat" is derived by energy conservation method

which can accurately calculate the range of generation capacity of cogeneration units under given residential heating and industrial heating. The above method can provide a reliable reference for the dispatching department to formulate a reasonable power generation plan for thermal power plants.