As the proportion of new energy power generation continues to increase

the stability of grid frequency is severely challenged

and the role of conventional thermal power units in grid frequency regulation has become increasingly prominent. However

the adjustment rate and accuracy of some thermal power units are difficult to meet the demand of grid load fluctuations. Therefore

a response performance optimization strategy for flywheelthermal power system automatic generation control based on load forecasting was proposed. Firstly

the load is predicted

using the tree-based pipeline optimization tool TPOT library to automatically machine learning to match and train the load regression prediction model

and the automatic generation control day-ahead planned value is introduced into the training data to reduce the prediction error. Then

according to the load prediction value and the current flywheel system

with the optimization goal of minimizing the regulation rate of thermal power units

the flywheel energy storage system is acted firstly in load distribution

and the state of charge of the flywheel is adjusted meanwhile. Finally

a simulation experiment is carried out based on the actual operation data of a power plant in Hubei

and the experimental results prove that the proposed method can effectively improve the frequency modulation performance of thermal power units.