With the increasing random disturbance of power grid caused by changes in electric utility structure and large-scale grid-connection of renewable energy

the grid side needs to strengthen the overall dispatch of controllable and tunable power saurce(like thermal power units) to reduce the short-term uncontrollable risk. An integrated method combining wavelet transform and large-scale autoregressive integrated moving average(ARIMA) model is proposed in this study for load forecasting of thermal power units. Concerning about the characteristics of load signals of thermal power units

firstly

the historical load signals are decomposed into regular signals with good regularity and the detail signals with strong randomness by wavelet transform. Then the ARIMA model is established

fitted and forecast separately. Finally

the final prediction is obtained by weighted summation. The forecasting and simulation of load data of a unit under normal and abnormal conditions shows that

the method in this study has higher precision than the ARIMA model.