Flexible load adjustable power evaluation method based on incremental learning

The invention discloses a flexible load adjustable power evaluation method based on incremental learning. The method comprises the following steps: constructing an air conditioner load group adjustable power calculation model; historical data are obtained and preprocessed, and related characteristic factors of adjustable power are found; the processed data are input into an air conditioner load group adjustable power calculation model based on E2EIL, and the model performance is improved; and air conditioner load data are collected, and the air conditioner load aggregation cluster adjustable power is calculated and output through the air conditioner load group adjustable power model. The problem of error accumulation between learning stages can be avoided; and the network structure can be dynamically adjusted according to the number of the new categories, so that the limitation of a fixed network structure is avoided, and various incremental learning tasks are better adapted. Knowledge learned in the past can