부산대학교 도서관

Monthly load forecast is a prerequisite for making monthly power generation plan and maintenance arrangement. Scientific forecasting is the basis and guarantee for correct decision-making. Previous forecasting methods do not study the mechanism of load formation deeply enough, and the complexity of historical load data is high, which makes it difficult to forecast the load accurately. To address this difficulty, this paper proposes a method of monthly power load interval forecasting based on mechanism analysis. The proposed method decomposes the complex load data into the two parts of trend load and sudden change load according to the causes, and statistically distributes the error distribution law to give the prediction intervals. This paper first explains the specific meaning and causes of the formation of each part of the load, and discusses the corresponding prediction model in detail. Finally, an example of load prediction in a region of Hainan is provided. The proposed method is close to the actual situation and has strong engineering practicability.