A Bus Load Forecasting Method Based on Multi-xgBoost Model Fusion

Abstract

The invention discloses a bus load prediction method based on the fusion of multiple XGBoost models, which includes the following steps: firstly, preprocessing the bus load data and related data, and constructing a sample set; secondly, constructing a hierarchical prediction system, a total of 2 In the first layer, N different XGBoost models are established, and in the second layer, one XGBoost model is established, and the output of the N XGBoost models in the first layer is used as its input, that is, the XGBoost model of the first layer is used for one-time learning. Then, the XGBoost model of the second layer performs secondary learning on the learning results of the first layer; the particle swarm optimization (PSO) is used to optimize the parameters of the layered forecasting system; finally, the sample set is used for training and testing, and the bus load forecasting results are output. The method of the invention strengthens the learning effect, improves the generalization ability, and is suitable for solving the prediction problem of the bus load with strong randomness.

Description

technical field [0001] The invention relates to the technical field of bus load forecasting, in particular to a bus load forecasting method based on fusion of multiple XGBoost models. Background technique [0002] The bus load refers to the sum of the terminal loads supplied by the main transformer of the substation to a certain power supply area. Since the power supply area is usually small, the bus load level is low, generally only a few hundred megawatts. In addition, due to differences in the nature of users in different power supply areas, for example, some areas have more residential loads, while some areas have more industrial loads, so the load components of each bus are often different, making forecasting more difficult. [0003] In recent years, machine learning technology has developed rapidly and is widely used in the field of load forecasting. Extreme Gradient Boosting (XGBoost) is a machine learning model implemented based on gradient boosting, and it perform...

AI Technical Summary

Problems solved by technology

[0004] In summary, for the bus load forecasting problem, a single machine learning model is difficult to effectively deal with, so it can be considered to solve the bus load forecasting problem by combining the model fusion method

Images