Wind speed prediction method combining SWLSTM with GPR

Abstract

The invention discloses a wind speed prediction method combining a shared weight long short-term memory network (SWLSTM) with Gaussian process regression (GPR). The method mainly comprises the following steps of: simplifying the structure of a standard long short-term memory network (LSTM) by adopting a shared weight; using an Adam optimization algorithm combined with a mini-batch mechanism to train SWLSTM to acquire a wind speed point prediction result with high accuracy; using the point prediction result obtained by the SWLSTM as the input of the GPR, and performing secondary prediction to obtain a wind speed probability prediction result; and selecting a confidence coefficient, and obtaining a wind speed interval prediction result under the corresponding confidence coefficient through Gaussian distribution. According to the prediction method disclosed by the invention, the training time of the LSTM is shortened by sharing the weight, and the SWLSTM is enabled to be capable of performing probability prediction and interval prediction by combining the GPR. SWLSTM-GPR can obtain a high-precision wind speed point prediction result, a suitable wind speed interval prediction result and reliable wind speed probability prediction distribution, which is of great significance for wind power planning and application.

Description

technical field [0001] The present invention relates to the technical field of wind speed prediction, and more specifically, to a wind speed prediction method based on Shared Weight Long-Short Term Memory (SWLSTM) combined with Gaussian Process Regression (GPR). Background technique [0002] Wind energy is a clean and economical renewable energy. Wind speed is the most influential factor for wind power production. High-precision and reliable wind speed prediction plays an important role in various aspects of wind power planning, dispatching and decision-making management, and is of great significance to the rational use of wind energy resources. However, the formation process of wind is affected by factors such as air pressure, geographical location, and the rotation of the earth, resulting in highly nonlinear, fluctuating, and uncertain characteristics of wind speed. These characteristics make wind speed prediction difficult, and traditional machine learning methods have ...

AI Technical Summary

Problems solved by technology

[0003] In view of the defects of the prior art, the purpose of the present invention is to solve the technical problem that the existing LSTM-based wind s

Images