Wind Power Forecasting Method and System Based on Deep Deterministic Strategy Gradient Algorithm
A technology of wind power forecasting and gradient algorithm, which is applied in forecasting, wind power generation, calculation, etc., can solve the problems of changing weights and not being able to extract effectively, and achieve the effect of improving forecasting accuracy
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0057] According to an embodiment of the present invention, a wind power variable weight combination forecasting method based on a deep deterministic strategy gradient algorithm is disclosed, refer to figure 1 , including the following steps:
[0058] (1) Obtain historical wind power series and forecast data of wind speed and direction at the moment to be forecast;
[0059] (2) Inputting the data into the trained support vector machine regression model, artificial neural network model and extreme gradient boosting tree model respectively to obtain the predicted value of wind power of each sub-model;
[0060] (3) In the process of exploratory learning, the deep deterministic policy gradient algorithm perceives the current state from the prediction environment, and outputs the weighted action with exploration noise to the combined model according to the current strategy, and the algorithm continuously updates and optimizes the update strategy according to the returned rewards; ...
Embodiment 2
[0159] A wind power prediction system based on a deep deterministic strategy gradient algorithm, comprising:
[0160] a data acquisition module, configured to acquire relevant data of wind power prediction;
[0161] The forecasting sub-model forecasting module is configured to input the obtained data into the trained multiple different forecasting sub-models respectively, and obtain the wind power forecast value of the corresponding forecasting sub-model;
[0162] a combined model building module configured to construct a combined model, the combined model is a combination of each prediction sub-model, and each prediction sub-model is assigned its own weight;
[0163] The combined model optimization module is configured to use the deep deterministic policy gradient algorithm to perceive the current state from the prediction environment at the moment to be predicted, determine the strategy according to the current state, obtain the weight with exploration noise, and assign it to ...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More