Renewable Plant Storage Forecasting for Accurate Energy Demand Dispatch
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Renewable power plants with storage systems struggle to accurately predict energy demand, limiting their participation in wholesale electricity markets due to penalties for non-fulfillment of actual energy demand.
Innovation Solution
A method using machine learning techniques, including uncertainty post-processing and predictive modeling, to enhance the accuracy of energy demand prediction, allowing for intelligent dispatch planning and integration with conventional power plants.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If known prediction methods are used, then the system is simple to operate, but the prediction accuracy is too low to participate in wholesale electricity markets
Solution Approach 1:
The patent combines optimization algorithms that generate forecasted variables with machine learning techniques for uncertainty post-processing. This merging of two different computational approaches creates a hybrid system that achieves high prediction accuracy required for wholesale electricity market participation while managing the complexity through integrated automated processing.
Solution Approach 2:
The patent introduces an intermediary machine learning component that processes the output of optimization algorithms. This intermediary performs uncertainty post-processing by comparing forecasted variables with historical data and perfect forecasts, thereby enhancing prediction accuracy without requiring complete system redesign.
2Reliability
If optimization algorithms are used to generate forecasted variables, then the energy scheduling becomes more accurate, but the computational complexity increases
Solution Approach 1:
The patent performs preliminary actions by generating forecasted variables using optimization algorithms before final energy scheduling decisions. This preliminary optimization creates a robust foundation that improves scheduling reliability, while the subsequent machine learning post-processing refines these predictions without requiring complete recomputation.
Solution Approach 2:
The patent implements feedback mechanisms where machine learning techniques analyze the relationship between forecasted variables and actual outcomes using historical data. This feedback loop continuously improves prediction reliability by adjusting uncertainty post-processing based on past performance patterns, reducing the need for increasingly complex optimization algorithms.
3Measurement precision
If historical data is used for training machine learning models, then the prediction becomes more accurate, but the data processing time increases
Solution Approach 1:
The patent applies partial action by using historical data selectively for uncertainty post-processing rather than complete retraining. The machine learning model leverages historical patterns to adjust forecasts, achieving high accuracy without processing entire historical datasets each time, thus reducing data processing time while maintaining prediction quality.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In order to provide a reliable method for predicting an energy demand for operating a renewable power plant with a storage with improved accuracy, a method is proposed that comprises: - generating a set of forecasted variables (102) based on at least one set of uncertainty-based variables (108) by using an optimization algorithm (107); and - uncertainty post-processing (105) of the set of forecasted variables (102) using machine learning techniques (104), wherein the machine learning techniques (104) are performed by a machine learning device (120) that makes use of historical data (123) relating to at least one energy demand according to a perfect forecast (148) in the past.