Regional Energy Power Forecasting With Weather Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting regional short-term energy power face inaccuracies due to errors and fluctuations in weather forecasts, particularly when relying solely on meteorological data or historical data, which affects the stability of new energy power generation systems and power grid planning.
Innovation Solution
A method that combines weight assignment for errors between measured and network meteorological data with historical data, using real-time and historical error data to correct weather forecasts and improve prediction accuracy by integrating them with a trend in weather forecasting.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If weather forecast data is directly used for calculation, then the method is simple and easy to operate, but the prediction accuracy is reduced due to errors in weather forecast values
Solution Approach 1:
The patent introduces an intermediary error correction mechanism that mediates between the simple weather forecast data and the final power/load prediction. Historical error data and real-time error data serve as intermediaries to correct the weather forecast values before they are used in calculations, thereby maintaining operational simplicity while improving prediction accuracy through error compensation.
Solution Approach 2:
The patent implements feedback by using historical error data from previous predictions and real-time error data from current measurements to continuously correct weather forecast values. This feedback loop allows the system to learn from past inaccuracies and adjust current predictions, improving overall accuracy without complicating the basic prediction workflow.
2Measurement precision
If a large amount of historical data is input into a neural network for calculation, then the method can sift corresponding meteorological data, but the complexity of the method increases and high requirements for historical data quality are needed
Solution Approach 1:
The patent extracts only the essential error components from historical data rather than using entire historical datasets for neural network training. By extracting and utilizing only the error differences between predicted and actual values, the system achieves accurate corrections without the complexity of processing large volumes of raw historical meteorological data through neural networks.
Solution Approach 2:
The patent transforms the prediction approach by changing from direct prediction of meteorological parameters to prediction of error parameters. Instead of using neural networks to predict weather values directly from historical data, the system predicts error values and applies them as corrections, thereby reducing data requirements and computational complexity while maintaining accuracy.
3Reliability
If historical data is used to predict future weather, then the method can account for weather patterns, but the accuracy is reduced due to randomness of weather changes
Solution Approach 1:
The patent performs preliminary action by calculating and storing error data in advance from historical measurements and forecasts. This pre-computed error data is then applied as corrections to current predictions, allowing the system to account for historical weather patterns and their associated errors without relying on direct historical weather predictions, thereby handling weather randomness more effectively.
Solution Approach 2:
The patent converts the harmful effect of weather randomness and forecast errors into a beneficial correction mechanism. By analyzing historical errors caused by weather variability and using them as correction factors, the system transforms the problem of weather unpredictability into a solution that actively compensates for such uncertainties in real-time predictions.
Data Source
AI summary
A method and system for predicting regional short-term energy power by taking weather into consideration includes: obtaining meteorological data of all moments in a set time in the future through a network; extracting respectively, from a historical database according to the obtained meteorological data, historical weather station meteorological data, historical network API meteorological data, and historical measured power generation power data within a set time period that meet meteorological conditions corresponding to all the moments; obtaining historical total error data; obtaining real-time error meteorological data; obtaining total error meteorological data; combining the obtained meteorological data of all the moments in the set time in the future with total error meteorological data of all the moments to obtain predicted meteorological data; obtaining predicted power data according to the predicted meteorological data; and optimizing an energy generation plan of a system according to the obtained predicted power data.