Power Load Prediction Using Similar-Day Temperature Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current load prediction methods in power systems lack accuracy, particularly in high-temperature weather conditions, due to the influence of air temperature on load prediction results.
Innovation Solution
A load prediction method that determines a similar day from historical days based on air temperature differences, using feature vectors and regression equations to calculate load differences, thereby improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a pre-trained prediction model is used for load prediction, then the prediction process is simple and fast, but the prediction accuracy is insufficient especially in high-temperature weather conditions
Solution Approach 1:
The prediction method is segmented into two distinct parts: (1) a pre-trained prediction model for baseline predictions, and (2) a temperature correction module that activates when high temperature conditions are detected. This segmentation allows the system to use the simple pre-trained model for normal conditions while applying the more complex correction only when needed, thus improving accuracy without permanently increasing overall system complexity.
Solution Approach 2:
The prediction system dynamically adjusts its complexity based on temperature conditions. When the temperature exceeds a threshold, the system transitions from using only the pre-trained model to incorporating the temperature correction mechanism. This dynamic adaptation allows the system to maintain simplicity during normal operation while achieving higher accuracy during critical high-temperature periods.
2Measurement precision
If air temperature difference is considered in load prediction, then prediction accuracy in high-temperature weather is improved, but the calculation complexity increases
Solution Approach 1:
The system performs preliminary actions by pre-training the prediction model on historical data that includes temperature variations. This pre-training embeds temperature-related patterns into the model's weights, so that during actual prediction, the model can automatically account for temperature effects without requiring complex real-time calculations. The heavy computational work is done in advance during the training phase.
Solution Approach 2:
The patent introduces a temperature correction mechanism as an intermediary between the pre-trained model and the final prediction result. This intermediary component specifically handles the temperature difference calculations and adjusts the baseline prediction accordingly, isolating the complex temperature-related calculations from the main prediction model and making the overall system more manageable.
Data Source
AI summary
A load prediction method includes determining for a current prediction time point of a prediction day, a similar days of a prediction day from historical days before the prediction day; acquiring an air temperature difference between an air temperature at the time point of the prediction day and an air temperature at the time point of the similar day; determining a difference of loads at the time point of the prediction day and the similar day according to the air temperature difference; and determining a predicted load value at the time point of the prediction day according to a load value at the time point of the similar day and the load difference.