Power Management Grouping by Prediction Error Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current power management systems face limitations in predicting power consumption accurately, leading to prediction errors that hinder the success rate of Demand Response (DR) mechanisms, where power providers request reduced consumption from facilities connected to the grid.
Innovation Solution
A management method that estimates error patterns in power consumption predictions for each entity and groups them based on positive and negative errors to minimize the total prediction error, thereby enhancing the success rate of DR by combining entities with opposite prediction errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single entity is selected for power reduction request, then the prediction accuracy for that entity can be optimized, but the total prediction error cannot be minimized due to individual entity prediction limitations
Solution Approach 1:
The patent combines multiple entities into groups based on their error patterns (positive or negative prediction errors). By aggregating entities with opposite error tendencies, the total prediction error for the group is minimized, thereby improving DR success rate while maintaining manageable prediction complexity.
2Reliability
If entities are grouped to minimize total prediction error, then the DR success rate increases, but the complexity of error pattern estimation and group determination increases
Solution Approach 1:
The patent segments the error estimation process into distinct steps: first estimating error patterns for individual entities, then determining groups based on these patterns. This segmentation allows the complex problem to be solved in manageable stages, reducing the overall computational burden while achieving minimal total prediction error.
Solution Approach 2:
The patent performs preliminary error pattern estimation for each entity before forming groups. By pre-characterizing each entity's error tendency (positive or negative), the subsequent grouping process becomes simpler and more efficient, as entities can be systematically combined based on their pre-assessed error patterns rather than requiring complex real-time optimization.
Data Source
AI summary
A management method includes a step A of estimating, for each of the plurality of entities, an error pattern related to a prediction error occurring when a power consumption amount in a future period is predicted; and a step B of determining a group of entities from among the plurality of entities based on the error pattern, wherein the group is a destination of a power reduction request for reducing the power consumption amount. The step A includes a step of estimating whether the prediction error is a positive error or a negative error. The step B includes a step of determining the group such that a total prediction error of the group is minimized, by combining an entity of which the prediction error is estimated to be a positive error and an entity of which the prediction error is estimated to be a negative error.