Energy Demand Prediction Using Representative Building Clusters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Demand side management systems face challenges in accurately predicting energy demand and adjusting consumption in community settings due to intentional consumption adjustments and social constraints, making it difficult to distinguish actual energy consumption from real demand.
Innovation Solution
A system that predicts future energy demand by identifying non-destination demanders with similar environmental conditions and using their actual energy consumption data to estimate community demand, then generates targeted consumption adjustment requests to selected demanders.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If demand side management is implemented to adjust energy consumption, then load concentration is reduced and surplus electricity is utilized, but accurate demand prediction becomes difficult because actual consumption data no longer reflects true demand
Solution Approach 1:
The patent segments the community into multiple zones and further divides zones into building clusters. By selecting representative buildings within each cluster as destination demanders, the system can predict overall community demand based on consumption patterns of these representatives, maintaining prediction accuracy while implementing demand side management.
Solution Approach 2:
The patent introduces zone-level aggregation as an intermediary layer between individual building consumption and community-wide demand prediction. By aggregating consumption data at the zone level and using representative buildings, the system creates a mediator that preserves demand information despite consumption adjustments by individual demanders.
2Productivity
If consumption adjustment requests are sent to all demanders, then community energy consumption is optimized, but it becomes impossible to distinguish actual consumption from adjusted consumption for prediction purposes
Solution Approach 1:
The patent extracts representative demanders from each building cluster to serve as destination demanders for consumption adjustment requests. By selecting only these representatives rather than all demanders, the system maintains unadjusted consumption data from non-destination demanders that can be used to infer true demand patterns while still achieving community-wide optimization.
3Ease of manufacture
If demand prediction is based on past consumption data from demanders who received adjustment requests, then the prediction model becomes inaccurate because the data reflects adjusted rather than actual demand
Solution Approach 1:
The patent dynamically selects destination demanders for each prediction cycle based on their representative status in building clusters. This dynamic selection ensures that consumption data from non-destination demanders (who did not receive adjustment requests) is used for prediction, maintaining data accuracy while allowing the system to adapt to changing community consumption patterns.
Data Source
AI summary
Energy demand in an entire community including multiple demanders is properly predicted and energy consumption thereof is properly controlled. Every day, some demanders are selected from the community and a request to adjust energy consumption for the following day is sent to the selected demanders. To decide a content of the request, energy demand for the following day is predicted. In this case, demanders which did not receive a request on past days are specified, and demand in the community for the following day is predicted based on the actual energy consumption by those demanders on the past days.