Synthetic Energy Time-Series for Seasonal Facility Load Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for optimizing electrical energy consumption in industrial sites face challenges when real-time data is unavailable, often relying on generic or repeated data that neglects seasonal trends, leading to inaccurate energy time-series estimates.
Innovation Solution
A system and method for synthesizing energy time-series data using a time-series generator that receives key attributes of a facility and generates a plausible time-series by modifying or creating a new series based on reference data, employing machine learning models and similarity metrics to ensure accuracy and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If real consumption data is repeated to fill a complete year, then data availability is improved, but seasonal trends are lost
Solution Approach 1:
The system performs preliminary actions by collecting and storing multiple years of historical consumption data before synthesis is needed. This pre-collected diverse data pool enables the synthesis algorithm to reconstruct seasonal patterns without simply repeating limited data, thus preserving seasonal trends while ensuring data availability.
Solution Approach 2:
Instead of copying limited real data repeatedly, the system creates synthetic copies through algorithmic generation that mimics the statistical properties and seasonal patterns of historical data. This produces multiple realistic variations that maintain seasonal trends while providing sufficient data quantity for complete year analysis.
2Ease of manufacture
If a generic time-series is selected from a database, then data acquisition is simplified, but accuracy for specific facilities is reduced
Solution Approach 1:
The system applies local quality by incorporating facility-specific attributes (size, type, location, equipment) into the synthesis process. Rather than using a single generic time-series, it generates customized time-series for each facility by weighting and combining reference data according to the specific facility's characteristics, thus maintaining both simplicity and accuracy.
Solution Approach 2:
The system changes parameters by adjusting the synthesis algorithm based on facility-specific input parameters such as facility size, sector, location, and equipment types. These parameter variations transform generic reference data into customized time-series that accurately reflect the specific facility's consumption patterns while keeping the overall process automated and simple.
3Measurement precision
If detailed data is required for accurate optimization, then data collection complexity increases, but the patent enables accurate estimation without detailed data
Solution Approach 1:
The system introduces an intermediary layer - the synthesis algorithm - that bridges the gap between limited input data (facility attributes) and the required detailed time-series output. This intermediary processes simple input parameters through sophisticated algorithms to generate detailed consumption patterns, achieving optimization accuracy without requiring direct collection of detailed operational data.
Solution Approach 2:
The system replaces the mechanical approach of manually collecting detailed operational data with an automated computational system. The synthesis algorithm automatically generates detailed time-series from minimal input parameters, substituting complex data collection processes with intelligent data generation, thus maintaining accuracy while reducing complexity.
Data Source
AI summary
A system and method for synthesizing energy time-series data for a facility includes a time-series generator configured to receive an input set of attributes comprising attributes characterizing the facility, and to output a synthesized time-series representing estimated energy data for the facility. The synthesized time-series are generated on the basis of the input set of attributes and one or more reference time-series associated with respective reference sets of attributes.