Polynomial Surface Fitting for Long-Term Resource Demand Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting resource demands, such as electrical power and water supply, lack accuracy and adaptability beyond short time spans, particularly failing to provide reliable forecasts for 12 to 24 months, which is crucial for effective distribution system planning and capital budget allocation.
Innovation Solution
A computer-based method that generates a topologic space and polynomial equation from sequential resource usage data, allowing for accurate prediction of future events by plotting day and hour data in a three-dimensional space and using regression analysis to derive coefficients for a predictive polynomial equation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional forecasting algorithms (fuzzy logic, neural nets, stochastics) are used for short-term prediction, then accuracy is high, but accuracy drops dramatically for predictions beyond one to two weeks
Solution Approach 1:
The patent transforms the time-series prediction problem into a spatial surface fitting problem by creating a topologic space with three dimensions: day of year, hour of day, and load value. This dimensional transformation allows the use of polynomial surface fitting to capture seasonal and diurnal patterns simultaneously, enabling accurate long-term predictions (12-24 months) while maintaining the accuracy benefits of traditional algorithms for short-term forecasts.
2Ease of operation
If system peak load hour data with substantial margin for error is used for planning, then planning simplicity is maintained, but infrastructure utilization efficiency decreases
Solution Approach 1:
The patent replaces the mechanical approach of adding substantial safety margins to peak load estimates with a mathematical modeling approach using polynomial surface fitting. This substitution enables more precise prediction of load patterns at different times and locations on the grid, allowing planners to optimize infrastructure utilization without excessive margins while maintaining planning simplicity through automated equation-based predictions.
3Adaptability or versatility
If weather-adjusted peak load data is used for planning, then adaptability to weather conditions is improved, but accuracy for specific equipment loading prediction deteriorates
Solution Approach 1:
The patent segments the overall load prediction into location-specific predictions by creating separate polynomial equations for different geographic locations and equipment types. Each location receives its own tailored equation based on local historical data, allowing weather adaptability at the regional level while maintaining precise predictions for specific equipment loading conditions.
Data Source
AI summary
Methods, Apparatus and Program Products for predicting resource usage data, weather data and econometric data, such as: demands on resources such as electrical power, water supply, communications infrastructure; temperature, humidity, wind speed, solar radiation, and degree days; and commodity price, gross domestic product, and a price index.