You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Use constrained smoothing splines to fit robust models of electric load as a function of temperature
Description
ASP learns the relationship between electric load and temperature for individual geographies (e.g. ReEDS regions, states, etc.) based on one or more years of historical data. The smoothing splines are used to learn the "aggregate" relationship between daily average load and daily average temperature, with simple averaging by month used to learn hourly offsets from the daily average. Together, these techniques yield a simple, lightweight method for fitting robust models that perform well both for description and prediction even with limited (e.g., a single year) and noisy (e.g., representing small or highly meteorologically variable geographies) data.
Mathematical Description
Spline fits are constrained to have a single critical point when modeling the relationship between daily average load and daily average temperature, such that forecasted loads are guaranteed to increase as temperatures become more extreme (both hot and cold).
Forthcoming (2023) publication from NREL study "Enhanced Simulation Tools for Scheduling Solar Plus Storage Power Plants." PI: Sourabh Dalvi.
Forthcoming (2023) publication from NREL study "Foundational Assistance to ISO/RTOs under Electricity Market Transformation." PI: Bethany Frew.
Use Cases
Electric load forecasting as a function of temperature for both description and forecasting. Particularly relevant in limited-data-availability contexts.
Infrastructure Sector
Atmospheric dispersion
Agriculture
Biomass
Buildings
Communications
Cooling
Ecosystems
Electric
District heating
Forestry
Health
Hydrogen
Individual heating
Land use
Liquid fuels
Natural Gas
Transportation
Water
Represented Behavior
Earth Systems
Employment
Built Infrastructure
Financial
Macro-economy
Micro-economy
Policy
Social
Modeling Paradigm
Analytics
Data
Discrete Simulation
Dynamic Simulation
Equilibrium
Engineering/Design
Optimization
Visualization
Capabilities
No response
Programming Language
C – ISO/IEC 9899
C++ (C plus plus) – ISO/IEC 14882
C# (C sharp) – ISO/IEC 23270
Delphi
GAMS (General Algebraic Modeling System)
Go
Haskell
Java
JavaScript(Scripting language)
Julia
Kotlin
LabVIEW
Lua
MATLAB
Modelica
Nim
Object Pascal
Octave
Pascal Script
Python
R
Rust
Simulink
Swift (Apple programming language)
WebAssembly
Zig
Required Dependencies
N/A
What is the software tool's license?
3-clause BSD License (BSD-3-Clause)
Operating System Support
Windows
Mac OSX
Linux
iOS
Android
User Interface
Programmatic
Command line
Web based
Graphical user
Menu driven
Form based
Natural language
Parallel Computing Paradigm
Multi-threaded computing
Multi-core computing
Distributed computing
Cluster computing
Massively parallel computing
Grid computing
Reconfigurable computing with field-programmable gate arrays (FPGA)
General-purpose computing on graphics processing units
Application-specific integrated circuits
Vector processors
What is the highest temporal resolution supported by the tool?
Hours
What is the typical temporal resolution supported by the tool?
Hours
What is the largest temporal scope supported by the tool?
Years
What is the typical temporal scope supported by the tool?
Years
What is the highest spatial resolution supported by the tool?
Facility
What is the typical spatial resolution supported by the tool?
State
What is the largest spatial scope supported by the tool?
Global
What is the typical spatial scope supported by the tool?
Name
Adaptive Splines for Prediction
Screenshots
Focus Topic
Electric load forecasting tool
Primary Purpose
Use constrained smoothing splines to fit robust models of electric load as a function of temperature
Description
ASP learns the relationship between electric load and temperature for individual geographies (e.g. ReEDS regions, states, etc.) based on one or more years of historical data. The smoothing splines are used to learn the "aggregate" relationship between daily average load and daily average temperature, with simple averaging by month used to learn hourly offsets from the daily average. Together, these techniques yield a simple, lightweight method for fitting robust models that perform well both for description and prediction even with limited (e.g., a single year) and noisy (e.g., representing small or highly meteorologically variable geographies) data.
Mathematical Description
Spline fits are constrained to have a single critical point when modeling the relationship between daily average load and daily average temperature, such that forecasted loads are guaranteed to increase as temperatures become more extreme (both hot and cold).
Website
https://github.com/NREL/adaptive-splines
Documentation
https://github.com/NREL/adaptive-splines
Source
https://github.com/NREL/adaptive-splines
Year
2022
Institution
National Renewable Energy Laboratory
Funding Source
DOE
Publications
2
Publication List
Forthcoming (2023) publication from NREL study "Enhanced Simulation Tools for Scheduling Solar Plus Storage Power Plants." PI: Sourabh Dalvi.
Forthcoming (2023) publication from NREL study "Foundational Assistance to ISO/RTOs under Electricity Market Transformation." PI: Bethany Frew.
Use Cases
Electric load forecasting as a function of temperature for both description and forecasting. Particularly relevant in limited-data-availability contexts.
Infrastructure Sector
Represented Behavior
Modeling Paradigm
Capabilities
No response
Programming Language
Required Dependencies
N/A
What is the software tool's license?
3-clause BSD License (BSD-3-Clause)
Operating System Support
User Interface
Parallel Computing Paradigm
What is the highest temporal resolution supported by the tool?
Hours
What is the typical temporal resolution supported by the tool?
Hours
What is the largest temporal scope supported by the tool?
Years
What is the typical temporal scope supported by the tool?
Years
What is the highest spatial resolution supported by the tool?
Facility
What is the typical spatial resolution supported by the tool?
State
What is the largest spatial scope supported by the tool?
Global
What is the typical spatial scope supported by the tool?
Region
Input Data Format
CSV
Input Data Description
Hourly temperature and load data
Output Data Format
CSV
Output Data Description
CSV files describing the fitted models
Contact Details
sinnott.murphy@nrel.gov
Interface, Integration, and Linkage
No response
The text was updated successfully, but these errors were encountered: