Abstract:Long-term planning of a robust power system requires the understanding of changing demand patterns. Electricity demand is highly weather sensitive. Thus, the supply side variation from introducing intermittent renewable sources, juxtaposed with variable demand, will introduce additional challenges in the grid planning process. By understanding the spatial and temporal variability of temperature over the US, the response of demand to natural variability and climate change-related effects on temperature can be separated, especially because the effects due to the former factor are not known. Through this project, we aim to better support the technology & policy development process for power systems by developing machine and deep learning 'back-forecasting' models to reconstruct multidecadal demand records and study the natural variability of temperature and its influence on demand.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to reconstruct long - term historical electricity demand data to better support the planning of power systems and the development of technologies and policies. Specifically: 1. **Understanding the changes in demand patterns**: In order to conduct robust long - term planning of power systems, it is necessary to understand the constantly changing demand patterns. Electricity demand is highly dependent on weather conditions, and the introduction of intermittent renewable energy sources (such as wind and solar energy) will increase the challenges in the grid planning process. 2. **Distinguishing the impact of natural variability from that of climate change**: By studying the spatio - temporal variability of temperature in the continental United States, the response of demand to natural variability (such as seasonal and inter - annual changes) can be distinguished from the impact of climate change on temperature, especially since the impact of the former has largely not been fully understood. 3. **Filling in the gaps in historical data**: Due to the lack of multi - decade historical demand records, especially in the Balancing Authority (BA) database, it is difficult to assess the capacity gap after the integration of clean energy and to understand the changes in electricity demand. Therefore, it is necessary to reconstruct the hourly historical demand data from 1980 to 2014 to cover a time span of multiple decades. ### Specific objectives - **Develop machine learning and deep learning models**: Use the "back - casting" method to reconstruct multi - decade demand records based on key predictor variables such as temperature. - **Analyze the natural variability of temperature and its impact on demand**: By studying the natural inter - annual variability (IAV) of temperature, understand its specific impact on electricity demand. - **Improve the accuracy of power system planning**: By providing more complete historical demand data, help planners better understand and respond to future changes in electricity demand, especially in the context of considering climate change. ### Method overview - **Model selection**: The effects of piecewise linear regression and support vector regression (SVR) were compared, and finally the long - short - term memory network (LSTM) was selected as the main model because it can effectively handle complex non - linear relationships in time - series data. - **Data processing**: Use temperature data from the MERRA re - analysis data set and the hourly demand records of ERCOT for training and validation. - **Performance evaluation**: Evaluate model performance through the root - mean - square error (RMSE) and the adjusted \(R^{2}\) value. In conclusion, this study aims to reconstruct long - term historical electricity demand data through advanced machine - learning techniques, thereby providing more reliable data support for future power - system planning.

Reconstruction of Long-Term Historical Demand Data

Electricity Demand Forecasting Using HWT Model with Fourfold Seasonality

Predicting Energy Demand Using Machine Learning: Exploring Temporal and Weather-Related Patterns, Variations, and Impacts

Medium- and long-term demand estimation integrating emerging technologies

Forecasting for demand response in smart grids: An analysis on use of anthropologic and structural data and short term multiple loads forecasting

Household Electricity Demand Forecasting -- Benchmarking State-of-the-Art Methods

mshw, a forecasting library to predict short-term electricity demand based on multiple seasonal Holt-Winters

Developing Reliable Hourly Electricity Demand Data Through Screening and Imputation.

Representative Days and Hours with Piecewise Linear Transitions for Power System Planning

Electricity Demand Forecasting with Hybrid Statistical and Machine Learning Algorithms: Case Study of Ukraine

Hierarchical demand forecasting benchmark for the distribution grid

U.S. Test System with High Spatial and Temporal Resolution for Renewable Integration Studies

Generating Synthetic Electricity Load Time Series at District Scale Using Probabilistic Forecasts

Cross-Temporal Hierarchical Forecast Reconciliation of Natural Gas Demand

Local Cold Load Pick-up Estimation Using Customer Energy Consumption Measurements

From High-Resolution Data to High-Resolution Probabilistic Load Forecasts

Future electricity demand for Europe: Unraveling the dynamics of the Temperature Response Function

Weather Sensitive High Spatio-Temporal Resolution Transportation Electric Load Profiles For Multiple Decarbonization Pathways

Long and Short-Term Power Supply and Demand Forecasting Based on Time Series Analysis under High Proportion Clean Energy Integration

Data-Driven Load Profiles and the Dynamics of Residential Electric PowerConsumption