An electricity smart meter dataset of Spanish households: insights into consumption patterns

Abstract

Smart meters are devices that provide detailed information about the energy consumed by specific electricity supply points, such as homes, offices, and businesses. Data from smart meters are useful for modeling energy systems, predicting electricity consumption, and understanding human behavior. We present the first smart meter dataset from Spanish electricity supply points, expanding the geographic diversity of available data on energy consumption at the household level and reducing biases in existing data, which typically come from a limited number of countries. The dataset consists of 25,559 raw hourly time series with an average length of nearly three years, spanning from November 2014 to June 2022. It also includes three subsets obtained by segmenting and cleaning the raw time series data, each focusing on the periods before, during, and after the COVID-19 lockdowns in Spain. This dataset is a valuable resource for studying electricity consumption patterns and behaviors that emerge in response to different natural experiments, such as nationwide and regional lockdowns, nighttime curfews, and changes in electricity pricing.

Leveraging Transformer-Based Non-Parametric Probabilistic Prediction Model for Distributed Energy Storage System Dispatch

“…To simulate dynamic loads, we aggregated hourly electricity consumption data from encrypted smart meters of users provided by the Spanish company GoiEner for each bus. This dataset [35], which was publicly released in January 2024, includes load data for various types of users, i.e., industrial, commercial, and residential. We extract the load data from 1 June 2021, 00:00 to 31 May 2022, 23:00, ensuring that the average load level matches the static load in the original network model for each bus, and the ratio of reactive power to active power is maintained.…”