Divide-Conquer Transformer Learning for Predicting Electric Vehicle Charging Events Using Smart Meter Data

Fucai Ke,Hao Wang

2024-03-20

Abstract:Predicting electric vehicle (EV) charging events is crucial for load scheduling and energy management, promoting seamless transportation electrification and decarbonization. While prior studies have focused on EV charging demand prediction, primarily for public charging stations using historical charging data, home charging prediction is equally essential. However, existing prediction methods may not be suitable due to the unavailability of or limited access to home charging data. To address this research gap, inspired by the concept of non-intrusive load monitoring (NILM), we develop a home charging prediction method using historical smart meter data. Different from NILM detecting EV charging that has already occurred, our method provides predictive information of future EV charging occurrences, thus enhancing its utility for charging management. Specifically, our method, leverages a self-attention mechanism-based transformer model, employing a ``divide-conquer'' strategy, to process historical meter data to effectively and learn EV charging representation for charging occurrence prediction. Our method enables prediction at one-minute interval hour-ahead. Experimental results demonstrate the effectiveness of our method, achieving consistently high accuracy of over 96.81\% across different prediction time spans. Notably, our method achieves high prediction performance solely using smart meter data, making it a practical and suitable solution for grid operators.

Computers and Society

What problem does this paper attempt to address?

The paper primarily addresses the issue of predicting home charging events for electric vehicles (EVs) and proposes a new solution. Existing research mostly focuses on predicting charging demand at public charging stations, while home charging prediction has received relatively less attention. This is especially true when home charging data is unavailable or access is limited, making existing methods potentially unsuitable. Therefore, the paper proposes a new approach to fill this research gap. Specifically, the study develops a home charging prediction method based on historical data from smart meters. This method differs from traditional non-intrusive load monitoring (NILM), which is typically used to detect EV charging events that have already occurred. The new method can predict future EV charging events, making it more practical for charging management. The core of this method is a Transformer model based on a self-attention mechanism, and it employs a "divide and conquer" strategy to process historical meter data, effectively learning EV charging representations and predicting charging events. The method can achieve one-hour-ahead predictions with a one-minute interval and demonstrated a high accuracy of over 96.81% in experiments. This indicates that the method can efficiently predict charging events using only smart meter data, making it a practical and feasible solution for grid operators. Additionally, the paper discusses the trade-off between input history length, prediction time span, and model performance, as well as the impact of probability threshold selection on model performance.

Divide-Conquer Transformer Learning for Predicting Electric Vehicle Charging Events Using Smart Meter Data

Prediction of Electric Vehicles Charging Load Using Long Short-Term Memory Model

Prediction of Electric Vehicles Charging Demand: A Transformer-Based Deep Learning Approach

Transformer-Based User Charging Duration Prediction Using Privacy Protection and Data Aggregation

Orderly Charging Control of Electric Vehicles: A Smart Meter-Based Approach

Revenue Estimation for Orderly Charging of Electric Vehicles Based on Smart Meters

Data-driven spatial-temporal prediction of electric vehicle load profile considering charging behavior

Online Electric Vehicle Charging Detection Based on Memory-based Transformer using Smart Meter Data

Probabilistic Forecasting of Electric Vehicle Charging Load Using Composite Quantile Regression LSTM

Coordinated Charging of Electric Vehicle Group Using Smart Meters

Deep Reinforcement Learning Based Charging Scheduling for Household Electric Vehicles in Active Distribution Network

Forecasting method of electric vehicle charging load based on virtual prediction parameter estimation strategy

Attention-based Citywide Electric Vehicle Charging Demand Prediction Approach Considering Urban Region and Dynamic Influences

Prediction of EV Charging Load Using Two-Stage Time Series Decomposition and DeepBiLSTM Model

Multi-step electric vehicles charging loads forecasting: An autoformer variant with feature extraction, frequency enhancement, and error correction blocks

Non-Intrusive Load Monitoring for Feeder-Level EV Charging Detection: Sliding Window-based Approaches to Offline and Online Detection

FMGCN: Federated Meta Learning-Augmented Graph Convolutional Network for EV Charging Demand Forecasting

Ensemble machine learning-based algorithm for electric vehicle user behavior prediction

Short-Term Forecasting of EV Charging Load Using Prophet-BiLSTM

Unveiling electric vehicle (EV) charging patterns and their transformative role in electricity balancing and delivery: Insights from real-world data in Sweden

Hybrid Predictive Modeling for Charging Demand Prediction of Electric Vehicles