EV Charging Detection from Interval Data for Grid Load Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Electrical utility grids face challenges in detecting and managing the increased load from electric vehicle (EV) charging, as existing technologies are inadequate in identifying EV charging events in real-time across residential installations, leading to potential grid overload.
Innovation Solution
The implementation of a system utilizing a convolutional neural network model, a midnight outlier usage model, and a monolithic shape model to detect EV charging activities by analyzing time-series electricity consumption data, temperature normalization, and identifying distinctive consumption patterns indicative of EV charging, such as significant changes, off-peak usage, and sustained energy consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If traditional electricity consumption monitoring is used, then infrastructure complexity is low, but EV charging detection capability is insufficient
Solution Approach 1:
The patent replaces traditional mechanical/electrical monitoring systems with machine learning models (convolutional neural networks, random forest classifiers) that analyze electricity consumption data patterns. These software-based systems detect EV charging events by recognizing distinctive consumption signatures without requiring additional hardware sensors or complex electrical infrastructure modifications at customer premises.
Solution Approach 2:
The patent introduces an intermediary data analysis layer between existing smart meters and utility systems. Machine learning models process raw electricity consumption data from standard meters, extracting EV charging information without requiring direct communication with EV chargers or modification of metering infrastructure. This intermediary layer enables detection using existing measurement capabilities.
2Productivity
If real-time EV charging detection is implemented, then grid load management capability is improved, but data processing requirements increase
Solution Approach 1:
The patent extracts only the essential features from raw electricity consumption data that are indicative of EV charging events. Machine learning models identify and extract distinctive patterns such as sustained high-power consumption, specific temporal profiles, and load signature characteristics, discarding redundant data. This extraction approach reduces processing requirements while maintaining detection accuracy.
Solution Approach 2:
The patent segments the electricity consumption data analysis into distinct processing stages: data collection from smart meters, feature extraction (identifying consumption patterns), model inference (classifying EV charging events), and results aggregation. This segmentation allows parallel processing and optimization at each stage, improving overall data processing efficiency for real-time detection.
3Measurement precision
If multiple detection models are deployed, then detection accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent employs multiple machine learning models with different detection strategies (convolutional neural networks for pattern recognition, random forest classifiers for feature-based detection, outlier detection algorithms) that can be dynamically selected or combined based on data characteristics and detection needs. This dynamic multi-model approach improves accuracy by leveraging complementary strengths of different algorithms while allowing computational resources to be allocated efficiently.
Data Source
AI summary
The disclosure describes three complimentary, synergistically interacting, and yet individually capable techniques for detecting electrical vehicle charging activities. In one example, convolutional neural network techniques find “edges” or points of significant change in electricity consumption. Time-series of electricity consumption are examined, and temperature is considered to normalize for changes in heating, ventilation, and air conditioning consumption. In an example, a time-series of electrical-consumption data of a service site is obtained over a time-range. The time-series of electrical-consumption data is converted into a time-series of consumption-change data. Temperature data may be associated with terms of the time-series of consumption-change data to thereby create input data for a machine-learned algorithm over the time-range. The input data is provided to a machine-learned model. The input data is processed over the time-range in the machine-learned model to generate output, such as a likelihood value of at least one EV charging event during the time-range.