CNN Anomaly Detection for Transactive Energy Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection methods for transactive energy systems are inadequate in addressing cyber and physical layer attacks, as they often require domain expert intervention, are costly, and fail to account for interrelations between subsystems, leading to inefficiencies and limitations in detecting anomalies across different configurations.
Innovation Solution
A deep convolutional neural network (CNN) is employed to analyze time series data from multiple subsystems, including electricity demand and supply, pricing, and weather data, to detect anomalies and generate alerts, with the ability to retrain based on feedback, reducing the need for manual feature crafting and enhancing adaptability across various system configurations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If domain expert intervention is used to construct physical models for anomaly detection, then measurement precision is improved, but device complexity and loss of time increase
Solution Approach 1:
The system employs automated machine learning algorithms that self-adjust parameters and construct detection models without requiring domain expert intervention. The algorithm automatically learns from historical data and adjusts its parameters to optimize anomaly detection performance.
Solution Approach 2:
The patent replaces manual domain expert modeling with automated computational algorithms. Instead of experts constructing physical models, the system uses machine learning algorithms to automatically learn patterns and relationships from data, substituting human intellectual labor with automated computational processes.
2Measurement precision
If domain expert intervention is used to construct physical models for anomaly detection, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The system performs preliminary actions by pre-processing historical data and pre-training detection algorithms during system setup. This allows the model to be ready for deployment without requiring time-consuming expert configuration at the time of use.
Solution Approach 2:
The automated algorithm performs self-configuration and self-optimization without requiring expert time investment. The system automatically learns from data and adjusts parameters, eliminating the need for experts to spend time constructing and tuning physical models.
3Device complexity
If monitoring is performed on a single subsystem, then device complexity is reduced, but reliability decreases due to inability to detect inter-subsystem inconsistencies
Solution Approach 1:
The patent implements a universal monitoring framework that can detect anomalies across multiple subsystems using a single integrated system. The algorithm is designed to handle diverse data types and subsystem configurations uniformly, providing multi-functional anomaly detection capabilities.
Solution Approach 2:
The system merges monitoring of multiple subsystems into a unified detection framework. By combining data from different subsystems and analyzing them together, the system can detect inter-subsystem inconsistencies that would be missed by isolated monitoring approaches.
4Manufacturing precision
If features are constructed for one specific subsystem configuration, then manufacturing precision is improved for that configuration, but adaptability decreases for other configurations
Solution Approach 1:
The system employs dynamic feature construction that adapts to different subsystem configurations. Rather than using fixed features designed for one configuration, the algorithm dynamically adjusts feature extraction and selection based on the specific data characteristics and configuration being analyzed.
Solution Approach 2:
The patent implements parameter changes to achieve adaptability. The algorithm modifies its parameters, feature extraction methods, and detection thresholds based on the specific subsystem configuration and data characteristics, allowing it to maintain high precision across diverse configurations without requiring manual reconfiguration.
Data Source
AI summary
A computer-implemented method for power grid anomaly detection using a convolutional neural network (CNN) trained to detect anomalies in electricity demand data and electricity supply data includes receiving (i) electricity demand data comprising time series measurements of consumption of electricity by a plurality of consumers, and (ii) electricity supply data comprising time series measurements of availability of electricity by one or more producers. An input matrix is generated that comprises the electricity demand data and the electricity supply data. The CNN is applied to the input matrix to yield a probability of anomaly in the electricity demand data and the electricity supply data. If the probability of anomaly is above a threshold value, an alert message is generated for one or more system operators.