Non-Intrusive Load Monitoring With Transient Event Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing non-intrusive load monitoring (NILM) systems fail to leverage high-accuracy electrical sensors and efficient machine learning for accurate differentiation between devices and processes, leading to inefficiencies and inaccurate energy consumption analysis.
Innovation Solution
A system utilizing a high-accuracy current sensor with transient event detection, scanning and scaling techniques, and machine learning to identify and classify electrical device operations, employing multiple scanning windows and pattern recognition algorithms to capture and process transient activity data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional NILM approaches are used with limited sensors, then system complexity is reduced and ease of installation is improved, but measurement precision and device differentiation accuracy deteriorate
Solution Approach 1:
The patent replaces traditional mechanical sensor-based measurement systems with an AI-driven virtual sensor system. The AI model processes data from existing utility meters to generate precise device-level measurements without requiring physical installation of multiple sensors, thereby maintaining measurement precision while reducing system complexity
Solution Approach 2:
The patent introduces an AI processing layer as an intermediary between utility meter data and device-level insights. This intermediary layer extracts and analyzes transient patterns to differentiate devices accurately without requiring direct physical measurement at each device, resolving the contradiction between accuracy and complexity
2Measurement precision
If high-accuracy current sensors with high report rates are deployed, then measurement precision and transient event detection are improved, but device complexity and installation complexity increase
Solution Approach 1:
The patent extracts only the essential transient event features from the high-rate data stream using AI analysis, rather than processing or storing the entire high-volume data stream. This extraction approach maintains transient detection accuracy while reducing the complexity of the sensor and processing system
Solution Approach 2:
The AI model is pre-trained to recognize and detect transient events associated with specific devices. This preliminary training allows the system to achieve high measurement precision for transient events without requiring complex real-time processing infrastructure, as the detection logic is already embedded in the trained model
3Measurement precision
If AI models are trained on transient patterns from multiple devices, then device classification accuracy is improved, but loss of information and data processing complexity increase
Solution Approach 1:
The patent transforms the raw transient pattern data into optimized feature representations that capture the essential characteristics needed for device classification. By changing the parameter representation from raw time-series data to extracted features, the system maintains classification accuracy while preserving critical information and reducing data volume
Data Source
AI summary
A system may include a non-intrusive sensor circuitry configured to provide electrical measurement data, including, for example, current data, voltage data, power factor data, active power consumption data, reactive power consumption data, or a combination thereof. A transient event detector may sweep the electrical measurement data with a first window and a second window, the first window adjacent to the second window. The transient event detector may identify a start and an end of transient activity based on electrical measurement data referenced by separate adjacent windows. The transient event detector may capture a transient activity data segment comprising a portion of the electrical measurement data between first index and the second index.