Energy Consumption Data Segmentation for Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems face challenges in quickly understanding and analyzing energy consumption patterns across sites like residential buildings, commercial premises, and factories, as they generate vast amounts of data that require efficient processing and modeling to identify anomalies and optimize performance.
Innovation Solution
A data processing method involving a computer system that selects and segments time series data, projects it into a multi-dimensional space based on similarity, and uses digital models to group and analyze data sections, facilitating the identification of aberrant data and performance monitoring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If measurement devices collect comprehensive energy consumption data and physical parameters at high sampling rates, then measurement precision and data completeness are improved, but the quantity of data increases significantly making analysis difficult
Solution Approach 1:
The patent segments the large volume of measurement data into smaller time-series segments or data chunks, which are then processed and analyzed individually. This segmentation allows the system to handle comprehensive measurement data without being overwhelmed by the total quantity, maintaining measurement precision while making the data manageable for analysis.
Solution Approach 2:
The patent extracts key features and characteristics from the comprehensive measurement data, such as typical consumption patterns, anomalies, and relevant parameters. By extracting only the essential information from the large dataset, the system maintains measurement precision while reducing the quantity of data that needs to be stored and analyzed.
2Loss of information
If comprehensive measurement data is collected from multiple sensors and meters, then data completeness is improved, but the complexity of processing and analyzing the data increases
Solution Approach 1:
The patent merges data from multiple sensors and measurement devices into a unified processing framework. By combining the data streams and applying integrated processing algorithms, the system maintains complete information from all sources while reducing the overall complexity of handling multiple separate data streams through standardized processing procedures.
Solution Approach 2:
The patent implements a universal data processing platform that can handle various types of measurement data (energy consumption, physical parameters, operational data) through a single system architecture. This multi-functional approach maintains data completeness from diverse sources while avoiding the complexity of separate processing systems for each data type.
3Measurement precision
If detailed analysis of energy consumption patterns is performed, then detection precision of anomalies is improved, but the time required for analysis increases
Solution Approach 1:
The patent performs preliminary processing of measurement data, including segmentation, feature extraction, and pre-analysis, before detailed anomaly detection is conducted. By preparing the data in advance and identifying obvious patterns early, the system achieves high anomaly detection precision while reducing the time required for the final detailed analysis phase.
Solution Approach 2:
The patent replaces manual or mechanical analysis methods with automated computational algorithms for detecting anomalies in energy consumption patterns. This substitution enables detailed analysis of comprehensive data with high precision while significantly reducing the time required, as automated systems can process data much faster than human analysts.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Improved data processing method for analyzing the energy consumption of a site from measurement data including: - selection of time series of data, - segmentation of the time series into chunks, - projection of the chunks, - display of the projections of the data chunks, - selection of at least a first group of chunks, - establishment of a numerical classification model characterizing the first group of data chunks selected from calendar metadata - establishment of a numerical classification model characterizing the first group of data chunks selected from the profiles of the chunks.