Monitoring System Calibrates Feature Amounts for New Electrical Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current monitoring systems face challenges in efficiently generating and managing training data for inferring the operation states of electrical devices, particularly due to the labor-intensive process of preparing feature amount information and the increased cost and burden of installing measurement instruments on each device.
Innovation Solution
A monitoring system comprising a server and monitoring devices that communicate to extract and calibrate feature amounts from measurement data, including current consumption, power consumption, and input voltage, to generate training data for new electrical devices, reducing the load on individual monitoring units and enabling unified management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If measurement instruments are mounted on each electrical device to obtain power consumption values, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The system segments the measurement function from the electrical devices themselves. Instead of mounting instruments on each device, a single measurement instrument measures total power consumption in the trunk line, and the monitoring device segments this total into individual device consumptions through inference processing.
Solution Approach 2:
The monitoring device acts as an intermediary between the single measurement instrument and the multiple electrical devices. It receives total power consumption data, extracts feature amounts, compares them with training data, and infers individual device operation states without direct measurement from each device.
2Measurement precision
If measurement instruments are installed on each electrical device, then measurement precision is improved, but the work burden and cost increase
Solution Approach 1:
The system separates the measurement function (performed by a single instrument in the trunk line) from the analysis function (performed by the monitoring device). This segmentation eliminates the need for physical installation on each device while maintaining the capability to obtain individual device consumption data.
Solution Approach 2:
The system creates a virtual model of each electrical device's power consumption pattern through training data and feature amount extraction. This digital copy allows the monitoring device to infer individual device consumption from total consumption measurements without physical contact with each device.
3Measurement precision
If feature amount information is prepared for each electrical device individually, then inference accuracy is improved, but productivity decreases
Solution Approach 1:
The system merges the training data preparation process across multiple devices. Instead of requiring separate feature amount extraction for each device, the monitoring device collects operation data from multiple devices, extracts feature amounts from total consumption patterns, and builds training data collectively, significantly improving preparation efficiency.
Solution Approach 2:
The monitoring device performs multiple functions: it measures total power consumption, extracts feature amounts, compares with training data, infers individual device states, and updates training data. This multi-functional approach eliminates the need for separate specialized processes for each device while maintaining high inference accuracy.
Data Source
AI summary
A monitoring system including a plurality of monitoring devices (20) and a server (10) is provided. When new device data including a feature amount of a new electrical device different form existing electrical devices is extracted from measurement data (current consumption or the like), the monitoring device (20) calibrates the feature amount of the new electrical device using calibration data generated based on calibration data of the exiting electrical device, and transmits the new device data to the server (10) in association with identification information of the new electrical device. When the calibrated new device data of the new electrical device is received from the plurality of monitoring devices (20), the server (10) generates and registers training data of the new electrical device based on the plurality of pieces of calibrated new device data.