Heat Meter Fault Prioritization Using ML Failure Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Calorimeters installed in households for central heating systems are difficult to maintain, prone to physical failure, and often go unnoticed, leading to potential fairness issues in heat usage billing and increased troubleshooting time due to inaccurate failure identification.
Innovation Solution
A calorimeter abnormality determination device and method using machine learning models to analyze measurement data from multiple calorimeters, determining failure probability and cause, and prioritizing repairs based on both factors to efficiently manage and maintain a large number of calorimeters with a small workforce.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If meter readers periodically check calorimeters in multiple households, then failure detection capability is improved, but labor cost and time consumption increase significantly
Solution Approach 1:
The calorimeter system performs self-diagnosis by automatically monitoring its own operational parameters (flow rate, temperature differential, measured heat) and comparing them against expected ranges. The system generates its own failure alerts without requiring external inspection, enabling households to autonomously detect and report calorimeter issues.
Solution Approach 2:
The manual mechanical inspection process by meter readers is replaced with an automated electronic monitoring system that continuously collects sensor data, processes it through analysis algorithms, and generates failure notifications automatically, eliminating the need for periodic human visits.
2Measurement precision
If meter readers troubleshoot calorimeters without accurate failure cause identification, then repair time and number of revisits increase
Solution Approach 1:
The system performs preliminary failure cause analysis by continuously monitoring operational parameters and comparing them against known failure patterns before a meter reader arrives. This preliminary diagnosis provides the technician with specific guidance on the likely failure cause, enabling them to proceed directly to the repair task rather than performing extensive troubleshooting.
Solution Approach 2:
The system provides continuous feedback on calorimeter performance by monitoring operational parameters and comparing them against expected ranges. When deviations are detected, the system analyzes the pattern of deviations to identify the specific failure cause and communicates this information to the maintenance personnel.
3Reliability
If calorimeters are not regularly maintained, then reliability deteriorates over time, but maintenance cost and effort increase when failures occur
Solution Approach 1:
The system performs preliminary maintenance by continuously monitoring operational parameters and detecting early signs of degradation or failure. By identifying potential issues before they result in complete failure, the system enables proactive maintenance scheduling that prevents reliability deterioration while optimizing maintenance resource allocation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The disclosure provides a device and method for determining an abnormality in a calorimeter, comprising a collecting device collecting measurement data from a plurality of calorimeters each measuring a calorie consumed in a specific household and a failure determination device determining a failure probability and a cause of failure for the plurality of calorimeters based on the measurement data. In this case, the failure determination device may include a receiving unit receiving the measurement data from the collecting device, a failure probability determination unit inputting the measurement data to a first machine learning model to determine failure probability information about the plurality of calorimeters, a failure cause determination unit inputting the measurement data to a second machine learning model to determine failure cause information about the plurality of calorimeters, and a priority determination unit determining a priority for a target calorimeter, which is one or more calorimeters among the plurality of calorimeters, based on the failure probability information and the failure cause information.