ML Compressed Air Leak Detection via Shutdown Event Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Detecting leaks in compressed air systems is challenging due to the complexity of these systems, the difficulty in distinguishing between consumed and leaked air, and the invisibility of air, making conventional methods inaccurate and inefficient.
Innovation Solution
A computer-implemented method using Machine Learning (ML) models trained to identify compressed air leak patterns based on pressure and flow rate data during shutdown events, allowing for the detection of leaks without the need for additional equipment or manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional leak detection methods are used, then the system can operate continuously, but the detection accuracy is low and leaks cannot be reliably distinguished from normal consumption
Solution Approach 1:
The system performs preliminary analysis by identifying shutdown events and collecting pressure and flow rate data during these events before applying ML models. This preliminary data collection and event identification prepares the system to detect leaks more accurately when the system is actually running, rather than requiring full shutdowns for detection.
Solution Approach 2:
The patent replaces conventional mechanical/physical leak detection methods (audible means, visual inspection) with Machine Learning models that analyze pressure and flow rate data. This substitution enables automated, accurate leak detection without requiring manual intervention or system shutdowns.
2Measurement precision
If full system shutdowns are performed to detect leaks, then detection accuracy improves, but production disruption and time loss increase
Solution Approach 1:
Instead of requiring complete system shutdowns, the system uses partial shutdown events (where demand is at least partially reduced) combined with ML analysis of pressure and flow rate data. This partial approach maintains production continuity while still enabling accurate leak detection through the trained models.
Solution Approach 2:
The patent replaces the mechanical approach of full system shutdowns with an intelligent ML-based analysis system that can detect leaks during normal operations by analyzing pressure and flow rate patterns, eliminating the need for production stoppages.
3Difficulty of detecting and measuring
If additional detection equipment is deployed to improve leak detection, then detection capability improves, but device complexity and cost increase
Solution Approach 1:
The system makes existing pressure and flow rate sensors serve multiple functions: both for normal system operation monitoring and for leak detection. The ML models analyze the same operational data for dual purposes, eliminating the need for separate dedicated leak detection equipment.
Solution Approach 2:
The system uses its own operational data (pressure and flow rate measurements already taken for system control) to perform self-diagnosis for leak detection. The ML models analyze data that the system already collects during normal operation, enabling the system to detect its own leaks without external detection equipment.
Data Source
AI summary
Disclosed here is a computer implemented method of detecting leaks in a compressed air system using Machine Learning (ML), comprising identifying one or more shutdown events in a compressed air system comprising one or more compressed air clients consuming compressed air delivered by one or more air compressors, during each shutdown event demand of compressed air by the compressed air client(s) is at least partially reduced, receiving pressure data and flow rate data measured in the compressed air system during each shutdown event, and detecting one or more compressed air leaks in the compressed air system using one or more trained ML models applied to the pressure data and the flow rate data. The ML model(s) are trained to identify a plurality of compressed air leak patterns based on pressure and flow rate data.