Visual and Acoustic Event Detection for Industrial Process Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional analytics for industrial processes are limited by reliance on expert-defined patterns, lack of real-time analysis, and high costs associated with intrusive sensor installations, leading to unnoticed abnormalities and delayed remediation.
Innovation Solution
Implementing visual and acoustic analytics using commodity cameras and microphones to acquire data, which is analyzed in real-time by a machine learning engine that adapts to detect events and generate control signals, allowing for automatic identification and control of industrial processes without manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional instrument-based data acquisition is used, then measurement reliability is improved, but device complexity and installation cost increase due to intrusive sensor installations and wiring
Solution Approach 1:
The patent replaces traditional mechanical/electrical sensors with cameras and microphones that capture visual and acoustic data. This substitution eliminates the need for intrusive physical sensors and wiring while maintaining monitoring capabilities through non-contact data acquisition methods.
Solution Approach 2:
The patent uses cameras to create visual copies (images and videos) of the industrial process environment, and microphones to capture acoustic copies of operational sounds. These copies serve as surrogate data sources that replace direct physical measurement by traditional sensors, enabling monitoring without physical intrusion.
2Measurement precision
If expert-defined predetermined patterns are used for event detection, then detection accuracy for known events is improved, but adaptability to unknown abnormalities deteriorates
Solution Approach 1:
The patent implements a machine learning engine that automatically learns and adapts to the specific industrial environment without requiring manual programming of detection patterns by experts. The system serves itself by continuously training on environment-specific data, enabling automatic adaptation to both known and unknown abnormalities.
Solution Approach 2:
The patent transforms the static, fixed pattern-matching approach into a dynamic system where the machine learning model continuously adapts and evolves based on new data. The detection patterns are not predetermined but dynamically generated and updated through ongoing machine learning training processes.
3Measurement precision
If manual analysis of recorded data is performed, then detection thoroughness is improved, but productivity and real-time response deteriorate
Solution Approach 1:
The patent replaces manual human analysis with an automated machine learning-based analysis system. This substitution enables continuous real-time processing of visual and acoustic data streams, eliminating the bottleneck of manual review while maintaining or improving detection accuracy through algorithmic pattern recognition.
Solution Approach 2:
The patent implements continuous real-time analysis of incoming data streams from cameras and microphones. The machine learning engine processes data continuously without interruption, enabling immediate detection and response to events as they occur, rather than batch processing or periodic manual review.
4Device complexity
If data is not analyzed in real-time, then system complexity is reduced, but loss of time for remediation increases
Solution Approach 1:
The patent performs preliminary actions by continuously pre-processing and analyzing data in real-time before critical events occur. The machine learning engine is constantly trained and updated with incoming data, so when abnormalities occur, the system is already prepared to detect and respond immediately, reducing remediation time.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and methods are provided for detecting events in industrial processes. An acquisition system may include one of a camera and an audio recorder to acquire monitoring data in the form of one of imaging data and acoustic data, respectively. A computer system, may include a machine learning engine and may be programmed to classify the monitoring data under a classifier, quantify, based on the classifier, the monitoring data with at least one quantifier, and detect an event when the at least one quantifier satisfies a predetermined rule corresponding to the at least one quantifier.