Validated Waveform Extraction for Manufacturing Abnormality Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing abnormality detection systems erroneously determine detection object waveforms when mixed waveforms are present, leading to inaccurate abnormality detection in manufacturing processes.
Innovation Solution
An abnormality detection device that includes a signal input unit, input buffer, waveform extraction units, and determination units using machine learning algorithms to accurately identify and extract detection object waveforms, reducing calculation load and power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If trigger condition-based waveform extraction is used, then waveform extraction simplicity is improved, but detection accuracy deteriorates due to erroneous determination of detection object waveforms
Solution Approach 1:
The patent introduces a determination unit as an intermediary component between the waveform extraction unit and abnormality detection unit. This determination unit verifies whether extracted waveforms are genuine detection object waveforms before they proceed to abnormality detection, preventing erroneous determinations while maintaining the simplicity of trigger condition-based extraction methodology
Solution Approach 2:
The patent segments the waveform processing system into distinct functional units: a waveform extraction unit that uses simple trigger conditions, a determination unit that validates extracted waveforms, and an abnormality detection unit that analyzes validated waveforms. This segmentation allows each unit to perform its specific function optimally without compromising overall accuracy
2Reliability
If constant waveform monitoring is performed, then abnormality detection coverage is improved, but energy consumption increases
Solution Approach 1:
The patent implements periodic action by activating the abnormality detection function only during specific periods when detection object waveforms are actually occurring. The system uses trigger conditions to identify these periods and performs waveform extraction and abnormality detection only during these intervals, rather than continuously monitoring all waveform data
Solution Approach 2:
The patent extracts only the relevant detection object waveforms from the continuous waveform data stream using trigger conditions. By extracting only the portions of waveforms that contain actual detection objects rather than processing the entire continuous stream, the system maintains comprehensive abnormality detection coverage while significantly reducing energy consumption
Data Source
AI summary
The abnormality detection device includes a first extraction unit extracting an input waveform data from a waveform data input to the first extraction unit, a first determination unit for determining whether the input waveform data includes a detection object waveform data, a second extraction unit for extracting and outputting the detection object waveform data from the input waveform data when the input waveform data is determined to include the detection object waveform data by the first determination unit; and a second determination unit for determining whether the detection target device has an abnormality based on whether the detection object waveform output from the second extraction unit indicates an abnormality.