Process Quality Monitoring Using Segmented Multi-Image Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current in-situ/inline monitoring techniques for manufacturing processes, such as laser welding, often rely on single-image analysis for quality control, which is inadequate for dynamic processes with temporal and spatial variations, and may be affected by phenomena like gas emissions or oscillations, leading to unreliable judgments of process state.
Innovation Solution
The system employs temporal discretization by segmenting process data into specific portions, using multiple sensors with band-pass filters to capture relevant wavelengths, and processing algorithms to analyze batches of images for statistical analysis and anomaly detection, enabling robust monitoring and control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If single-image analysis is used for quality control, then the system complexity is low, but the measurement precision and reliability of process state judgment deteriorate due to temporal and spatial variations and phenomena like gas emissions or oscillations
Solution Approach 1:
The process is segmented into multiple discrete time segments, with each segment containing multiple images. Instead of analyzing single images or continuous streams, the system divides the temporal domain into manageable segments, allowing statistical analysis within each segment while maintaining overall process monitoring. This segmentation enables robust quality control by analyzing patterns across multiple images rather than relying on single-image snapshots.
2Loss of time
If single-image analysis is used for quality control, then the processing time is short, but the reliability of quality judgment deteriorates due to inability to capture temporal distributions of anomalies
Solution Approach 1:
The system performs preliminary segmentation of the process into discrete time segments before detailed analysis. By pre-organizing images into segments based on temporal characteristics, the system prepares data structures that facilitate efficient statistical analysis. This preliminary action enables faster processing compared to analyzing entire continuous streams, while still capturing temporal distributions of anomalies through segment-based statistics.
3Measurement precision
If batches of images are processed for statistical analysis, then the measurement precision and reliability improve, but the device complexity and processing time increase
Solution Approach 1:
The system segments the image batch into multiple process segments, where each segment contains a subset of images analyzed for specific statistical characteristics. This segmentation reduces the computational complexity of analyzing entire batches at once, while still capturing temporal and spatial distributions of anomalies. Each segment can be processed independently, enabling parallel computation and reducing overall processing burden.
Solution Approach 2:
The system performs statistical analysis at the segment level rather than requiring complete batch processing for every quality judgment. By using partial action (analyzing segments rather than full batches continuously), the system achieves improved measurement precision for anomaly detection while reducing processing time and complexity compared to exhaustive batch analysis of all images.
4Reliability
If segment-based statistical analysis is implemented, then the robustness of quality control improves, but the loss of time for data processing increases
Solution Approach 1:
The system divides the continuous process into discrete time segments, with each segment containing a manageable number of images for statistical analysis. This segmentation enables parallel processing of multiple segments, reducing overall data processing time while maintaining robust quality control through segment-based statistics. The segmented approach allows the system to process segments independently rather than waiting for complete batch accumulation.
Solution Approach 2:
The system implements periodic analysis at segment boundaries rather than continuous processing of every image. By performing statistical analysis periodically at defined segment intervals, the system achieves robust quality control through consistent segment-based evaluation while reducing processing time compared to continuous frame-by-frame analysis. This periodic action maintains reliability while optimizing processing efficiency.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach provides a more accurate and reliable evaluation of process quality by analyzing temporal and spatial distributions of anomalies, reducing the risk of false judgments and improving the robustness of quality control in dynamic manufacturing environments.
Implementation Method 1
using multiple sensors with band-pass filters to capture relevant wavelengths
Implementation Method 2
light emissions from processes can be captured to provide data regarding a quality state of a process by using photo-sensitive devices that are sensitive to different wavelength ranges
Data Source
Figure 1
Figure 2~3
Figure 4A
AI summary
A system for evaluating at least one state of a process is provided the system having means for segmenting the process into a plurality of process segments, one or more sensors configured to capture information related to each process segment of the plurality of process segments generated by the segmenting means, the information comprising a plurality of samples, and processing means configured to process the plurality of samples related to each process segment of the plurality of segments, and, based on the processing, provide an indication associated with the at least one state of the process.