Multi-Section Neural Network for Time-Series Abnormality Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing abnormality detection techniques in manufacturing processes, such as semiconductor production, struggle to achieve high precision in identifying abnormalities from time series data sets.
Innovation Solution
The development of an abnormality detecting device that utilizes a neural network with multiple network sections and a concatenation section to process time series data sets, allowing for the training of a model that accurately identifies abnormality levels by comparing combined output results with predefined abnormality levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single neural network is used for abnormality detection, then the device complexity is low, but the measurement precision of abnormality detection is insufficient
Solution Approach 1:
The neural network is divided into multiple independent network sections (first network section, second network section, etc.), each processing different aspects of the time series data. This segmentation allows each section to specialize in detecting specific patterns while maintaining overall system manageability and improving detection precision through diversified analysis.
Solution Approach 2:
Multiple network sections are combined through a concatenation section that integrates their output data. This merging strategy consolidates the detection results from different network sections, creating a comprehensive abnormality detection model that leverages the strengths of each individual section while achieving high precision.
2Measurement precision
If multiple network sections are used to process time series data sets, then the measurement precision improves, but the device complexity increases
Solution Approach 1:
The complex neural network is segmented into multiple functional network sections, each handling specific processing tasks. This segmentation reduces the complexity burden on individual components while maintaining high overall precision through coordinated operation of multiple specialized sections.
Solution Approach 2:
Each network section is designed to process time series data sets with universal applicability to different abnormality types. The modular design allows the same structural pattern to be replicated across multiple sections, reducing configuration complexity through standardization while maintaining detection accuracy.
3Measurement precision
If time series data sets are processed through multiple network sections with concatenation, then the abnormality detection precision increases, but the processing time increases
Solution Approach 1:
Multiple network sections process time series data sets in parallel rather than sequentially, performing preliminary analysis simultaneously. This parallel processing approach eliminates the time penalty of multiple processing stages while maintaining the precision benefits of comprehensive analysis through the concatenation section.
Data Source
AI summary
An abnormality detection device trains a model using multiple network sections each configured to process acquired time series data sets and a concatenation section configured to combine output data output from each of the multiple network sections and to output, as a combined result, a result of combining the output data output from each of the multiple network sections. The trained model is then applied to adapt a unit of process performed during manufacture of a processed object.