Display Panel Transistor Characterization With ML Noise Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing display panel fabrication technologies face challenges in accurately detecting and analyzing the driving characteristics of switching transistors, which are crucial for precise control of light-emitting elements in organic light-emitting display devices, due to noise interference in measurement processes.
Innovation Solution
An apparatus and method utilizing a primary learning model to remove noise during encoding and decoding processes, and a secondary learning model to refine the detected characteristics, creating a database for accurate classification and correction of switching transistor parameters, enabling precise detection and correction of output current, voltage, and threshold voltage ranges.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional measurement methods are used to detect switching transistor characteristics, then the measurement process is simple, but noise interference reduces measurement precision
Solution Approach 1:
The patent introduces learning models (primary and secondary) as intermediary components between the measurement module and the final characteristic detection results. These learning models process the raw measurement data, remove noise through encoding-decoding operations, and generate corrected characteristic values, thereby eliminating noise interference while maintaining measurement simplicity
Solution Approach 2:
The patent replaces traditional signal filtering methods with machine learning-based noise removal. The learning models use data-driven approaches to distinguish signal from noise, substituting conventional mechanical/electrical filtering mechanisms with intelligent algorithms that adaptively remove noise while preserving measurement accuracy
2Measurement precision
If multiple learning models and processing steps are introduced to remove noise and correct characteristics, then measurement precision improves, but device complexity increases
Solution Approach 1:
The patent divides the noise removal and characteristic correction process into distinct segments: a primary learning model for initial noise removal, encoding-decoding processing for further refinement, and a secondary learning model for final correction. This segmentation allows each component to focus on specific tasks, improving overall precision while maintaining manageable system complexity through modular architecture
Solution Approach 2:
The patent implements a continuous processing chain where measurement data flows through multiple learning models and processing steps without interruption. The primary learning model continuously processes raw data, the encoding-decoding operations continuously refine the results, and the secondary learning model continuously generates corrected characteristics, ensuring uninterrupted high-precision measurement while optimizing system efficiency
3Manufacturing precision
If learning models are used to correct switching transistor characteristics, then manufacturing precision improves, but loss of time in processing increases
Solution Approach 1:
The patent performs preliminary noise removal using the primary learning model before subsequent processing steps. By removing the majority of noise early in the process, the encoding-decoding operations and secondary learning model work with cleaner data, reducing their processing time and computational load, thereby minimizing overall time loss while maintaining high manufacturing precision
Solution Approach 2:
The patent applies partial processing to different aspects of the measurement data: the primary learning model handles noise removal, encoding-decoding handles intermediate refinement, and the secondary learning model handles final correction. This partial action approach processes only the necessary aspects at each stage rather than reprocessing all data, reducing total processing time while achieving comprehensive precision improvement
Data Source
AI summary
An apparatus for fabricating a display panel includes: a sample module in which a switching transistor is formed, a measurement module having input/output terminals electrically connected to the switching transistor of the sample module, a characteristic detecting unit configured to detect primary operation characteristic information including an output current, an output voltage value and a threshold voltage range of the switching transistor, a model learning unit configured to correct at least one of the output current, the output voltage value and the threshold voltage range in the primary operation characteristic information using a learning program included in at least one learning model, and to extract correction results as learning result data, and a correction learning unit configured to classify the learning result data containing the primary operation characteristic information according to fabrication characteristics of sample switching transistors and a list of classifications.