Improving Modifiability and Performance of an Automated Display Testing Software Architecture
Ojaluoma, Tuomo (2020)
Ojaluoma, Tuomo
2020
Automaatiotekniikan DI-tutkinto-ohjelma - Degree Programme in Automation Engineering, MSc (Tech)
Tekniikan ja luonnontieteiden tiedekunta - Faculty of Engineering and Natural Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2020-05-12
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202004294368
https://urn.fi/URN:NBN:fi:tuni-202004294368
Tiivistelmä
A need to objectively measure image quality of display devices has created a demand to automate the process. One of the solutions for this automation task is a display inspection system developed by OptoFidelity called GoldenEye. The system has demonstrated its ability to measure image quality and defects but it has also become evident that the system should perform the measurements faster while also being more flexible in terms of hardware and measurement configurations.
This thesis takes the existing platform and designs a new software architecture to both complete the tasks the earlier architecture was set to do while also searching for solutions for speed improvements and better modifiability which should enable the architecture to support multiple hardware configurations and deployment environments.
The thesis goes through different tools and techniques that are available for the architectural design and the theoretical background behind them. Some existing solutions, standards and research from the field of factory automation are explored to see how they can be used in the design work. An OKD-MES concept is chosen as a basis for the new architecture and the software logic and functionality is divided into separate blocks based on it.
The basic design of each of the blocks is covered in the thesis and in the end,some estimations are given how well the new architecture answers the goals set to it. The execution time for a single measurement and analysis is measured both for the old and new architecture to get an estimation on how much the new architecture improves the performance. In the tests, the measurement-analysis sequence execution time was cut by 82% and the analysis time was cut down by 97%.
The significant performance increases proved that traditional multiprocessing is much more efficient solution for image analysis parallellization than a task queue based approach. Abstracting the software into blocks based on OKD-MES also seems to increase the modifiability of the software. Particularly the separation of sequence generation to it’s own block increases the ability to quickly change the behaviour of the measurement software
This thesis takes the existing platform and designs a new software architecture to both complete the tasks the earlier architecture was set to do while also searching for solutions for speed improvements and better modifiability which should enable the architecture to support multiple hardware configurations and deployment environments.
The thesis goes through different tools and techniques that are available for the architectural design and the theoretical background behind them. Some existing solutions, standards and research from the field of factory automation are explored to see how they can be used in the design work. An OKD-MES concept is chosen as a basis for the new architecture and the software logic and functionality is divided into separate blocks based on it.
The basic design of each of the blocks is covered in the thesis and in the end,some estimations are given how well the new architecture answers the goals set to it. The execution time for a single measurement and analysis is measured both for the old and new architecture to get an estimation on how much the new architecture improves the performance. In the tests, the measurement-analysis sequence execution time was cut by 82% and the analysis time was cut down by 97%.
The significant performance increases proved that traditional multiprocessing is much more efficient solution for image analysis parallellization than a task queue based approach. Abstracting the software into blocks based on OKD-MES also seems to increase the modifiability of the software. Particularly the separation of sequence generation to it’s own block increases the ability to quickly change the behaviour of the measurement software