Publication Date
Spring 2024
Degree Type
Thesis
Degree Name
Master of Science (MS)
Department
Computer Engineering
Advisor
Mahima Agumbe Suresh; Jorjeta Jetcheva; Armin Moghadam
Abstract
Process monitoring is regarded as one of the most important aspects of the manufacturing industry. It involves monitoring and controlling the many phases of production to uphold high standards and foster a culture of continuous development. A company can establish a delicate balance between quality assurance, yield optimization, and productivity increase by having this key role. In a typical manufacturing business, the usual methods for process monitoring would involve employing physical sensors and human labor for inspections which are time consuming, prone to errors, less efficient and adds to the cost of the final product. Synthetic data approach can be leveraged as a potential solution to these problems by generating substantial quantities of training data, which can then be analyzed by deep learning algorithms to identify complex patterns and correlations. The purpose of this research is to use the outcomes of deep learning models that have been trained on simulated data to analyze real-world data and assess their efficacy. This research will highlight that deep learning models can be effectively trained on synthetic data, achieving the same level of performance as models trained on real data. We observed that the methods used to generate the synthetic data may have a significant influence on the performance of the model. The findings may be assessed to demonstrate the capacity of synthetic data and deep learning to improve process monitoring in production and shed light on the practical challenges that must be considered when using these techniques in the workplace.
Recommended Citation
Bhatia, Sahib S., "Synthetic Data and Deep Learning for Process Monitoring in Manufacturing" (2024). Master's Theses. 5495.
DOI: https://doi.org/10.31979/etd.vmqk-ncaw
https://scholarworks.sjsu.edu/etd_theses/5495
