Thesis
PREDICTION OF GREEN CONCRETE QUALITY USING MACHINE LEARNING AND COBOT
Washington State University
Master of Science (MS), Washington State University
01/2021
DOI:
https://doi.org/10.7273/000001867
Handle:
https://hdl.handle.net/2376/122187
Abstract
Concrete is a construction product that has been used for thousands of years. The compressive strength properties as well as its workability and low cost make concrete a leading material in structural applications. Typical quality tests for concrete include compression tests, where the concrete is allowed to cure for 7, 14, or 28 days. It is desirable to quickly and accurately assess the quality of the uncured no slump concrete moments after manufacturing. The benefits include reassuring customers about the concrete quality while saving time, money, resources and identifying inadequate product early in the production stages. Currently, an operator stands by the machine to check quality by visually inspecting and poking the concrete blocks. The same tactile and visual approach is repeated for all product and after machine adjustments. The approach is highly subjective as the decisions vary due to several factors. In this research, we aim to develop a robotic quality inspection station using a cobot equipped with a force sensing finger to automate the inspection process. The research involved several setups to collect data for a machine learning (ML) model utilizing the K-nearest neighbor (KNN) algorithm. The model has been incorporated into the cobot control so that the cobot can evaluate blocks in real time against the ML model to determine good or bad block quality. Design of the inspection systemalong with the experiments conducted with quartz flour and concrete blocks are explained. Details of the ML model and the integration into the cobot operation are discussed. Results from automatic quality assessment experiments with the cobot are compared to operator predictions of quartz flour and concrete blocks. Recommendations are made to continue with the development of the inspection system.
Metrics
13 File views/ downloads
97 Record Views
Details
- Title
- PREDICTION OF GREEN CONCRETE QUALITY USING MACHINE LEARNING AND COBOT
- Creators
- Aaron James Burke
- Contributors
- Hakan Gurocak (Advisor)Xiaolin (Linda) Chen (Committee Member)Jong-Hoon Kim (Committee Member)
- Awarding Institution
- Washington State University
- Academic Unit
- Engineering and Computer Science (VANC), School of
- Theses and Dissertations
- Master of Science (MS), Washington State University
- Publisher
- Washington State University
- Number of pages
- 58
- Identifiers
- 99900606554501842
- Language
- English
- Resource Type
- Thesis