Date of Award
12-2013
Document Type
Thesis
Degree Name
Master of Science (MS)
Legacy Department
Mechanical Engineering
Committee Chair/Advisor
Gregory M. Mocko
Committee Member
Lonny L. Thompson
Committee Member
Mary E. Kurz
Abstract
The objective of this thesis is to design, develop and implement an automated approach to support processing of historical assembly data to extract useful knowledge about assembly instructions and time studies to facilitate the development of decision support systems, for a large automotive original equipment manufacturer (OEM). At a conceptual level, this research establishes a framework for sustainable and scalable approach to extract knowledge from big data using techniques from Natural Language Processing (NLP) and Machine Learning (ML). Process sheets are text documents that contain detailed instructions to assemble a portion of the vehicle, specification of parts and tools to be used, and time study. To maintain consistency in the authorship process, assembly process sheets are required to be written in a standardized structure using controlled language. To realize this goal, 567 work instructions from 236 process sheets are parsed using Stanford parser using Natural Language Toolkit (NLTK) as a platform and a standard vocabulary consisting of 31 verbs is formed. Time study is the process of estimating assembly times from a predetermined motion time system, known as MTM, based on factors such as the activity performed by the associate, difficulty in assembling, parts and tools used, distance covered. The MTM compromises of a set of tables, constructed through statistical analysis and best-suited for batch production. These MTM tables are suggested based on the activity described in the work instruction text. The process of performing time studies for the process sheets is time consuming, labor intensive and error-prone. A set of (IF AND THEN ) rules are developed, by analyzing 1019 time study steps from 236 process sheets, that guide the user to an appropriate MTM table. These rules are computationally generated by a decision tree algorithm, J48, in WEKA, a machine learning software package. A decision support tool is developed to enable testing of the MTM mapping rules. The tool demonstrates how NLP techniques can be used to read work instructions authored in free-form text and provides MTM table suggestions to the planner. The accuracy of the MTM mapping rules is found to be 84.6%.
Recommended Citation
Koneru, Abhiram, "Knowledge Extraction from Work Instructions through Text Processing and Analysis" (2013). All Theses. 1769.
https://open.clemson.edu/all_theses/1769