Designing a Multivariate Process Control Procedures for Production System Case of Ethio Cement PLC
No Thumbnail Available
Date
2023-05
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
This dissertation explores the application of Statistical Process Control (SPC)
techniques in the manufacturing sector. Industries demand multivariate process
monitoring technique capable of identifying cause of variation, and conducting fast
and accurate fault detection analysis. However, the existing techniques fall short of
satisfying this demand. Hence, the research question is devised as follows: how to
design a multivariate process control procedure that can effectively monitor and
control the production system, identify the root causes of variation, and provide
solutions for improvement. The literature review conducted in this study revealed that
while SPC techniques have been extensively studied and applied in various industries,
the multivariate analysis of identifying cause of variation is relatively limited. Their
practical implementation and adaptation to the industry have not been thoroughly
explored. The main objective of this research is to design a procedure that can bridge
the theoretical gap that exist in the manufacturing sector. By addressing this gap, it is
anticipated that the productivity, quality, and overall performance of the production
system can be improved. To address these limitations, a novel approach called the
GANNT chart is introduced in this research. The GANNT chart incorporates three
key theories: Graph theory (G), Artificial Neural Networks (ANN), and Hotelling T2
(T). By combining these theories, the proposed approach aims to enhance the process
control technique used in the production system. The GANNT chart mimics human
decision-making processes and serves as a decision support system for both process
engineers and operators.The GANNT chart methodology offers several advantages. Firstly, it analyzes the
correlation effects between variables using Hotelling T2, allowing for a more
comprehensive understanding of process variation. Secondly, it leverages graph
theories to retain and utilize knowledge from previous successful operations,
facilitating continuous improvement. Lastly, the system is trained using Artificial
Neural Networks, enabling it to provide solutions to future challenges based on
learned patterns from past operations. The proposed model is validated in the cement
industry to assess its effectiveness and practicality. The results demonstrate that the
GANNT chart effectively addresses the identified gaps in the application of SPC
techniques to the cement production process. The model's ability to accurately detect
process deviations and provide insights into the causes of variation contributes to
improved productivity, quality, and overall performance. As a future research
direction, this study highlights two suggestions. The first one is examining and
extending the assumptions to design this model in such a way that it considers
different scenarios not covered by this research. The second direction is extend the
implementation of GANNT chart to various industries, including service giving
industries, and study and explore its applicability. In conclusion, bridging the gap
between theory and practice, this research aims to contribute to the advancement of
multivariate process control to the industry, ultimately leading to enhanced
operational efficiency and product quality.
Description
Keywords
Neural Network, Graph theory, Hotelling T2, GANNT chart, VARIMA model, cement process control