Adaptable algorithms for performance optimization of dynamic batch manufacturing processes

Teo, Kenneth Tze Kin (2018) Adaptable algorithms for performance optimization of dynamic batch manufacturing processes. Doctoral thesis, Universiti Malaysia Sabah.

[img] Text
24 PAGES.pdf

Download (527kB)
[img] Text
FULLTEXT.pdf
Restricted to Registered users only

Download (6MB)

Abstract

This study aims to explore the potential of implementing a multi-objective dynamic optimizer to acquire regular optimization on nonlinear characteristic of batch manufacturing process. The idea is proposed in an effort of accessing the functionality and practicability of the dynamic optimizer for the purpose of precision optimization. As such, yield productivity of maximizing the desired product while minimizing the undesired by-product could be improved. Traditionally, a fixed temperature profile has been identified and assumed as ‘nominal optimized’ for manufacturers to follow. Whether the process is fully optimized under inconsistent internal heat liberation, external environmental variations, model mismatches and process uncertainties remains as a challenging topic. In practice, the commercial batch process plant are often utilized to handle numerous production with different varieties of specific products and thus it is rarely being classified as dynamically optimized. This thesis investigates different approaches of integrating hybrid adaptable intelligent algorithms to accommodate the concept of precision optimization via simulated models of industry-scale and pilot-scale. The dynamic changes causing the need of dynamic modelling for a better dynamic optimization will be catered via a specifically formulated fitness function. With nowadays high end computation ability, revolutionary changes of implementing precision measurement is expectable and applicable to obtain expensive products. Central to precision manufacturing is artificial intelligence as this thesis presents the performance characteristics of tuning-based, rule-based, learning-based and evolutionary-based algorithms. Performance analyses are presented to validate how significant of the dynamic optimization based on the product molarity surpasses the existing conventional prescribed temperature approach. Various algorithms are designed, formulated and computed in MATLAB and then embedded to a real-time programmable microcontroller. The results indicate that the proposed algorithm is able to improve the percentage of yield by at least 10-15% under physical properties variation which could be a vital tool in the future era of precision manufacturing industries.

Item Type: Thesis (Doctoral)
Keyword: Multi-objective optimization, Dynamic optimizer, Batch manufacturing process, Precision optimization, Yield productivity, Nonlinear characteristics, Artificial intelligence, MATLAB, Microcontroller, Hybrid algorithms, Manufacturing optimization
Subjects: Q Science > QA Mathematics > QA1-939 Mathematics > QA1-43 General
Department: FACULTY > Faculty of Engineering
Depositing User: DG MASNIAH AHMAD -
Date Deposited: 24 Feb 2025 09:35
Last Modified: 24 Feb 2025 09:35
URI: https://eprints.ums.edu.my/id/eprint/42830

Actions (login required)

View Item View Item