Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.780250
Title: Performance modelling and adaptive control for linked sequential systems
Author: Morley, Paul Nicholas
Awarding Body: Open University
Current Institution: Open University
Date of Award: 2013
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
This thesis investigates the dynamics of linked sequential systems of machines in industrial laundries. Two aspects are considered: firstly the control of such systems and in particular the decision making point when a batch to be processed can be sent to one of many identical machines, and secondly the modelling of the whole system of linked machines. The decision making point in the control of these systems is frequently implemented in a sub-optimal manner, or a manner which becomes sub-optimal as conditions change. An adaptive system is preferable and an Evolutionary Artificial Neural Network approach (EANN) is proposed. The EANN is tested on simulations of real laundry systems and shown to be effective. Then it is applied to two abstract game playing problems in order to better understand its limitations. Limitations are found to include the fact that if learning does not appear to take place, it is not possible to determine if this is a failure of the Evolutionary approach or the Artificial Neural Network parameters. The dynamics and performance of Linked Sequential Systems in Industrial Laundries are not well understood or covered by theory in the literature. The theory of the performance of these systems is outlined, and an Agent Based Model (ABM) simulation presented. The ABM simulation is explained and then the simulation is compared to a real world system in an existing laundry. The performance of the existing system is measured and compared to the prediction of the ABM simulation. The ABM simulation is shown to offer a better understanding of the system than the previous static calculation. Finally the ABM is used in a design exercise to show how it could be used to specify a system more accurately than the static calculation at design stage.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.780250  DOI: Not available
Share: