On monitoring and intelligence in an integrated manufacturing system

Date
2003-04
Authors
Fourie, Cornelius J. (Cornelius Jacobus)
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch : Stellenbosch University
Abstract
ENGLISH ABSTRACT: Some concepts of manufacturing on their own playa decisive role in manufacturing like Integration, Intelligence and Remote Monitoring. They have been tried and tested with great success in various applications in manufacturing. However, very little has been written on the synergy that is created when all three is deployed in one system. It is the aim of this work to survey the attributes of each of these key concepts, to compare them on the grounds of applicability and to study the effects when combined into one system. Final conclusions are made after the hypotheses have been validated with the aid of an experimental model. The first objective of this work is to show how many techniques such as expert systems, fuzzy logic, neural networks and genetic algorithms are used to enable systems to perform intelligently. It is accepted that the competitiveness, growth and profitability of a company in future may depend on the level of its system intelligence. This is so because an intelligent system is able to act appropriately under rapidly changing conditions of customer customisation and demands on quicker throughputs. A further objective of this work is to show how integration adds the element of synergy to a system. This is done by showing several ways of achieving integration by non-technological means like departmental consolidation, plant consolidation, product rationalisation, more flexible working practices, etc. There are as many options for integration by technical means as well, ranging from group technology to process or transfer lines, and from flexible automation such as robots through to hard automation using special-purpose machinery and transfer lines. The third objective is to show how remote monitoring enhances the capabilities of manufacturing systems by synergising with the other two key concepts. With the technology of intelligent manufacturing and integration, larger and more complex manufacturing systems are becoming a reality. However, the danger exists that the shop floor machine tools remain isolated islands of automation. Plant machinery needs to be networked into the enterprise-wide information system. The ability to monitor a variety of process parameters and alert plant staff to changing conditions can greatly reduce downtime. This lack of connectivity therefore represents a huge constraint as far as productivity is concerned. For this reason, there is a great interest to study remote monitoring, analysis and diagnostic systems for application in modem manufacturing. The major contribution of this work is to study the synergy that is created by combining the three key concepts into one system and to validate the findings with the aid of the experimental model. The meaning of validation is to make legally valid; to grant official sanction to; to confirm the validity of something or to declare something as true. To validate is to support or corroborate a theory on a sound or authoritative basis by experiments designed to show a hypothesis as being true. The components of the validation model are a neural network, a simulator, a decision evaluator or critic, and a teacher. The neural network is used to make the decisions. Its inputs are the system parameters and its outputs are a vector of values between 0 and 1, the highest value indicates the decision being made (winner takes all). The simulator executes the decision it obtains from the network and thus changes the state of the system. The evaluator looks at how the system changed due to the decision made by the network and decides whether it was a good or a bad decision. The teacher then adjusts the output of the network accordingly and trains the network with the adjusted outputs. The results of the validation experiments show that intelligence is used to train the model, integration is achieved by combining the elements of the model with the mobile robot and remote monitoring is done by the model to analyse the condition of the system and to react accordingly. The main objective of this work is clearly met in that synergy was shown to be created by the three key concepts.
AFRIKAANSE OPSOMMING: Aspekte soos Integrasie, Intelligensie en Afstandsmonitering speel 'n deurslaggewende rol in vervaardiging en is al op hulle eie met groot sukses in vele toepassings gebruik. Daar is egter nog nie veel aangeteken oor die sinergie wat ontstaan wanneer hulle tesame in een stelsel gebruik word nie. Dit is die doel van hierdie werk om die kenmerke van elk van hierdie sleutel aspekte na te vors, dit op grond van toepaslikheid met mekaar te vergelyk en die uitwerking te bestudeer wanneer hulle in een stelsel saamgevoeg word. Nadat die hipoteses met behulp van 'n eksperimentele model gevalideer is, word finale gevolgtrekkings gemaak. Die eerste doelwit van hierdie werk is om aan te toon dat verskeie tegnieke soos genetiese algoritmes en neurale netwerke gebruik word om stelsels meer kundig te laat optree. Dit word aanvaar dat die toekomstige mededingendheid en groei van ondernemings mag afhang van die stelsel intelligentheidsvlak. Dit is omdat intelligente stelsels gepas kan optree onder snel-veranderende omstandighede. 'n Verdere doelwit is om aan te toon hoe integrasie sinergie kan toevoeg tot 'n stelsel. Dit word gedoen deur verskeie metodes te bespreek van hoe om integrasie op 'n nie-tegniese vlak te bewerkstellig. Die tegniese metodes van integrasie word ook bespreek en sluit tegnieke soos groeptegnologie, aanpasbare outomatisasie en robotika in. Die derde doelwit is om aan te toon hoe afstandsmonitering as sleutel aspek die ander twee sleutel aspekte kan versterk. Die tegnologië van intelligente vervaardiging en integrasie maak die skepping van groter en meer kompleks vervaardigingstelsels nou moontlik. Die gevaar bestaan egter dat hierdie masjiene slegs eilande van outomatisasie sal bly indien hulle nie met behulp van netwerke in die onderneming se inligtingstelselopgeneem word nie. Die vermoë om prosesveranderinge te monitor kan lei tot verminderde staantyd van masjiene en kan dus produktiwiteit verhoog. Om hierdie redes is die toepassing van afstandsmonitering en -diagnosering belangrik vir toepassing in vervaardiging. Die belangrikste bydrae van hierdie werk is die studie van die sinergie wat ontstaan wanneer die drie sleutel aspekte in een stelsel gekombineer word en om die bevindinge te valideer met behulp van 'n eksperimentele model. Om te valideer beteken om iets geldig te verklaar of om die geldigheid van iets te bevestig. Dit beteken verder om 'n teorie te ondersteun of te staaf op 'n grondige en deskundige basis met behulp van eksperimente. Die validasie model bestaan uit 'n neurale netwerk, 'n simulator, 'n besluitevalueerder of beoordelaar, en 'n onderwyser (terugvoerder). Die neurale netwerk neem die besluite met die stelselparameters as inset en die uitset 'n vektor met waardes tussen 0 en 1. Die simulator voer die besluit uit en verander so die toestand van die stelsel. Die evalueerder bepaal hoe die stelsel verander het as gevolg van die besluit en bepaalook of dit 'n goeie of slegte besluit was. Die onderwyser verstel dan die uitset van die netwerk dienooreenkomstig en lei die netwerk op met die verstelde uitsette. Die resultate van die validasie eksperiment toon aan dat intelligensie gebruik word om die modelop te lei, integrasie behaal word deur die elemente van die model te kombineer met die mobiele robot en afstandsmonitering toegepas word deur die toestand van die stelsel te monitor en te analiseer. Die hoofdoelwit van hierdie werk word dus duidelik behaal deur die beskrywing van die sinergie wat ontstaan deur die kombinasie van die drie sleutel aspekte.
Description
Thesis (PhD)--University of Stellenbosch, 2003.
Keywords
Manufacturing processes, Computer integrated manufacturing systems, Dissertations -- Industrial engineering, Theses -- Industrial engineering
Citation