High energy price and the increasing requirements of quality and low cost of products have created an urgent need to implement new technologies in current automated manufacturing environments. Condition monitoring systems of manufacturing processes have been recognised in recent years as one of the essential technologies that provide the competitive advantage in many manufacturing environments. This research aims to develop an effective sensor fusion model for turning processes for the detection of tool wear. Multi-sensors combined with a novelty detection algorithm and Learning Vector Quantisation (LVQ) neural networks are used in this research to detect tool wear and provide diagnostic and prognostic information. A novel approach, termed ASPST, (Automated Sensor and Signal Processing Selection System for Turning) is used to select the most appropriate sensors and signal processing methods. The aim is to reduce the number of sensors needed in the overall system and reduce the cost. The ASPST approach is based on simplifying complex sensory signals into a group of Sensory Characteristic Features (SCFs) and evaluating the sensitivity of these SCFs in detecting tool wear. A wide range of sensory signals (cutting forces, strain, acceleration, acoustic emission and sound) and signal processing methods are also implemented to verify the capability of the approach. A cost reduction method is also implemented based on eliminating the least utilised sensor in an attempt to reduce the overall cost of the system without sacrificing the capability of the condition monitoring system. The experimental results prove that the suggested approach provides a responsive and effective solution in monitoring tool wear in turning with reduced time and cost.

Industrial technology has excelled profoundly in the past few decades, helping organisations throughout the world to be more efficient in all processes and keeping costs down. However, despite the abundance of several IT solutions, there exist many problems where more than one decision has to be made. Among the techniques supporting a multi-decisional context, simulations can undoubtedly play an important role as they provide what-if analysis and hence help to evaluate quantitative benefits. This thesis develops a simulation model for breakdown in an industrial machine, the main crusher in a cement factory. It also examines three important parameters (Drill Head, Dusting and Lubrication) of the crusher machine with the use of Bayesian network modelling which allows determination of suitable influencing factors in a precise and dynamic manner. The model also supports integration with management systems such as J.I.T, and MRPII. Witness simulation software has been used in this work to model the breakdown frequency of the Crusher machine and the associated parameters. The Bayesian Network Modelling is used to consider historical data and expert opinions; the Bayes’ approach takes into consideration off all existing parameters that affect the machine breakdown directly or indirectly. This tool is capable of establishing a probability based on the information gathered about the parameters. The simulation model is developed further to enable the Bayesian Network Modelling to be applied via the Chain Rule to calculate the probability of failure. The findings of this research show the approach developed in this work, where the Bayesian probability development process is integrated into the simulation model. This provides a unique and dynamic tool to aid decision making in understanding machine breakdowns. The resulting simulator is a decision making tool capable of analysing the status of the machine and the associated influencing factors. This uses an approach based on multiple performance measures and a user-defined set of inputs based on historical and expert opinion. This work provides a methodology to study the importance of key parameters of the crusher machine. This in effect highlights the correlation between the governing parameters and the occurrence of breakdown.

No description available.

No description available.

No description available.

Manufacturing enterprises need to have agile manufacturing automation systems that can be responsive to changes in demand and operate efficiently within the context of a global business economy. Manufacturing automation systems that possess the capabilities to support flexibility, reconfigurability and reusability would enable enterprises to have a competitive edge in their markets. However, traditional approaches to manufacturing automation systems development and implementation are typically rigid in nature and thus do not provide the capabilities to adequately support the required levels of agility. This thesis has specified and developed a paradigm for developing and implementing an agile manufacturing automation system.

The development of next generation manufacturing systems is currently an active area of research worldwide. Globalisation is placing new demands on the manufacturing industry with products having shorter lifecycles and being required in more variants. Manufacturing systems must therefore be agile to support frequent manufacturing system reconfiguration involving globally distributed engineering partners. The research described in this thesis addresses one aspect within this research area, the Human Machine Interface (HMI) system that support the personnel involved in the monitoring, diagnostics and reconfiguration of automated manufacturing production machinery. Current HMI systems are monolithic in their design, generally offer poor connectivity to other manufacturing systems and require highly skilled personnel to develop and maintain them. The new approach established in the research and presented in this thesis provides a specification capture technique (using a novel storyboarding modelling notation) that enables the end users HMI functionality to be specified and rapidly developed into fully functional End User HMI's via automated generation tools. A novel feature in this HMI system architecture that all machine information is stored in a common unified machine data model which ensures consistent accurate machine data is available to all machine lifecycle engineering tools including the HMI. The system's run-time architecture enables remote monitoring and diagnostics capabilities to be available to geographically distributed engineering partners using standard internet technologies. The implementation of this novel HMI approach has been prototyped and evaluated using the industrial collaborators full scale demonstrator machines within cylinder head machining and engine assembly applications.

No description available.

Researchers posit that maximum benefit will accrue if there is a fit between the firm's strategies and its Advanced Manufacturing Technology (AMT) difftision. The existing literature discusses how the investment of AMT should be consistent with manufacturing strategy, however little attention has been given to an empirical examination of the link between manufacturing strategy and AMT, and the implications of the fit between these two on manufacturing performance. This thesis addresses that gap in the literature and thus seeks to examine the link between manufacturing strategy and AMT, and how the alignment between these two variables can affect the manufacturing performance.

Ce sujet de thèse concerne de manière générale l'évaluation des performances et l'ordonnancement dans des systèmes de production flexibles et principalement les problèmes d'ordonnancement d'atelier de type Flow-Shop et Flow-Shop hybride. Le problème d'ordonnancement d'un Flow-Shop peut être défini ainsi : un ensemble de N jobs composés chacun de M opérations, doivent passer sur M machines dans le même ordre. Une machine peut exécuter une seule opération à la fois, chaque job ne peut avoir qu'une seule opération en cours de réalisation simultanément et la préemption n'est pas autorisée. Dans le cas des Flow-Shops hybrides, Mk machines identiques sont disponibles à chaque étage k en un ou plusieurs exemplaires. Pour cette étude, notre objectif est toujours de minimiser le temps total d'exécution aussi appelé makespan. Les problèmes d'ordonnancement les plus répandus sont de type Flow-Shop classique où les espaces de stockage entre les machines sont considérées comme infinies. D'autres problèmes sont caractérisés par des capacités de stockage limitées ou nulles qui engendre une seule contrainte de blocage. Cette contrainte peut être un blocage classique (de type RSb) ou particulier (de type RCb ou RCb*). Dans nos travaux de recherche, nous présentons un cas général qui peut être tiré de l'industrie et modélisé sous forme de systèmes de type Flow-Shop et Flow-Shop hybride soumis simultanément à plusieurs types de blocage. Pour résoudre ce genre de problèmes, nous avons étudié dans cette thèse la complexité de ces systèmes et nous avons proposé des méthodes exactes, des méthodes approchées ainsi que des bornes inférieures / This thesis deals mainly with makespan minimization in Flow-Shop and hybrid Flow-Shop scheduling problems where mixed blocking constraints are considered. In Flow-Shop scheduling problem, a set of N jobs must be executed on a set of M machines. All jobs require the same operation order that must be executed according to the same manufacturing process. Each machine can only execute one job at any time. Pre-emptive operation is not authorized in presented work. In case of hybrid Flow-Shop, at any processing stage k, there exist one or more identical machines Mk. Objective function consists in determining best schedule in order to reduce makespan, i.e. time where all operations are completed. The most common scheduling problem is classical flowshop where buffer space capacity between machines is considered as unlimited. Other problems are characterized by the fact that the storage capacity is limited or null and which generates one blocking constraint. This constraint can be a classical blocking (RSb) or particular blocking (RCb or RCb*). In our works, we present a general case which can be derived from industry and modeled as Flow-Shop and hybrid Flow-Shop systems subject simultaneously to different blocking. To solve these problems, we studied in this thesis complexity of these systems and we proposed exact methods, approached methods and lower bounds