Uma vez que uma aplicação ou projeto tenha sido identificado como sujeito ao uso da simulação, decisões devem ser tomadas acerca de como conduzir os estudos. Embora não haja regras definitivas, alguns passos são geralmente recomendados, tais como: planejamento do estudo, definição do sistema, construção do modelo, execução dos experimentos, análise dos resultados e relatório final. E, a construção do modelo é uma das etapas mais demoradas e complexas. Na execução dos experimentos, e na análise dos resultados a Realidade Virtual (RV) é uma interface que pode fornecer um suporte importante para a tomada de decisões, pois a RV auxilia na criação de mundos virtuais semelhantes ao mundo real, que ajudam na compreensão do funcionamento dos sistemas. Este trabalho propõe um metamodelo para o desenvolvimento de sistemas de simulação de eventos discretos com interface de RV aplicados a ambientes ou estações de trabalho de manufatura. A proposta do metamodelo é composto por quatro módulos: i) de edição e validação de modelos, ii) de criação, edição e execução de ambientes virtuais, iii) de conexão para efetuar o elo entre os dois primeiros módulos e, iv) de controle para gerenciar a comunicação e controlar a simulação. Como resultado tem-se a simulação centralizada baseada no modelo descrito em Rede de Petris (RP) com interface de RV distribuída. O sistema computacional gerado pelo metamodelo auxiliará no aprendizado e compreensão do problema simulado, e ainda permitirá ao usuário o envolvimento através da iteração. / The VR aids in the creation of virtual worlds similar to the real world, helping in understanding how a system works. This research work is concerned with the development of a discrete-event system simulation based on Petri Nets, with Virtual Reality interface for manufacturing environment or workstations. The system is composed of four modules, the first is model editing and validation, the second module is a modeling tool for virtual objects and/or a library of virtual elements, the third module makes the link between the two previous modules, and finally, the fourth is a simulation and control module. The centralized simulation is based on models described by Petri Nets with distributed RV interface resulting in a flexible and consistent system. The system will help users in learning and understanding the simulation problem, and it will also allow the user\'s integration with the environment through the interaction, and through distributed interface.

Cette thèse contribue à l’identification de modèles compacts et expressifs de Systèmes à Evénements Discrets (SED) réactifs, à des fins de rétro-conception ou de certification. L’identification est passive, et boîte-noire, la connaissance se limitant aux signaux d’entrées/sorties du système. Les Réseaux de Petri Interprétés (RdPI) permettent de modéliser à la fois le comportement observable (causalités entrées/sorties observées directement), et le comportement non observable du système (évolutions de variables internes). Cette thèse vise à identifier des modèles RdPI à partir d’une séquence observée de vecteurs entrées/sorties. Notamment, l’enjeu étant de traiter des systèmes concurrents de taille réaliste, l’approche développée permet le passage à l’échelle de résultats précédents.La construction de la partie observable est d’abord améliorée par l’ajout d’un filtre. Celui-ci détecte et supprime les synchronisations parasites causées par le contrôleur en présence de systèmes concurrents. Une nouvelle approche est ensuite proposée pour découvrir la partie non observable, basée sur l’utilisation de projections, et garantissant la reproductibilité du comportement observé malgré la concurrence. Une heuristique permet de construire un modèle satisfaisant pour la rétro-ingénierie, à coût de calcul limité. Enfin, une approche distribuée est proposée pour réduire davantage le coût de calcul, en partitionnant automatiquement le système en sous-systèmes. L’effet cumulatif de ces contributions est illustré par l’identification de RdPI sur un système de taille raisonnable, validant leur efficacité. / This thesis proposes a method to identify compact and expressive models of closed-loop reactive Discrete Event Systems (DES), for reverse-engineering or certification. The identification is passive, and blackbox, accessible knowledge being limited to input/output signals. Interpreted Petri Nets (IPN) represent both the observable behaviour (direct input/output causalities) and the unobservable behaviour (internal state evolutions) of the system. This thesis aims at identifying IPN models from an observed sequence of I/O vectors. The proposed contributions extend previous results towards scalability, to deal with realistic systems who exhibit concurrency.Firstly, the construction of the observable part of the IPN is improved by the addition of a filter limiting the effect of concurrency. It detects and removes spurious synchronizations caused by the controller. Then, a new approach is proposed to improve the discovery of the unobservable part. It is based on the use of projections and guarantees the reproduction of the observed behaviour, despite concurrency. An efficient heuristic is proposed to compute a model adapted to reverse-engineering, limiting the computational cost. Finally, a distributed approach is proposed to further reduce the computational cost, by automatically partitioning the system into subsystems. The efficiency of the cumulative effect of these contributions is validated on a system of realistic size.

The Water Recovery System (WRS) is one of the necessary subsystems in an Advanced Life Support program. It regenerates potable water by processing wastewater generated on a space shuttle or on a space station. The Biological Water Processor (BWP) is one of the complex subsections of WRS that involves processes described by continuous time as well as discrete event dynamics. Such systems can be modeled as hybrid dynamical systems in MATLAB/Stateflow. The following state variables of the BWP have been simulated in MATLAB: Pressure drop across Organic Carbon Oxidation Reactor (OCOR), Pressure drop across nitrification reactor, nitrifier flow rate, total flow to OCOR, Gas-liquid separator level, feed pump modes. The developed model of the BWP can be used for the simulation of the complete WRS and also for synthesis of a distributed control laws for the BWP subsection.

<p> Hierarchical Interface-based Supervisory Control (HISC) decomposes a discrete-event system (DES) into a high-level subsystem which communicates with n ≥ 1 low-level subsystems, through separate interfaces which restrict the interaction of the subsystems. It provides a set of local conditions that can be used to verify global conditions such as nonblocking and controllability. As each clause of the definition can be verified using a single subsystem, the complete system model never needs to be stored in memory, offering potentially significant savings in computational resources.</p> <p> Currently, a designer must create the supervisors for a HISC system himself, and then verify that they satisfy the HISC conditions. In this thesis, we develop a synthesis method that respects the HISC hierarchical structure. We replace the supervisor for each level by a corresponding specification DES. We then do a per level synthesis to construct for each level a maximally permissive supervisor that satisfies the corresponding HISC conditions.</p> <p> We define a set of language based fixpoint operators and show that they compute the required level-wise supremal languages. We then present algorithms that implement the fixpoint operators. We present a complexity analysis for the algorithms and show that they potentially offer significant improvement over the monolithic approach.</p> <p> A large manufacturing system example (estimated worst case state space on the order of 10^22) extended from the AIP example is discussed. A software tool for synthesis and verification of HISC systems using our approach was also developed.</p> / Thesis / Master of Applied Science (MASc)

abstract: The adoption of the Service Oriented Architecture (SOA) as the foundation for developing a new generation of software systems - known as Service Based Software Systems (SBS), poses new challenges in system design. While simulation as a methodology serves a principal role in design, there is a growing recognition that simulation of SBS requires modeling capabilities beyond those that are developed for the traditional distributed software systems. In particular, while different component-based modeling approaches may lend themselves to simulating the logical process flows in Service Oriented Computing (SOC) systems, they are inadequate in terms of supporting SOA-compliant modeling. Furthermore, composite services must satisfy multiple QoS attributes under constrained service reconfigurations and hardware resources. A key desired capability, therefore, is to model and simulate not only the services consistent with SOA concepts and principles, but also the hardware and network components on which services must execute on. In this dissertation, SOC-DEVS - a novel co-design modeling methodology that enables simulation of software and hardware aspects of SBS for early architectural design evaluation is developed. A set of abstractions representing important service characteristics and service relationships are modeled. The proposed software/hardware co-design simulation capability is introduced into the DEVS-Suite simulator. Exemplar simulation models of a communication intensive Voice Communication System and a computation intensive Encryption System are developed and then validated using data from an existing real system. The applicability of the SOC-DEVS methodology is demonstrated in a simulation testbed aimed at facilitating the design & development of SBS. Furthermore, the simulation testbed is extended by integrating an existing prototype monitoring and adaptation system with the simulator to support basic experimentation towards design & development of Adaptive SBS. / Dissertation/Thesis / Ph.D. Computer Science 2011

Au CERN (Organisation européenne pour la recherche nucléaire), le contrôle et la supervision du plus grand accélérateur du monde, le LHC (Large Hadron Collider), sont basés sur des solutions industrielles (SCADA). Le LHC est composé de sous-systèmes disposant d’un grand nombre de capteurs et d’actionneurs qui rendent la surveillance de ces équipements un véritable défi pour les opérateurs. Même avec les solutions SCADA actuelles, l’occurrence d’un défaut déclenche de véritables avalanches d’alarmes, rendant le diagnostic de ces systèmes très difficile. Cette thèse propose une méthodologie d’aide au diagnostic à partir de données historiques du système. Les signatures des défauts déjà rencontrés et représentés par les listes d’alarmes qu’ils ont déclenchés sont comparées à la liste d’alarmes du défaut à diagnostiquer. Deux approches sont considérées. Dans la première, l’ordre d’apparition des alarmes n’est pas pris en compte et les listes d’alarmes sont représentées par un vecteur binaire. La comparaison se fait à l’aide d’une distance pondérée. Le poids de chaque alarme est évalué en fonction de son aptitude à caractériser chaque défaut. La seconde approche prend en compte l’ordre d’apparition des alarmes, les listes d’alarmes sont alors représentées sous forme de séquences symboliques. La comparaison entre ces deux séquences se fait à l’aide d’un algorithme dérivé de l’algorithme de Needleman et Wunsch utilisé dans le domaine de la Bio-Informatique. Les deux approches sont testées sur des données artificielles ainsi que sur des données extraites d’un simulateur très réaliste d’un des systèmes du LHC et montrent de bons résultats. / In the context of the CERN Large Hadron Collider (LHC), a large number of control systems have been built based on industrial control and SCADA solutions. Beyond the complexity of these systems, a large number of sensors and actuators are controlled which make the monitoring and diagnostic of these equipment a continuous and real challenge for human operators. Even with the existing SCADA monitoring tools, critical situations prompt alarms avalanches in the supervision that makes diagnostic more difficult. This thesis proposes a decision support methodology based on the use of historical data. Past faults signatures represented by alarm lists are compared with the alarm list of the fault to diagnose using pattern matching methods. Two approaches are considered. In the first one, the order of appearance is not taken into account, the alarm lists are then represented by a binary vector and compared to each other thanks to an original weighted distance. Every alarm is weighted according to its ability to represent correctly every past faults. The second approach takes into account the alarms order and uses a symbolic sequence to represent the faults. The comparison between the sequences is then made by an adapted version of the Needleman and Wunsch algorithm widely used in Bio-Informatic. The two methods are tested on artificial data and on simulated data extracted from a very realistic simulator of one of the CERN system. Both methods show good results.

De nombreux systèmes de transport peuvent être modélisées par des synchronisations ordinaires (pour tout k>=l, l'occurrence k de l'événement B se produit au moins t unités de temps après l'occurrence k-l de l'événement A). Ces systèmes sont linéaires dans l'algèbre (max,+). Pour certaines applications, il est primordial de modéliser la simultanéité entre événements. Comme la synchronisation ordinaire ne suffit pas à exprimer ce phénomène, nous introduisons la synchronisation partielle (l'événement B ne peut se produire que quand l'événement A se produit). Dans ce mémoire, des méthodes développées pour la modélisation et le contrôle de systèmes linéaires dans l'algèbre (max,+) sont étendues à des systèmes régis par des synchronisations ordinaires et partielles. Nous considérons uniquement des systèmes divisés en un système principal et un système secondaire et gouvernés par des synchronisations ordinaires entre événements dans le même système et des synchronisations partielles d'événements dans le système secondaire par des événements dans le système principal. Nous introduisons une commande optimale et une commande prédictive pour cette classe de systèmes par analogie avec les résultats disponibles pour les systèmes linéaires dans l'algèbre (max,+). En considérant un comportement donné pour le système principal, il est aussi possible de représenter le système secondaire par une fonction de transfert et de modifier sa dynamique pour suivre un modèle de référence. / Many transportation networks can be modeled by (max,+)-linear systems, i.e., discrete event systems ruled by standard synchronizations (conditions of the form: "for all k>=l, occurrence k of event B is at least t units of time after occurrence k-l of event A"). In some applications, it is also necessary to model simultaneity between events (e.g., for a road equipped with traffic lights, a vehicle can cross an intersection only when the associated traffic light is green). Such conditions cannot be expressed using standard synchronizations. Hence, we introduce the partial synchronization (condition of the form: "event B can only occur when event A occurs"). In this thesis, we consider a class of discrete event systems ruled by standard and partial synchronizations, called (max,+)-systems with partial synchronization. Such systems are split into a main system and a secondary system such that there exist only standard synchronizations between events in the same system and partial synchronizations of events in the secondary system by events in the main system. We adapt some modeling and control approaches developed for (max,+)-linear systems to (max,+)-systems with partial synchronization. Optimal feedforward control and model predictive control for (max,+)-linear systems are extended to (max,+)-systems with partial synchronization. Furthermore, transfer relation and model reference control are provided for the secondary system under a predefined behavior of the main system.

Sistemas de controle de edifícios inteligentes têm evoluído consideravelmente e a sua confiabilidade também tem melhorado. Mesmo assim, falhas são inevitáveis em sistemas concebidos, construídos e operados pelo homem e uma abordagem alternativa envolve a minimização das conseqüências das falhas. Isto é possível através da reconfiguração do sistema que se faz realocando funções de controle e escolhendo outras formas de interação entre os processos. Apesar disso, na literatura ainda não existe muito material que trate da modelagem de processos em que desde a sua concepção as falhas são consideradas. Assim, propõe-se neste trabalho o desenvolvimento de um procedimento para a modelagem de sistemas de controle considerando além das especificações funcionais, em situação normal, a ocorrência de falhas. Em função da base teórica existente e ferramentas disponíveis a abordagem proposta adota o conceito de sistemas a eventos discretos, holon e a técnica de rede de Petri e suas extensões para descrever a estrutura e o funcionamento deste sistema de controle. Um exemplo de aplicação é aqui apresentado para ilustrar as principais características deste procedimento. / Control systems of intelligent buildings have evolved and its reliability has evidently improved. However, faults are inevitable in systems conceived, constructed and operated by humans and an alternative approach is the minimization of the faults consequences. This is possible through the reconfiguration of the system, i.e., re-allocating control functions and choosing other forms of interaction between the processes. Despite this, there are few publications that deal with the modeling of processes where since its conception the faults are considered. Thus, this work proposes the development of a procedure for the modeling of control systems considering beyond the functional specifications in normal situation, the occurrence of faults. Based on the existing theoretical base and available tools the proposal adopts the concept of discrete event system, holon and Petri net technique with its extensions to describe the structure and the operation of these control systems. An application example is here presented to illustrate the principal characteristics of this proceeding.

I detta arbete har studerats hur flödet i en flerstegs- bearbetningsprocess påverkas av stokastiska fluktuationer och störningar i de enskilda processtegen. Mera bestämt har analys utförts av hur de stokastiska variationerna i operationstiderna kan och bör modelleras vid simuleringsstudier. Även hur påverkan av valet av sådana stokastiska modeller kan tänkas ha på processen i sin helhet, till exempel avseende total genomloppstid. Examensarbetet syftar till att undersöka hur val av fördelning på parametern operationstid, påverkar resultatfaktorn genomloppstid vid flödessimuleringar. För att finna svar på denna påverkan har en fallstudie utförts, med utgångspunkt av indata från en produkt som tillverkas på Volvo Aero. Denna produkt genomgår en tillverkningssekvens innehållande 18 stycken bearbetningsoperationer innefattande tre olika processtyper (automatisk, halvautomatisk och manuell). Dessa tre processtyper är i olika grad beroende av operatörers insats. De 18 bearbetningsoperationernas processtid har analyserats numeriskt och grafiskt. Programvaran Stat:fit har använts som hjälpmedel för att erhålla svar på lämplig fördelning per tillverkningsoperation samt vilka teoretiska fördelningar som är lämpliga att använda för de tre olika processtyperna. De rekommenderade fördelningsteorierna per tillverkningsoperation har genomgått fördelningstest (Chi2, Kolmogorov-Smirnov och Anderson-Darling) och använts som grund vid skapande av försöksplan till simuleringsstudien. Simuleringsstudien har utförts enligt försöksplan i programvaran Simul8. Samtliga körningar från simuleringsmodellen är statistiskt säkerställda med 95 % konfidensintervall. Fallstudien har visat att resultatpåverkan från operationstidernas fördelningstyp är relativt liten vid simulering av komplexa system där faktorer som nivå av tillverkningsvolym och tillgänglighet har större påverkan på resultatfaktorn genomloppstid. Vid enklare modeller utan begränsning i form av reducerad tillgänglighet synliggörs skillnad i simuleringsresultat av olika val av fördelning på parametern operationstid. Fördelningen av dessa simuleringsresultat styrks av den centrala gränsvärdessatsen, det vill säga att om antalet observerade värden är tillräckligt stort, uppträder resultatet som normalfördelat. / Discrete event simulation is used to imitate and analyze how systems change over time. The actual behavior of the variation in the system is interpreted by using discrete and continuous probability distributions. In the software program Simul8, simulation models are created based on the information collected from the production. Shifts, operation time and efficiency are examples of information required for the modeling process. The aim with this bachelor´s thesis was to investigate how different choice of probability distributions on the parameter operation time affects the result of a discrete event simulation. The thesis is a result of a case study performed at Volvo Aero Corporation, Sweden. The case study involves investigation of probability distribution for 18 manufacturing operations for a product. The manufacturing sequence consists of three different types of processes (automatic, semiautomatic and manual). These three types of processes need different level of instrumentality. The commercial statistical computer software, Stat:fit has been used to find proper probability distribution for each of the manufacturing operations. The results from Stat:fit have been used to analyze if there are any connections between the process type and the probability distributions. The recommended probability distributions have been tested with Goodness-of-fit tests (Chi2, Kolmogorov-Smirnov and Anderson-Darling) using Stat:fit and used in the simulation modeling. The simulation model has been validated and verified by a simulation advisor at Volvo Aero. Five different simulation models have been evaluated in Simul8, with five different types of distributions. All simulation runs have been statistical proved, in Simul8 with 95% confidence interval. The result of this study indicates that the variation of process time has limited effect for complex simulation models containing low level of efficiency and high load factors, concerning the result of throughput time. For simple models, excluded from restricted efficiency, the effect on the throughput time is featured.

This thesis is aimed at studying the open pit truck- shovel haulage systems using computer simulation approach. The main goal of the study is to enhance the analysis and comparison of heuristic truck dispatching policies currently available and search for an adaptive rule applicable to open pit mines. For this purpose, a stochastic truck dispatching and production simulation program is developed for a medium size open pit mine consisting of several production faces and a single dump site using GPSS/H software. Eight basic rules are modeled in separate program files. The program considers all components of truck cycle and normal distribution is used to model all these variables. The program asks the user to enter the number of trucks initially assigned to each shovel site. Full-factorial simulation experiments are made to investigate the effects of several factors including the dispatching rules, the number of trucks operating, the number of shovels operating, the variability in truck loading, hauling and return times, the distance between shovels and dump site, and availability of shovel and truck resources. The breakdown of shovel and trucks are modeled using exponential distribution. Three performance measures are selected as truck production, overall shovel utilization and overall truck utilizations. Statistical analysis of the simulation experiments is done using ANOVA method with Minitab software. Regression analysis gives coefficient of determination values, R2, of 56.7 %, 84.1 %, and 79.6 % for the three performance measures, respectively. Also, Tukey&rsquo / s method of mean comparison test is carried out to compare the basic dispatching rules. From the results of statistical analysis, it is concluded that the effects of basic truck dispatching rules on the system performance are not significant. But, the main factors affecting the performances are the number of trucks, the number of shovels, the distance between the shovels and dump site, finally the availability of shovel and truck resources. Also, there are significant interaction effects between these main factors. Finally, an adaptive rule using the standardized utilization of shovels and trucks is developed.