Electromechanical Actuators (EMAs) are being increasingly used in many applications. There is a need to augment good design of EMAs with continuous awareness of their operational capability and make them ‘intelligent’ for two key objectives: enhancing performance to address exigent task requirements and to track any changes from their ‘as-built and certified’ state for condition-based maintenance. These objectives are achieved using a decision making philosophy where the human system operator supervises EMA operation using performance criteria and decision surfaces; updated by in-situ measurement of the variables of interest via a suite of diverse sensors. However, operational decisions made on the basis of faulty data could result in unwelcome consequences. With unexpected variations in a sensor’s output from its anticipated values, the challenge is to determine if it indicates a problem in the sensor or the monitored system. In addressing this conundrum, it is also essential to account for the inherent uncertainties present in the values being analyzed. To this end, this dissertation presents the development of a novel Sensor and Process Fault Detection and Isolation (SPFDI) algorithm. This provides a framework to utilize data from all the available sensors in a holistic manner to detect any faults in individual sensors or the system components concurrently. The algorithm uses a Bayesian network to model a system; populated with extensive empirical data. The probabilistic foundations of this method allow for incorporating and propagating uncertainties. The construction of a modular testbed and its Bayesian network are discussed in detail. Several design/ operational criteria have been proposed to aid in the creation of more usable networks in the future. The SPFDI algorithm estimates multiple values for each measurand using different combinations of input variables and probabilistic inferencing. These values are compared against those indicated by the corresponding sensors; a difference between them is indicative of a potential problem. Quantitative indicators to track the condition of different system components and sensors, termed as belief values, are modified after each comparison. The final belief values obtained at the end of an iteration of the algorithm provide a definitive indication of the sources of anomalies in the observed data and can provide guidance to the operator on decisions such as whether or not to use data from a particular sensor for updating existing decision surfaces. The representative examples and experimental results confirm the efficacy of the algorithm in detecting and isolating single as well as multiple sensor faults. The algorithm has also been found to be capable of distinguishing between sensor and system/process faults. Special categories of faults and factors that influence the execution characteristics and quality of results from the algorithm were also explored meticulously and suitable modifications have been suggested to enable the algorithm to continue to function effectively in these situations. To demonstrate the flexibility of the proposed SPFDI algorithm, its potential utilization in four broad classes of applications consisting of complex systems monitored by multiple sensors was also explored in this report. / text

Les activités industrielles peuvent induire des risques de nature diverse, voire conduire à des événements catastrophiques. De façon à réduire ces risques à des niveaux acceptables, à en limiter les coûts aussi bien humains, matériels que financiers et à respecter la réglementation, il est crucial d’en prévenir la survenue (prévention) et à identifier des situations anormales en fonction d’informations récupérées sur le terrain. Ceci nécessite de résoudre deux points essentiels : détecter l’anomalie et faire un diagnostic, d’où l’importance des modèles de bon comportement qui modélisent des systèmes physique réels et les méthodes associées permettant d’effectuer la surveillance en continu. L’approche la plus utilisée dans la communauté FDI est l’approche structurelle qui consiste à construire des relations de redondance analytique (RRA). Le principe de l’approche structurelle est de prendre en compte seulement l’existence des relations entre les variables induites par les équations du modèle. En basant sur cette approche, ma thèse consiste à trouver une réponse qui satisfait un triple objectif :1. Extraire les relations valides à partir du modèle de bon comportement d'un système afin de prendre en compte l'évolution du système en éliminant les relations et des mesures invalides ayant pour but d'effectuer le diagnostic en ligne2. Construire, en utilisant une analyse symbolique couplée avec la théorie des graphes, les relations de redondance analytique symboliques (RRAS) pour la détection des défauts dans le système.3. Evaluer ces RRAS en utilisant le calcul par intervalle, an de prendre en compte les incertitudes présents dans les mesures, lors des tests de cohérence. / Large-scale complex process plants are safety-critical systems where the real-time diagnosis is very important. In a model based systems engineering approach, the structured development process from the concept to the production to the operation phase is organized around a coherent model of the system. This model contains, in particular, relations about the behavior of the system that could have been used for simulation in the design phase. The objective of this work is to use this information to design automatically on-line diagnosis algorithms using the hybrid dynamical information part and sensor measurements of the system model. In this thesis, the proposed approach allows:To extract the valid relations of system behavior to take into account system evolution by eliminating invalid constraints and measurements for establishing an on-line diagnosisTo build, using symbolic analysis and graph path search, analytical redundancy relations for the various system configurationsTo evaluate these ARRs in using set valued computations (interval arithmetic) to take into account model and measurements uncertainties

The concept of lights-out process control is explored in this work (specifically pertaining to the minerals processing industry). The term is derived from lights-out manufacturing, which is used in discrete component manufacturing to describe a fully automated production line, i.e. with no human intervention. Lights-out process control is therefore defined as the fully autonomous operation of a processing plant (as achieved through automatic process control), without operator interaction. / Thesis (PhD)--University of Pretoria, 2017. / National Research Foundation (NRF) / Electrical, Electronic and Computer Engineering / PhD / Unrestricted

<p>This thesis is a result of the increased requirements on availability and costs of the aircraft Jas 39 Gripen. The work has been to specify demands and to find methods suitable for development of a decision support system for the fault isolation of the aircraft. The work has also been to implement the chosen method. Two different methods are presented and a detailed comparison is performed with the demands as a starting point. The chosen method handle multiple faults in O(N2)-time where N is the number of components. The implementation shows how all demands are fulfilled and how new tests can be added during execution. Since the thesis covers the development of a prototype no practical evaluation with compare of manually isolation is done.</p>

This thesis is a result of the increased requirements on availability and costs of the aircraft Jas 39 Gripen. The work has been to specify demands and to find methods suitable for development of a decision support system for the fault isolation of the aircraft. The work has also been to implement the chosen method. Two different methods are presented and a detailed comparison is performed with the demands as a starting point. The chosen method handle multiple faults in O(N2)-time where N is the number of components. The implementation shows how all demands are fulfilled and how new tests can be added during execution. Since the thesis covers the development of a prototype no practical evaluation with compare of manually isolation is done.

O problema da tolerância a falhas em robôs manipuladores cooperativos conectados rigidamente a um objeto indeformável é estudado nesta tese. A tolerância a falhas é alcançada através de reconfiguração do sistema de controle. Primeiro, a falha é detectada e isolada. Então, o sistema de controle é reconfigurado de acordo com a falha isolada. As falhas em robôs manipuladores são primeiramente estudadas de acordo com suas consequências no sistema cooperativo. Quatro tipos de falhas são identificados: juntas com balanço livre (sem atuadores ativos), bloqueadas, com informação incorreta de posição e com informação incorreta de velocidade. A detecção e a isolação dos dois primeiros tipos de falhas são alcançadas através de um sistema utilizando redes neurais artificiais. Redes do tipo MLP são empregadas para mapear a dinâmica dos robôs cooperativos sem falhas e uma rede RBF é utilizada para a classificação do vetor de resíduos. As falhas do tipo informação incorreta de posição ou velocidade das juntas são detectadas e isoladas através do uso das restrições impostas pela cadeia cinemática fechada presente no sistema cooperativo. Quando falhas do tipo juntas com balanço livre ou bloqueadas são isoladas, as leis de controle são reconfiguradas. Para estes casos, controladores híbridos de movimento e esmagamento do objeto são deduzidos. Quando falhas do tipo informação incorreta de posição ou velocidade das juntas são isoladas, as medidas afetadas são substituídas por valores estimados. Resultados obtidos em simulações e em robôs cooperativos reais mostram que a metodologia proposta é viável. / The problem of fault tolerance in cooperative manipulators rigidly connected to an undeformable load is addressed in this work. Fault tolerance is reached by reconfiguration of the control system. The faults are firstly detected and isolated. Then, the control system is reconfigured according to the isolated fault. Four faults are considered: free-swinging joint faults, locked joint faults, incorrectly measured joint position faults, and incorrectly measured joint velocity faults. Free-swinging and locked joint faults are detected and isolated by artificial neural networks. MLP’s are utilized to reproduce the dynamics of the fault-free system and an RBF is used to classify the residual vector. Incorrectly measured joint position and velocity faults are detected and isolated based on the kinematic constraints imposed on the cooperative system. When free-swinging and locked joint faults are isolated, the control laws are reconfigured. Control laws for motion and squeeze of the object are developed in these cases. When incorrectly measured joint position faults and incorrectly measured joint velocity faults are isolated, the faulty measurements are replaced by their estimates. Results obtained in simulations and in real cooperative robots indicate that the proposed methodology is viable.

La thèse est dédiée à l'analyse et à la conception de la commande tolérante aux défauts (fault tolerant control - FTC) en se fondant sur des méthodes ensemblistes. Nous étudions l'apparition des défauts pour les systèmes multi-capteurs, et les modes de détection, ainsi que la conception de lois de commande qui assurent la stabilité en boucle fermée. L'utilisation des ensembles invariants/contractifs permet la caractérisation des signaux résiduels, qui sont utilisés par la suite dans le processus de détection et d'isolement des défauts. La décision est fondée sur le positionnement par rapport à des hyperplans de séparation avec des importantes réductions de temps de calcul. Un mécanisme dual mis en œuvre par un bloc de récupération, permet la certification de la récupération des capteurs précédemment affectés par ces défauts.Dans une perspective théorique, nous soulignons les conditions qui permettent l'inclusion du bloc FDI (fault detection and isolation) et sa raison d'être dans la conception des lois de commande. Cela conduit par exemple à la synthèse des gains de retour d'état statique, par résolution de problèmes d'optimisation efficace (linéaire/convexe).Selon les paramètres choisis pour le réglage, la conception de la FTC peut être complétée par un superviseur de référence ou d'une loi de commande prédictive, qui adapte la trajectoire d'état et l'action de commande par retour d'état, afin d'assurer l'identification et la détection des défauts. Les questions spécifiques à l'utilisation de méthodes ensemblistes sont détaillées et des améliorations diverses sont proposées, par exemple : la construction des ensembles invariants, des formulations moins complexes des problèmes de type Mixed Integer Programming (MIP), l'analyse de la stabilité des systèmes commutés (notion de ``dwell-time''). / The scope of the thesis is the analysis and design of fault tolerant control (FTC) schemes through the use of set-theoretic methods. In the framework of multisensor schemes, the faults appearance and the modalities to accurately detect them are investigated as well as the design of control laws which assure the closed-loop stability. By using invariant/contractive sets to describe the residual signals, a fault detection and isolation (FDI) mechanism with reduced computational demands is implemented based on set-separation. A dual mechanism, implemented by a recovery block, which certificates previously fault-affected sensors is also studied. From a broader theoretical perspective, we point to the conditions which allow the inclusion of {FDI} objectives in the control law design. This leads to static feedback gains synthesis by means of numerically attractive optimization problems. Depending on the parameters selected for tuning, is shown that the FTC design can be completed by a reference governor or a predictive control scheme which adapts the state trajectory and the feedback control action in order to assure {FDI}. When necessary, the specific issues originated by the use of set-theoretic methods are detailed and various improvements are proposed towards: invariant set construction, mixed integer programming (MIP), stability for switched systems (dwell-time notions).

O problema da tolerância a falhas em robôs manipuladores cooperativos conectados rigidamente a um objeto indeformável é estudado nesta tese. A tolerância a falhas é alcançada através de reconfiguração do sistema de controle. Primeiro, a falha é detectada e isolada. Então, o sistema de controle é reconfigurado de acordo com a falha isolada. As falhas em robôs manipuladores são primeiramente estudadas de acordo com suas consequências no sistema cooperativo. Quatro tipos de falhas são identificados: juntas com balanço livre (sem atuadores ativos), bloqueadas, com informação incorreta de posição e com informação incorreta de velocidade. A detecção e a isolação dos dois primeiros tipos de falhas são alcançadas através de um sistema utilizando redes neurais artificiais. Redes do tipo MLP são empregadas para mapear a dinâmica dos robôs cooperativos sem falhas e uma rede RBF é utilizada para a classificação do vetor de resíduos. As falhas do tipo informação incorreta de posição ou velocidade das juntas são detectadas e isoladas através do uso das restrições impostas pela cadeia cinemática fechada presente no sistema cooperativo. Quando falhas do tipo juntas com balanço livre ou bloqueadas são isoladas, as leis de controle são reconfiguradas. Para estes casos, controladores híbridos de movimento e esmagamento do objeto são deduzidos. Quando falhas do tipo informação incorreta de posição ou velocidade das juntas são isoladas, as medidas afetadas são substituídas por valores estimados. Resultados obtidos em simulações e em robôs cooperativos reais mostram que a metodologia proposta é viável. / The problem of fault tolerance in cooperative manipulators rigidly connected to an undeformable load is addressed in this work. Fault tolerance is reached by reconfiguration of the control system. The faults are firstly detected and isolated. Then, the control system is reconfigured according to the isolated fault. Four faults are considered: free-swinging joint faults, locked joint faults, incorrectly measured joint position faults, and incorrectly measured joint velocity faults. Free-swinging and locked joint faults are detected and isolated by artificial neural networks. MLP’s are utilized to reproduce the dynamics of the fault-free system and an RBF is used to classify the residual vector. Incorrectly measured joint position and velocity faults are detected and isolated based on the kinematic constraints imposed on the cooperative system. When free-swinging and locked joint faults are isolated, the control laws are reconfigured. Control laws for motion and squeeze of the object are developed in these cases. When incorrectly measured joint position faults and incorrectly measured joint velocity faults are isolated, the faulty measurements are replaced by their estimates. Results obtained in simulations and in real cooperative robots indicate that the proposed methodology is viable.

This dissertation focused on the IRIS design since this will likely be one of the designs of choice for future deployment in the U.S and developing countries. With a net 335 MWe output IRIS novel design falls in the “medium” size category and it is a potential candidate for the so called modular reactors, which may be appropriate for base load electricity generation, especially in regions with smaller electricity grids, but especially well suited for more specialized non-electrical energy applications such as district heating and process steam for desalination. The first objective of this dissertation is to evaluate and quantify the performance of a Nuclear Power Plant (NPP) comprised of two IRIS reactor modules operating simultaneously with a common steam header, which in turn is connected to a single turbine, resulting in a steam-mixing control problem with respect to “load-following” scenarios, such as varying load during the day or reduced consumption during the weekend. To solve this problem a single-module IRIS SIMULINK model previously developed by another researcher is modified to include a second module and was used to quantify the responses from both modules. In order to develop research related to instrumentation and control, and equipment and sensor monitoring, the second objective is to build a two-tank multivariate loop in the Nuclear Engineering Department at the University of Tennessee. This loop provides the framework necessary to investigate and test control strategies and fault detection in sensors, equipment and actuators. The third objective is to experimentally develop and demonstrate a fault-tolerant control strategy using this loop. Using six correlated variables in a single-tank configuration, five inferential models and one Auto-Associative Kernel Regression (AAKR) model were developed to detect faults in process sensors. Once detected the faulty measurements were successfully substituted with prediction values, which would provide the necessary flexibility and time to find the source of discrepancy and resolve it, such as in an operating power plant. Finally, using the same empirical models, an actuator failure was simulated and once detected the control was automatically transferred and reconfigured from one tank to another, providing survivability to the system.

Ce travail se compose de deux parties, dans la première, deux types de méthodologies sont proposées pour garantir l'observabilité sur des systèmes non uniformément observables. Premièrement sont présentées les méthodes basées sur le grammien d'observabilité et, à continuation, les méthodes basées directement sur l'équation de l'observateur. Dans la deuxième partie, diverses techniques sont détaillées pour la détection de défauts (fuites et obstructions) dans les canalisations sous pressions. Pour cela on construit plusieurs modèles en discrétisant les équations du coup de bélier par différences finies, implicites et explicites dans le temps. Sur ces modèles des techniques sont développés en utilisant des observateurs et des algorithmes d'optimisation. Les modèles discrets ainsi que certains observateurs ont été validés par une série d'expériences effectuées dans des canalisations d'essai. Des résultats de convergence, expérimentaux et en simulation sont exposés dans ce mémoire. / This work consists of two parts, in the first one, two types of methods are proposed to ensure the observability of non-uniformly observable systems. Firstly methods based on the observability gramian are presented, and then some methods based directly on the equation of the observer. In the second part, various techniques are detailed for the detection of defaults (leaks and obstructions) in a pipeline under pressure. For that, we built several models by discretizing the water hammer equations using finite differences explicit and implicit in time. Then some techniques are developed using observers and optimization algorithms. Discrete models and some observers were validated by a series of experiments in pipelines. Convergence, experimental and simulation results are presented in this manuscript.