Cette étude est le résultat d’une thèse CIFRE de trois ans entre le bureau d’étude d’Airbus (domaine du contrôle de l’avion) et le laboratoire TéSA à Toulouse. L’objectif principal est de proposer, développer et valider une solution logicielle pour la détection et le diagnostic d’un type spécifique de vibrations des gouvernes de profondeur et direction, appelée oscillation en cycle limite (limit cycle oscillation ou LCO en anglais), basée sur les signaux existants dans les avions civils. LCO est un terme mathématique générique définissant un mode périodique indépendant de conditions initiales et se produisant dans des systèmes non linéaires non conservatifs. Dans cette étude, nous nous intéressons au phénomène de LCO induit par les jeux mécaniques dans les gouvernes d’un avion civil. Les conséquences du LCO sont l’augmentation locale de la charge structurelle, la dégradation des qualités de vol, la réduction de la durée de vie de l’actionneur, la dégradation du confort du poste de pilotage et de la cabine, ainsi que l’augmentation des coûts de maintenance. L’état de l’art en matière de détection et de diagnostic du LCO induit par le jeu mécanique est basé sur la sensibilité du pilote aux vibrations et sur le contrôle périodique du jeu sur les gouvernes. Cette étude propose une solution basée sur les données (issues de la boucle d’asservissement des actionneurs qui agissent sur les gouvernes) pour aider au diagnostic du LCO et à l’isolement du jeu mécanique. L’objectif est d’améliorer encore plus la disponibilité des avions et de réduire les coûts de maintenance en fournissant aux compagnies aériennes un signal de contrôle pour le LCO et les jeux mécaniques. Pour cette raison, deux solutions algorithmiques pour le diagnostic des vibrations et des jeux ont été proposées. Un détecteur en temps réel pour la détection du LCO est tout d’abord proposé basé sur la théorie du rapport de vraisemblance généralisé (generalized likelihood ratio test ou GLRT en anglais). Certaines variantes et simplifications sont également proposées pour satisfaire les contraintes industrielles. Un détecteur de jeu mécanique est introduit basé sur l’identification d’un modèle de Wiener. Des approches paramétrique (estimateur de maximum de vraisemblance) et non paramétrique (régression par noyau) sont explorées, ainsi que certaines variantes des méthodes non paramétriques. En particulier, le problème de l’estimation d’un cycle d’hystérésis (choisi comme la non-linéarité de sortie d’un modèle de Wiener) est abordé. Ainsi, les problèmes avec et sans contraintes sont étudiés. Une analyse théorique, numérique (sur simulateur) et expérimentale (données de vol et laboratoire) est réalisée pour étudier les performances des détecteurs proposés et pour identifier les limitations et la faisabilité industrielle. Les résultats numériques et expérimentaux obtenus confirment que le GLRT proposé (et ses variantes / simplifications) est une méthode très efficace pour le diagnostic du LCO en termes de performance, robustesse et coût calculatoire. D’autre part, l’algorithme de diagnostic des jeux mécaniques est capable de détecter des niveaux de jeu relativement importants, mais il ne fournit pas de résultats cohérents pour des niveaux de jeu relativement faibles. En outre, des types d’entrée spécifiques sont nécessaires pour garantir des résultats répétitifs et cohérents. Des études complémentaires pourraient être menées afin de comparer les résultats de GLRT avec une approche Bayésienne et pour approfondir les possibilités et les limites de la méthode paramétrique proposée pour l’identification du modèle de Wiener. / This research study is the result of a 3 years CIFRE PhD thesis between the Airbus design office(Aircraft Control domain) and TéSA laboratory in Toulouse. The main goal is to propose, developand validate a software solution for the detection and diagnosis of a specific type of elevator andrudder vibration, called limit cycle oscillation (LCO), based on existing signals available in flightcontrol computers on board in-series aircraft. LCO is a generic mathematical term defining aninitial condition-independent periodic mode occurring in nonconservative nonlinear systems. Thisstudy focuses on the LCO phenomenon induced by mechanical freeplays in the control surface ofa civil aircraft. The LCO consequences are local structural load augmentation, flight handlingqualities deterioration, actuator operational life reduction, cockpit and cabin comfort deteriorationand maintenance cost augmentation. The state-of-the-art for freeplay induced LCO detection anddiagnosis is based on the pilot sensitivity to vibration and to periodic freeplay check on the controlsurfaces. This study is thought to propose a data-driven solution to help LCO and freeplaydiagnosis. The goal is to improve even more aircraft availability and reduce the maintenance costsby providing to the airlines a condition monitoring signal for LCO and freeplays. For this reason,two algorithmic solutions for vibration and freeplay diagnosis are investigated in this PhD thesis. Areal time detector for LCO diagnosis is first proposed based on the theory of the generalized likeli hood ratio test (GLRT). Some variants and simplifications are also proposed to be compliantwith the industrial constraints. In a second part of this work, a mechanical freeplay detector isintroduced based on the theory of Wiener model identification. Parametric (maximum likelihoodestimator) and non parametric (kernel regression) approaches are investigated, as well as somevariants to well-known nonparametric methods. In particular, the problem of hysteresis cycleestimation (as the output nonlinearity of a Wiener model) is tackled. Moreover, the constrainedand unconstrained problems are studied. A theoretical, numerical (simulator) and experimental(flight data and laboratory) analysis is carried out to investigate the performance of the proposeddetectors and to identify limitations and industrial feasibility. The obtained numerical andexperimental results confirm that the proposed GLR test (and its variants/simplifications) is a very appealing method for LCO diagnostic in terms of performance, robustness and computationalcost. On the other hand, the proposed freeplay diagnostic algorithm is able to detect relativelylarge freeplay levels, but it does not provide consistent results for relatively small freeplay levels. Moreover, specific input types are needed to guarantee repetitive and consistent results. Further studies should be carried out in order to compare the GLRT results with a Bayesian approach and to investigate more deeply the possibilities and limitations of the proposed parametric method for Wiener model identification.

Magnetgärdets transformatorstation är belägen i Ludvika och ägs av Västerbergslagens Energi AB. Transformatorstationen är nyligen ombyggd och nu mer placerad inomhus, vilket gör att påkänningar från väder och vind minskar. För att minimera behovet av personlig tillsyn och öka tillgängligheten behövs dock någon form av tillståndsövervakning. Syftet med examensarbetet har varit att undersöka om det med hjälp av värmekameror är möjligt att tillståndsövervaka de kritiska punkterna i transformatorstationen. För att fastställa detta har teori kring tillståndsövervakning, termografering och värmekameror sammanställts tillsammans med utförda mätningar och tester. Resultaten har sedan diskuterats för att se om ett möjligt lösningsförslag kan presenteras. Viktiga kriterier för att möjligöra tillståndsövervakning med hjälp av värmekameror har tagits fram. Slutsatsen är att det är möjligt, men kräver en komplicerad systemlösning. Lösningsförslaget bygger på flera samspelande delar och bör bli ett kraftfullt övervakningssystem. Huruvida värmekameror är det bästa och enda verktyget är dock tveksamt. Det anses därför väsentligt att även undersöka andra övervakningsmöligheter, som exempel termistorgivare. / The substation called Magnetgärdet is located in Ludvika, Sweden and it is owned by Västerbergslagens Energi AB. The substation has recently been renovated and is now placed indoors, which reduces the influence of weather on the station. To decrease the need for personal supervision and increase the availability of the station a monitoring solution is needed. The purpose of this degree thesis is to examine if it is possible to monitor the condition of critical items in the station with infrared cameras. To determine this, theory on condition monitoring, thermography and thermal imaging cameras has been compiled together with relevant measurements and tests. The results are then discussed to see if a possible solution can be presented. Important criteria for making condition monitoring with infrared cameras possible have also been formulated in this degree thesis. The conclusion is that it is possible, but a complicated system is required. The solution we have given is based on different parts interplaying with each other and it should become a powerful monitoring system. It is however uncertain whether infrared cameras are the only and best tools, therefore other tools should be considered. It might be possible to use thermistors with, or instead of infrared cameras.

Brushless Direct Current (BLDC) motors are one of the motor types rapidly gaining popularity. BLDC motors are being increasingly used in critical high performance industries such as appliances, automotive, aerospace, consumer, medical, industrial automation equipment and instrumentation. Fault detection and condition monitoring of BLDC machines is therefore assuming a new importance. The objective of this research is to advance the field of rotor and load fault diagnosis in BLDC machines operating in a variety of operating conditions ranging from constant speed to continuous transient operation. This objective is addressed as three parts in this research. The first part experimentally characterizes the effects of rotor faults in the stator current and voltage of the BLDC motor. This helps in better understanding the behavior of rotor defects in BLDC motors. The second part develops methods to detect faults in loads coupled to BLDC motors by monitoring the stator current. As most BLDC applications involve non-stationary operating conditions, the diagnosis of rotor faults in non-stationary conditions forms the third and most important part of this research. Several signal processing techniques are reviewed to analyze non-stationary signals. Three new algorithms are proposed that can track and detect rotor faults in non-stationary or transient current signals.

Stator current spectral analysis techniques are usually used to detect rotor faults in induction machines. Magnetic field anomalies in the airgap due to the rotor faults result in characteristic side-band harmonic components in the stator current spectrum, which can be measured as rotor fault signatures. A position-varying load torque oscillation at multiples of the rotational speed, however, has exactly the same effect. Stator current harmonics due to a load torque oscillation often obscure and even overwhelm rotor eccentricity fault detection since the magnitude of load oscillation induced harmonics is usually much larger. Although previous research has suggested some methods to differentiate between these two effects, most of them rely heavily on the accurate estimation of motor parameters. The objective of this research is to develop a far more practical and computationally efficient method to detect rotor faults effectively in the presence of a load torque oscillation. A significant advantage of the proposed scheme is that it does not need any knowledge of motor parameters. The normalized negative sequence information induced by a mixed rotor eccentricity in the stator current or terminal voltage space vector spectra, serves as a reliable rotor fault indicator to eliminate load oscillation effects. Detailed airgap magnetic field analysis for an eccentric motor is performed and all machine inductance matrices as well as their derivatives are reformulated accordingly. Careful observation of these inductance matrices provides a fundamental understanding of motor operation characteristics under a fault condition. Simulation results based on both induction motor dynamic model and Maxwell 2D Finite Element Model demonstrate clearly the existence of the predicted rotor fault indicator. Extensive experimental results also validate the effectiveness and feasibility of the proposed detection scheme.

The objective of this research is to develop a bearing fault detection scheme for electric machines via stator current. A new method, called the stator current noise cancellation method, is proposed to separate bearing fault-related components in the stator current. This method is based on the concept of viewing all bearing-unrelated components as noise and defining the bearing detection problem as a low signal-to-noise ratio (SNR) problem. In this method, a noise cancellation algorithm based on Wiener filtering is employed to solve the problem. Furthermore, a statistical method is proposed to process the data of noise-cancelled stator current, which enables bearing conditions to be evaluated solely based on stator current measurements. A detailed theoretical analysis of the proposed methods is presented. Several online tests are also performed in this research to validate the proposed methods. It is shown in this work that a bearing fault can be detected by measuring the variation of the RMS of noise-cancelled stator current by using statistical methods such as the Statistical Process Control. In contrast to most existing current monitoring techniques, the detection methods proposed in this research are designed to detect generalized-roughness bearing faults. In addition, the information about machine parameters and bearing dimensions are not required in the implementation.

This dissertation proposes several simple, robust, and non-intrusive condition monitoring methods for induction motors fed by closed-loop inverters and claw-pole generators with built-in rectifiers. While the flexible energy forms synthesized by power electronic converters greatly enhance the performance and expand the operating region of induction motors and claw-pole generators, they also significantly alter the fault behavior of these electric machines and complicate the fault detection and protection. In this dissertation, special characteristics of the connected closed-loop inverter and rectifier have been thoroughly analyzed, with particular interest in their impact on fault behaviors of the induction motor and the claw-pole generator. Based on the findings obtained from the theoretical and experimental analysis, several sensorless thermal, mechanical, and insulation monitoring methods are proposed by smartly utilizing special features and capabilities of the connected power electronic converter. A simple and sensitive stator turn-fault detector is proposed for induction motors fed by closed-loop inverter. In addition, a stator thermal monitoring method based on active DC current injection and direct voltage estimation is also proposed to prevent the closed-loop controlled induction motors from thermally overloading. The performance of both methods is demonstrated by extensive experimental results. Methods to detect serpentine belt slip, serpentine belt defect, rotor eccentricity have been proposed for claw-pole generators using only the available electric sensor information. Methods to detect and protect stator turn faults in claw-pole generators are also presented in this dissertation. Lastly, a novel method to detect the generalized bearing roughness fault is proposed. All the proposed condition monitoring techniques have been validated by experimental results.

Detecting abnormal operating conditions, which will lead to faults developing later, has important economic implications for industries trying to meet their performance and production goals. It is unacceptable to wait for failures that have potential safety, environmental and financial consequences. Moving from a “reactive” strategy to a “proactive” strategy can improve critical equipment reliability and availability while constraining maintenance costs, reducing production deferrals, decreasing the need for spare parts. Once the fault initiates, predicting its progression and deterioration can enable timely interventions without risk to personnel safety or to equipment integrity. This work presents an online-integrated condition monitoring and prognostics framework that addresses the above issues holistically. The proposed framework aligns fully with ISO 17359:2011 and derives from the I-P and P-F curve. Depending upon the running state of machine with respect to its I-P and P-F curve an algorithm will do one of the following: (1) Predict the ideal behaviour and any departure from the normal operating envelope using a combination of Evolving Clustering Method (ECM), a normalised fuzzy weighted distance and tracking signal method. (2) Identify the cause of the departure through an automated diagnostics system using a modified version of ECM for classification. (3) Predict the short-term progression of fault using a modified version of the Dynamic Evolving Neuro-Fuzzy Inference System (DENFIS), called here MDENFIS and a tracking signal method. (4) Predict the long term progression of fault (Prognostics) using a combination of Autoregressive Integrated Moving Average (ARIMA)- Empirical Mode Decomposition (EMD) for predicting the future input values and MDENFIS for predicting the long term progression of fault (output). The proposed model was tested and compared against other models in the literature using benchmarks and field data. This work demonstrates four noticeable improvements over previous methods: (1) Enhanced testing prediction accuracy, (2) comparable processing time if not better, (3) the ability to detect sudden changes in the process and finally (4) the ability to identify and isolate the problem source with high accuracy.

Condition monitoring of critical machinery has many economic benefits. The primary objective is to detect faults, for example on rolling element bearings, at an early stage to take corrective action prior to the catastrophic failure of a component. In this context, it is important to be able to discriminate between stable and deteriorating fault conditions. A number of conventional vibration analysis techniques exist by which certain faults in rotating machinery may be identified. However, under circumstances involving multiple fault conditions conventional condition monitoring techniques may fail, e.g. by indicating deteriorating fault conditions for stable fault situations or vice versa. Condition monitoring of rotating machinery that may have multiple, possibly simultaneous, fault conditions is investigated in this thesis. Different combinations of interacting fault conditions are studied both through experimental methods and simulated models. Novel signal processing techniques (such as cepstral analysis and equidistant Fourier transforms) and pattern recognition techniques (based on the nearest neighbour algorithm) are applied to vibration problems of this nature. A set of signal processing and pattern recognition techniques is developed for the detection of small incipient mechanical faults in the presence of noise and dynamic load (imbalance). In the case investigated the dynamic loading consisted of varying degrees of imbalance. It is demonstrated that the proposed techniques may be applied successfully to the detection of multiple fault conditions. / Thesis (Ph.D. (Electronical Engineering))--North-West University, Potchefstroom Campus, 2004.

Condition monitoring of critical machinery has many economic benefits. The primary objective is to detect faults, for example on rolling element bearings, at an early stage to take corrective action prior to the catastrophic failure of a component. In this context, it is important to be able to discriminate between stable and deteriorating fault conditions. A number of conventional vibration analysis techniques exist by which certain faults in rotating machinery may be identified. However, under circumstances involving multiple fault conditions conventional condition monitoring techniques may fail, e.g. by indicating deteriorating fault conditions for stable fault situations or vice versa. Condition monitoring of rotating machinery that may have multiple, possibly simultaneous, fault conditions is investigated in this thesis. Different combinations of interacting fault conditions are studied both through experimental methods and simulated models. Novel signal processing techniques (such as cepstral analysis and equidistant Fourier transforms) and pattern recognition techniques (based on the nearest neighbour algorithm) are applied to vibration problems of this nature. A set of signal processing and pattern recognition techniques is developed for the detection of small incipient mechanical faults in the presence of noise and dynamic load (imbalance). In the case investigated the dynamic loading consisted of varying degrees of imbalance. It is demonstrated that the proposed techniques may be applied successfully to the detection of multiple fault conditions. / Thesis (Ph.D. (Electronical Engineering))--North-West University, Potchefstroom Campus, 2004.

Monitoring and controlling smart grid assets in a timely and reliable manner is highly desired for emerging smart grid applications. Wireless Sensor Networks (WSNs) are anticipated to be widely utilized in a broad range of smart grid applications due to their numerous advantages along with their successful adoption in various critical areas including military and health care. Despite these advantages, the use of WSNs in such critical applications has brought forward a new challenge of ful lling the Quality of Service (QoS) requirements of these applications. Providing QoS support is a challenging issue due to highly resource constrained nature of sensor nodes, unreliable wireless links and harsh operation environments. In this thesis we critically investigate the problem of QoS provisioning in WSNs. We identify challenges, limitations and requirements for applying QoS provisioning for WSNs in smart grid applications. We nd that the topic of data prioritization techniques at the MAC layer to provide delay bounds in condition monitoring applications is not well developed. We develop six novel QoS schemes that provide data di erentiation and reduce the latency of high priority tra c in a smart grid context. These schemes are namely; Delay-Responsive Cross layer (DRX), Fair and Delay-aware Cross layer (FDRX), Delay-Responsive Cross layer with Linear backo (LDRX), Adaptive Realistic and Stable Model (ARSM), Adaptive Inter-cluster head Delay Control (AIDC) and QoS-aware GTS Allocation (QGA). Furthermore, we propose a new Markov-based model for IEEE 802.15.4 MAC namely, Realistic and Stable Markovbased (RSM). RSM considers actual network conditions and enhances the stability of the WSNs. We show through analytical and simulation results that all of the presented schemes reduce the end-to-end delay while maintaining good energy consumption and data delivery values.