The industrial world is in constant demand for faster, cheaper and higher quality manufacturing. Robot utilisation and automation has evolved to become a necessary asset to master in order to stay competitive in the global market. With the growing dependency on robots, unexpected downtime and brakedowns can cause devastating loss of revenue. Consequently, this has lead to an increased importance for an accurate condition based way of performing robotic maintenance. As of writing, robots are predominantly maintained through time dependent maintenance. Part replacement is based on statistical models where maintenance is performed without taking the actual robot condition into consideration. As a result an overall level of uncertainty is ensued, where lacking the ability to properly diagnose the robot, also leads to superfluous repairs. Because of the costly impact this has on production, a condition based maintenance approach to robots would yield increased reliability at a lower cost of maintenance. This research focuses on trying to monitor vibrations in a robot, so as to infer about wear and to provide a first step in vibration based Robot Condition Monitoring. This research has been of multidisciplinary nature where robotics, tribology, mechanical component, signal analysis and diagnosis theory have overlapped in several areas throughout the project. The research has provided a vibration baseline and trends of the theoretical bearing defect frequencies for a hypocycloid gearbox installed on an ABB IRB6600 robot. The gearbox was not worn to a level that a severe gearbox degradation was irrefutably detectable and analysable. Accelerometers normally used on wind turbines were used for the project, and are believed to be sufficiently successful in capturing bearing related signals to accredit it for continued use at the preliminary stages of Robot Condition Monitoring development. A worn RV410F hypocycloid gearbox, was dismantled and analysed. Bearings found inside indicate high degrees of moisture corrosion and extensive surface wear. These findings had decisive roles in what future work recommendations where presented. Areas with great potential are condition monitoring through the use of Acoustic Emission and lubrication analysis. Further recommendations include investigating signal analysis techniques such as cepstrum pre-whitening and discrete wavelet transforms.

A novelty detection algorithm inspired by human audio pattern recognition is conceptualized and experimentally tested. This anomaly detection technique can be used to monitor the health of a machine or could also be coupled with a current state of the art system to enhance its fault detection capabilities. Time-domain data obtained from a microphone is processed by applying a short-time FFT, which returns time-frequency patterns. Such patterns are fed to a machine learning algorithm, which is designed to detect novel signals and identify windows in the frequency domain where such novelties occur. The algorithm presented in this paper uses one-dimensional kernel density estimation for different frequency bins. This process eliminates the need for data dimension reduction algorithms. The method of "pseudo-likelihood cross validation" is used to find an independent optimal kernel bandwidth for each frequency bin. Metrics such as the "Individual Node Relative Difference" and "Total Novelty Score" are presented in this work, and used to assess the degree of novelty of a new signal. Experimental datasets containing synthetic and real novelties are used to illustrate and test the novelty detection algorithm. Novelties are successfully detected in all experiments. The presented novelty detection technique could greatly enhance the performance of current state-of-the art condition monitoring systems, or could also be used as a stand-alone system.

Les travaux présentés contribuent à un des enjeux majeurs de l'Entreprise Etendue liée au domaine de la production d'énergie électrique. L'objectif est de maintenir en dynamique la qualité des services rendus par les processus de production. Ces travaux ont ainsi pour objet, en se référant au cadre de modélisation d'Entreprise GERAM, de proposer une méthodologie réutilisable pour l'automatisation intégrée des centrales hydroélectriques. Ces dernières étant structurellement des systèmes stables, cette méthodologie est basée sur une approche orientée processus et aboutit au développement de modèles pérennes et réutilisables. Le point central de cette méthodologie consiste en la définition d'un modèle de référence ICMMS (Intelligent Control-Maintenance-technical Management Systems) formalisant la connaissance générique, de niveau terrain, applicable à l'automatisation de toute centrale hydroélectrique. La mise en œuvre de ce modèle de référence conduit à la proposition d'une architecture HSAS (Hybrid Smart Automation System) qui intègre en un tout cohérent sur les points de vue Contrôle, Maintenance et Gestion Technique, les différents composants d'automatisation distribués, supportés par des actionneurs, capteurs, ou contrôleurs conventionnels de niveau terrain. Par rapport à cette architecture, les concepts innovants de "Surveillance Conditionnelle" pour l'îlot Maintenance et d' "Atténuation de Perturbations" pour l'îlot Contrôle sont définis et étudiés afin d'être intégrés au système ICMMS. De plus, nous proposons, pour la Gestion Technique, des concepts, critères et outils pour l'évaluation de performances des HGUs (Hydroelectric Generating Units). Cette contribution est basée sur la définition d'un système d'évaluation des performances économiques utilisant des descripteurs quantitatifs mesurant l'état d'efficacité, le niveau de gestion de l'exploitation et l'état de maintenance de ces unités. Une nouvelle stratégie en lien avec la maintenance, intitulée EBM (Economic performance Based Maintenance), est ainsi formalisée. L'ensemble de nos propositions est validée sur une étude de cas.

New reactor designs and the license extensions of the current reactors has created new condition monitoring challenges. A major challenge is the creation of a data-based model for a reactor that has never been built or operated and has no historical data. This is the motivation behind the creation of a hybrid modeling technique based on first principle models that adapts to include operating reactor data as it becomes available. An Adaptive Non-Parametric Model (ANPM) was developed for adaptive monitoring of small to medium size reactors (SMR) but would be applicable to all designs. Ideally, an adaptive model should have the ability to adapt to new operational conditions while maintaining the ability to differentiate faults from nominal conditions. This has been achieved by focusing on two main abilities. The first ability is to adjust the model to adapt from simulated conditions to actual operating conditions, and the second ability is to adapt to expanded operating conditions. In each case the system will not learn new conditions which represent faulted or degraded operations. The ANPM architecture is used to adapt the model's memory matrix from data from a First Principle Model (FPM) to data from actual system operation. This produces a more accurate model with the capability to adjust to system fluctuations. This newly developed adaptive modeling technique was tested with two pilot applications. The first application was a heat exchanger model that was simulated in both a low and high fidelity method in SIMULINK. The ANPM was applied to the heat exchanger and improved the monitoring performance over a first principle model by increasing the model accuracy from an average MSE of 0.1451 to 0.0028 over the range of operation. The second pilot application was a flow loop built at the University of Tennessee and simulated in SIMULINK. An improvement in monitoring system performance was observed with the accuracy of the model improving from an average MSE of 0.302 to an MSE of 0.013 over the adaptation range of operation. This research focused on the theory, development, and testing of the ANPM and the corresponding elements in the surveillance system.

To increase efficiency, shafts are made lighter and more flexible, and are designed to rotate faster to increase the system's power-to-weight ratio. The demand for higher efficiency in rotordynamic systems has led to increased susceptibility to transverse fatigue cracking of the shaft. Shaft cracks are often detected and repaired during scheduled periods of off-line maintenance. Off-line maintenance can be expensive and time consuming; on-line condition monitoring allows maintenance to be performed as-needed. However, inadequate (or a lack of) monitoring can allow rapidly propagating cracks to result in catastrophic shaft failure. It is therefore imperative to develop on-line condition monitoring techniques to detect a crack and diagnose its severity. A particularly useful method for transverse shaft crack detection/diagnosis is vibration monitoring. Detection, and especially diagnosis, of transverse fatigue cracks in rotordynamic systems has proven difficult. Whereas detection assesses only the presence of a crack, diagnosis estimates important crack parameters, such as crack depth and location. Diagnosis can provide the operator with quantitative information to assess further machinery operation. Furthermore, diagnosis provides initial conditions and predictive parameters on which to base prognostic calculations. There is a two-fold challenge for on-line diagnosis of transverse fatigue crack parameters. First, crack characterization involves specifying two important parameters: the crack's depth and location. Second, the nature of rotating machinery permits response measurement at only specific locations. Cracks are typically categorized as breathing or gaping; breathing cracks open and close with shaft rotation, while gaping cracks remain open. This work concerns the diagnosis of gaping crack parameters; the goal is to provide metrics to diagnose a crack's depth and location. To this end, a comprehensive approach is presented for modeling an overhung cracked shaft. Two linear gaping crack models are developed: a notch and a gaping fatigue crack. The notch model best approximates experimentally manufactured cracks, whereas the gaping fatigue crack model is likely more suited for real fatigue cracks. Crack diagnosis routines are established using free and forced response characteristics. Equations of motion are derived for both crack models, including excitation due to gravity and imbalance. Transfer matrix techniques are established to expediently obtain the steady-state system response. A novel transfer matrix technique, the Complex Transfer Matrix, is developed to distinguish forward and backward whirl components. The rotor's angular response is primarily employed in this work for crack detection and diagnosis. The overhung shaft induces an increased sensitivity to variations in crack depth and location. In addition, an available overhung rotordynamic experimental test rig allows for comparison of the current analytic results to previously obtained experimental results. Under the influence of gravity, the steady-state response of the cracked system includes a prominent 2X harmonic component, appearing at a frequency equal to twice the shaft speed. The magnitude of the 2X harmonic is strongly influenced by the shaft speed. A resonant response occurs when the shaft speed reaches half of a system natural frequency. This work demonstrates that the profile of the 2X harmonic versus shaft speed is a capable diagnostic tool. Identification of the 2X resonance frequency restricts the crack parameters to certain pairs of location and depth. Following this limiting process, the magnitude of the 2X harmonic is used to identify the crack's depth and location. Orbital shapes at the rotor are discussed, as are orbital modes of the shaft deflection. Quantitative results and qualitative observations are provided concerning the difficulty of crack detection and diagnosis.

Induction motors are widely used in industrial processes. The malfunction of a motor may not only lead to high repair costs, but also cause immense financial losses due to unexpected process downtime. Since thermal overload is one of the major root causes of stator winding insulation failure, an accurate and reliable monitoring of the stator winding temperature is crucial to increase the mean time to catastrophic motor breakdown, and to reduce the extraordinary financial losses due to unexpected process downtime. To provide a reliable thermal protection for induction motors fed by motor control devices, a dc signal-injection method is proposed for in-service induction motors fed by soft-starter and variable-frequency drives. The stator winding temperature can be monitored based on the estimated stator winding resistance using the dc model of induction motors. In addition, a cooling capability monitoring technique is proposed to monitor the cooling capability of induction motors and to warn the user for proactive inspection and maintenance in the case of cooling capability deterioration. The proposed cooling capability monitoring technique, combined with the proposed stator winding temperature monitoring technique, can provide a complete thermal protection for in-service induction motors fed by motor control devices. Aside from online thermal protection during a motor's normal operation, the thermal protection of de-energized motors is also essential to prolong a motor's lifetime. Moisture condensation is one of the major causes to motor degradation especially in high-humidity environments. To prevent moisture condensation, a non-intrusive motor heating technique is proposed by injecting currents into the motor stator winding using soft-starters. A motor's temperature can be kept above the ambient temperature due to the heat dissipation, so that the moisture condensation can be avoided. To sum up, active stator winding temperature estimation techniques for induction motors under both operating and de-energization conditions are proposed in this dissertation for both thermal protection and optimizing the operation of a motor system. The importance of these proposed techniques lies in their non-intrusive nature: only the existing hardware in a motor control device is required for implementation; a motor's normal operation is not interrupted.

Unscheduled aircraft maintenance causes a lot problems and costs for aircraft operators. This is due to the fact that aircraft cause significant costs if flights have to be delayed or canceled and because spares are not always available at any place and sometimes have to be shipped across the world. Reducing the number of unscheduled maintenance is thus a great costs factor for aircraft operators. This thesis describes three methods for aircraft health monitoring and prediction; one method for system monitoring, one method for forecasting of time series and one method that combines the two other methods for one complete monitoring and prediction process. Together the three methods allow the forecasting of possible failures. The two base methods use decision trees for decision making in the processes and genetic optimization to improve the performance of the decision trees and to reduce the need for human interaction. Decision trees have the advantage that the generated code can be fast and easily processed, they can be altered by human experts without much work and they are readable by humans. The human readability and modification of the results is especially important to include special knowledge and to remove errors, which the automated code generation produced.

[Truncated abstract] Centrifugal pumps are the most prevalent, electrically powered rotating machines used today. Each pump is designed to deliver fluid of a given flow rate at a certain pressure. The point at which electrical energy is converted most efficiently into increased pressure is known as the Best Efficiency Point. For a variety of reasons, pumps often operate away from this point (intentionally or otherwise), which not only reduces efficiency, but also increases the likelihood of premature component failure. Acoustic emissions (AE) are high frequency elastic waves, in the range of 20-2000kHz, released when a material undergoes localised plastic deformation. Acoustic emission testing is the process of measuring and analysing these stress waves in an attempt to diagnose the nature and severity of the underlying fault. AE sensors mounted on the surface of a machine or structure also detect any stress waves generated within the fluid being transmitted through to the structure. Unfortunately, attempts to detect incipient component faults in centrifugal pumps using acoustic emission analysis have been complicated by the sensitivity of AE to a pump?s operating state. Therefore, the aim of this thesis was to determine how acoustic emission monitoring could be used to identify the hydraulic conditions within a pump. Data was collected during performance tests from a variety of small end-suction pumps and from one much larger double-suction pump. A system was developed to collect, process and analyse any number of AE features (be they related to discrete AE events, or due to the continuous background AE level) from continuously operating equipment. ... Unfortunately, results from smaller pumps were less conclusive, particularly at low flows, probably due to the relatively small changes in hydraulic energy across the range of flows, and consequent sensitivity to the testing process. However, even in these pumps consistent patterns in hit energies were observed resulting in the conclusion that low to medium flows in centrifugal pumps are typified by a very large number of very low energy (VLE) events. These decrease in number and increase in energy as flow approaches BEP and/or is reduced to very low flows. High flows above BEP are marked by an absence of these VLE events, with bursts having significantly higher energies and spread over a much greater range. Unfortunately, these VLE events are too small to affect averaged trends, indicating that further work on a suitable filter is required. vi

A complete chain of analysis and synthesis system identification tools for detectability assessment and adaptive identification of parameters with physical interpretation that can be found commonly in control-oriented powertrain models is presented. This research is motivated from the fact that future powertrain control and monitoring systems will depend increasingly on physically oriented system models to reduce the complexity of existing control strategies and open the road to new environmentally friendly technologies. At the outset of this study a physics-based control-oriented dynamic model of a complete transient engine testing facility, consisting of a single cylinder engine, an alternating current dynamometer and a coupling shaft unit, is developed to investigate the functional relationships of the inputs, outputs and parameters of the system. Having understood these, algorithms for identifiability analysis and adaptive identification of parameters with physical interpretation are proposed. The efficacy of the recommended algorithms is illustrated with three novel practical applications. These are, the development of an on-line health monitoring system for engine dynamometer coupling shafts based on recursive estimation of shaft’s physical parameters, the sensitivity analysis and adaptive identification of engine friction parameters, and the non-linear recursive parameter estimation with parameter estimability analysis of physical and semi-physical cyclic engine torque model parameters. The findings of this research suggest that the combination of physics-based control oriented models with adaptive identification algorithms can lead to the development of component-based diagnosis and control strategies. Ultimately, this work contributes in the area of on-line fault diagnosis, fault tolerant and adaptive control for vehicular systems.

This thesis is concerned with developing a control and condition monitoring system for a class of fault tolerant actuators with high levels of redundancy. The High Redundancy Actuator (HRA) is a concept inspired by biomimetics that aims to provide fault tolerance using relatively large numbers of actuation elements which are assembled in parallel and series configurations to form a single actuator. Each actuation element provides a small contribution to the overall force and displacement of the system. Since the capability of each actuation element is small, the effect of faults within the individual element of the overall system is also small. Hence, the HRA will gracefully degrade instead of going from fully functional to total failure in the presence of faults. Previous research on HRA using electromechanical technology has focused on a relatively low number of actuation elements (i.e. 4 elements), which were controlled with multiple loop control methods. The objective of this thesis is to expand upon this, by considering an HRA with a larger number of actuation elements (i.e. 12 elements). First, a mathematical model of a general n-by-m HRA is derived from first principles. This method can be used to represent any size of electromechanical HRA with actuation elements arranged in a matrix form. Then, a mathematical model of a 4-by-3 HRA is obtained from the general n-by-m model and verified experimentally using the HRA test rig. This actuator model is then used as a foundation for the controller design and condition monitoring development. For control design, two classical and control method-based controllers are compared with an H_infinity approach. The objective for the control design is to make the HRA track a position demand signal in both health and faulty conditions. For the classical PI controller design, the first approach uses twelve local controllers (1 per actuator) and the second uses only a single global controller. For the H_infinity control design, a mixed sensitivity functions is used to obtain good tracking performance and robustness to modelling uncertainties. Both of these methods demonstrate good tracking performance, with a slower response in the presence of faults. As expected, the H_infinity control method's robustness to modelling uncertainties, results in a smaller performance degradation in the presence of faults, compared with the classical designs. Unlike previous work, the thesis also makes a novel contribution to the condition monitoring of HRA. The proposed algorithm does not require the use of multiple sensors. The condition monitoring scheme is based on least-squares parameter estimation and fuzzy logic inference. The least-squares parameter estimation estimates the physical parameters of the electromechanical actuator based on input-output data collected from real-time experiments, while the fuzzy logic inference determines the health condition of the actuator based on the estimated physical parameters. Hence, overall, a new approach to both control and monitoring of an HRA is proposed and demonstrated on a twelve elements HRA test rig.