Search

Considering the innately connected neutral point in a multi-functional converter system (MFCS) fed three-phase surface-mounted permanent magnet synchronous machine (SPMSM) drive, this paper develops its fault self-tolerance towards open-phase faults (OPFs) without assisted hardware. First, the operating principle of the MFCS fed SPMSM drive is explained by building a zero-sequence model. Second, to achieve fault-tolerant operations, an unconventional current vector trajectory is proposed, aiming at maintaining the torque and average input power invariable. The shape of the proposed trajectory tends to be like a heart rather than a circle with the rise of motor speed. Moreover, d-axis and zero-sequence currents vary with the position of rotor. By considering the limits of inverter voltage and current, the post-fault output capability of the drive is explicitly analyzed. Third, a new fault-tolerant control strategy is thus proposed, in which the OPF is detected by comparing the deviation between predictive and actual currents. To track the time-varying current references, a differential flatness-based current controller is further developed. Finally, experimental results are given to verify the effectiveness.

International audience; This paper tackles the problem of robust output set-point tracking for a power flow controller (PFC) for meshed DC micro-grids. The PFC is a power electronics device used to control the power (or current) flow in the lines of the grid and act as a DC circuit breaker. The state-space model of this system is bilinear with uncertain dynamics and a polynomial output. In the proposed design, the plant is first extend with an integral action on the tracking error. The cascade model composed by the plant and the integrator is then stabilised by means of a state-feedback law, with a forwarding approach. If the plant parameters are sufficiently close to their nominal value, robust regulation is then achieved. Simulations are given to validate the results.

This paper presents a state-averaged model and the design of nonlinear observers for an on/off pneumatic actuator. The actuator is composed of two chambers and four on/off solenoid valves. The elaborated state-averaged model has the advantage of using only one continuous input instead of four binary inputs. Based on this new model, a high-gain observer and a sliding-mode observer are designed using the actuator position and the pressure measurements in one of the chambers. Finally, their closed-loop performance is verified and compared on an experimental benchmark.

The existing fault-tolerant three-phase ac drives need split capacitors, auxiliary legs, and triode for alternating current (TRIACs) upon a standard three-phase two-level voltage source inverter (VSI) to obtain the self-healing ability toward open-phase faults (OPFs). In order to reduce the cost, volume, and weight, this article proposes a novel fault-tolerant three-phase surface-mounted permanent magnet synchronous machine (SPMSM) drive without split capacitors, auxiliary legs, or TRIACs. The proposed drive is still built upon a three-phase two-level VSI, whereas its dc-source is directly connected to the neutral point of the motor. First, mathematical model and equivalent circuit are established to reveal the existence of an equivalent dc/dc boost converter in the drive system. Second, the brand-new d -0 axes current references are proposed in order to preserve the motor's performance when an OPF occurs. In this case, both $\boldsymbol{i_{d}}$ and $\boldsymbol{i_{0}}$ are nonconstant. Third, an overall control strategy including the healthy and fault-tolerant modes is presented. In the fault-tolerant mode, a deadbeat current controller is developed to track the nonconstant current references. Finally, experiments are carried out on an SPMSM test-bench to verify the effectiveness of the proposed scheme.

International audience; In this article, parallel interconnection of dc/dc converters is considered. For this topology of converters feeding a common load, it has been recently shown that dynamics related to voltage regulation can be completely separated from the current distribution without considering frequency separation arguments, which inevitably limits achievable performance. Within the Hamiltonian framework, this article shows that this separation between current distribution and voltage regulation is linked to the energy conservative quantities: the Casimir functions. Furthermore, a robust control law is given in this framework to get around the fact that the load might be unknown. In this article, we also ensure that the system converges to the optimal current repartition, without requiring explicit expression of the optimal locus. Finally, resulting control law efficiency is assessed through experimental results.

International audience; The multi-functional converter system (MFCS) is a special topology of three-phase AC motor drives, which inherently integrates a DC-voltage regulation function. Firstly, in order to explain the mechanism of the MFCS, an average model is derived. An equivalent circuit is developed, which explicitly reveals that the MFCS is equivalent to a cascaded DC/DC converter-DC motor drive system. Secondly, the inverter's mean duty-cycle is identified as an independent control degree of freedom for the DC-bus voltage. Then, the influence of pulse width modulation (PWM) on the DC-bus voltage is studied. It concludes that space vector PWM is invalid for the MFCS. To address this problem, the zero-sequence voltage injection PWM (ZSVIPWM) is proposed and actually sinusoidal PWM is a special case of the ZSVIPWM. Thirdly, an overall control strategy is proposed for the MFCS, which includes not only the vector control for the motor but also a closed-loop control for the DC-bus voltage. Finally, comparative experiments with the standard topology are carried out on a permanent magnet synchronous machine prototype to verify these theoretical analyses and the control strategy.

International audience; In this paper, a robust relay control of a buckconverter is proposed. The controller is synthesized based onthe design of a continuous controller for the average model ofa buck converter usually used when Pulse Width Modulation(PWM) is considered. To deal with parameter uncertainties, thecontroller implements an integral action. In this paper, localstability is proven for the hybrid system, thereby ensuring zerosteady state error on the voltage output. A subset of the robustbasin of attraction is explicitly constructed. The effectivenessof this control is illustrated experimentally.

International audience; In this paper, we present a Transverse Electric (TE) polarized Bessel-beam antenna in the microwave range based on a leaky radial waveguide. According to the dispersion equation, an inductive sheet over a ground plane is needed. To achieve the inductive sheet, a grid structure printed on a dielectric substrate is designed. Besides, the radial waveguide is coaxially fed with a quasi-loop shaped structure to excite TE mode. The proposed antenna is investigated by simulation and measurement. In the measurement, a small loop is used to probe near magnetic field, and accurate measured magnetic intensity profiles are presented. The measurements show that TE-polarized Bessel beams can be launched within the non-diffractive range of 145 mm (2.803 λ0) at 5.8 GHz.

Torque ripples are usually found in brushless DC (BLDC) motor control applications. This article presents the design of non-sinusoidal back electromotive forces (BEMF) observer for a BLDC motor in order to reduce electromechanical torque ripples. For their robustness quality, a high order sliding mode observer is used to observe the non-sinusoidal BEMF which allows to determine magnetic field angular relative position leading to a so-called extended Park’s transformation. Using this extended transformation, speed and torque vector control can be made with better performances than the classical Park’s transformation. Simulation results are performed to show the effectiveness of the proposed approach.

International audience; Energy storage elements such as supercapacitors are widely used in high-power applications. However, due to single cell voltage limitation, an energy storage system with a large number of supercapacitors is often employed. Energy management systems are associated to energy storage systems in order to assure user and equipment safety. Balancing circuits, which enable the equalization of the voltage of each element in series, are a part of the energy management system device. The work presented in this paper aims to enhance the lifetime of energy storage systems. It relies on better controlling balancing circuits on the terminals of the storage system elements. With the conventional function of balancing circuit, the energy storage system is limited by its weakest element that may fail prematurely. Thus, a new balancing approach is presented, discussed, and analyzed. It is based on the elements degradation level prediction. The model predictive control used with the new approach aims to equalize aging speed between elements of a module and ensures a maximum lifetime to the energy storage system. A comparison with the conventional control shows that adopting this new approach, with the same equipment, can enhance the storage system's lifetime by dozens percent.

International audience; This paper presents a control algorithm for the parallel interconnection of heterogeneous power converters. A single resistive load is assumed to be fed by an arbitrary number of buck converters via a common DC bus. The approach is based on control allocation theory and a constrained quadratic optimization algorithm. The strategy achieves a fast voltage response with an optimal current distribution among the converters, while taking into account the current limits, the dynamic response, and the efficiency of the individual converters. An interesting by-product of the approach is the ability to put converters in and out of service through trivial adjustments of the code. The benefits of the approach are assessed through simulations and an experimental evaluation.

International audience; Electrical traction using induction machine sensorless control requires high observer performance for all speed ranges, even for low speed or regenerative braking conditions which appear frequently during long time. It is well known that the speed of induction motors is unobservable at very low stator frequencies. This paper uses an observability index to continuously analyze speed observability for sensorless control of induction machines. The correlation between observability-index and observer performance is illustrated in a Hardware in the Loop (HIL) experimental test-bench combining the well-known vector control with an extended Kalman filter. Thanks to the observability-index information, an optimal strategy is proposed to design controllers to guide the system away from undesirable behavior and avoid the weak observability-index region by taking into account all working constraints. A simplified case is presented to improve the speed observer performance, which was tested in the same conditions with the same HIL test-bench to experimentally validate the proposed sensorless control for traction applications.

International audience; This paper focuses on the observer-based methods for induction machine sensorless drive. It is well known that the use of speed observer is limited at a very low stator frequency due to speed unobservability. Some existing methods propose to avoid the zero-stator-frequency working points. These methods are nonetheless limited to high-torque operations. The first contribution of this paper is to analyze finely the speed observability, which leads to the definition of an observability index. The correlation between this index and the observer performance is analyzed and illustrated. The second contribution is to propose a control strategy that maximizes the observability index and improves the observer performance. The proposed strategy is tested first in open loop to evaluate the speed observer performance, and then, in closed loop to illustrate its pertinence for induction machine sensorless drive. All these results are validated by an experiment. The performed experimentation shows that the proposed control strategy gives better results than the existing method that avoid the zero-frequency working points. It offers a new path for induction machine sensorless drives.

International audience; Despite their alias-free operation, continuous time ADCs can have their power consumption overwhelmed when presented out-of-band, high frequency, high amplitude signals. A new continuous time ADC architecture which, thanks to its embedded low-pass filter transfer function is proposed, enables a power efficient quantisation in the presence of such signals. Moreover, the half-digital nature of this approach simplifies the implementation of the proposed system compared to a standard filter-first solution.

International audience; Digital control technique is becoming an attractive alternative for high-frequency switching mode power supply (SMPS). This paper introduces a fully synthesizable digital sliding-mode (SM) controller for high-frequency dc/dc SMPS. The proposed SM controller is associated with a digital pulsewidth modulation (DPWM) block to operate at constant and high switching frequency compared to already published approaches. The digital SM controller is experimentally validated by means of field-programmable gate array in a 4-MHz step-down converter with 10-bit analog-to-digital conversion (ADC) and 11-bit DPWM. Measured dynamic performances are superior to conventional digital controllers for dc/dc SMPS.

International audience; The development of systems-on-chip requires embedded power management solutions due to the large number of power domains. The switched-capacitor voltage regulator is a suitable candidate as capacitors may be integrated whereas inductors still suffer limitations in that respect. Literature covers proposals of optimized power stages and several dedicated controllers for switched-capacitor DC-DC converters. Unfortunately the results do not cover systematic stability analyses. The paper proposes an original and systematic approach for the stability analysis of a switched-capacitor voltage regulator using sampled-data modeling. An application is given for a one-phase converter with a frequency-modulation based controller. The stability of the open-loop and closed-loop model is proposed and can be extended to multi-phase configurations.

International audience; This paper presents a new control method applied to the electro-pneumatic field. This strategy originates from the hybrid control theory recently developed for the control of asynchronous or synchronous electrical motors, e.g. Retif (2004). The interest of this strategy concerns the possibility of using standard on/off distributors instead of the usual servodistributors (components issued from proportional technology) for the force control of a pneumatic cylinder. Distributor components have less performance but are cheaper than a servodistributor. The aim is to obtain, with a distributor, the same performances as servodistributors on the global system. Based on both cylinder and distributor models, the hybrid control presented here chooses the best state for each on/off distributor to reach the desired force value. Experimental results are presented and discussed.

International audience; Power management of mobile digital systems-on-chip necessitates large voltage regulation performances. DC/DC converters used to supply digital cores are facing stringent constraints with respect to load transients, reference tracking and accuracy. The fast transient response offered by sliding-mode control is well suited to DC/DC converters with fast varying load. Fixed switching-frequency is also mandatory for EMI and noise related issues. Fixed-switching frequency sliding-mode control has been experimented as a multi-loop system. There is a necessity to investigate the stability when multi-loops interact and the parameter set for each loop is quite large. The paper details a methodology for the stability analysis of a frequency-controlled sliding-mode DC/DC converter. The proposed stability analysis takes the possible cross-interactions into account and can be used to optimise the DC/DC controller without the limitations of a non-interaction criterion. As a verification a DC/DC converter is implemented in CMOS 130 nm. Experimental results agree with the analysis main recommendations.

International audience; Listening music with a headset is now a classical feature of mobile phones. A linear amplifier is still employed to drive the headset as the audio quality must challenge dedicated audio players with somewhat 100 dB of signal-to-noise ratio. This feature must also have the lowest impact on the platform autonomy, size, and cost. A symmetrical power supply is required by such audio amplifiers due to the standard jack connector of headsets. Using an inductive switch mode power supply (SMPS) is the most energy efficient solution to supply audio amplifiers from a battery. Classically, two dc/dc converters are employed to generate the required two symmetrical supply rails from a standard Lithium-Ion battery. It is at the cost of numerous bulky and expensive external components. A solution to reduce the number of external passive components is to consider a single-inductor, bipolar-outputs, (SIBO) dc/dc converter. An early prototype of such a converter has been realized on a 130-nm silicon process and offers primary results, nonoptimal but promising. The feedback control structure is studied based on a small signal model that serves to define the pairing of the loops for designing a decentralized controller. A set of controllers is designed by a pole placement technique and transient performances are demonstrated on audio patterns thanks to a piecewise linear model of the converter. Eighty percent peak efficiency is measured with the primary demonstrator but latter simulation results yield an expected improvement to 90%. The audio specifications appear very constraintful and the proposed SMPS does not meet entirely the latter figures. Limitations are detailed. However, many other applications can benefit from the proposed SIBO dc/dc converter

International audience; Mobile applications necessitate nowadays huge digital resources. Power management of a digital systems-on-chip is based on dynamic voltage scaling. DC/DC converters used to supply the digital system-on-a-chips are facing stringent constraints with respect to load transients, line transients, and reference tracking. Hysteretic control is known as the most convenient control scheme with a fair tradeoff between transient performances, analog implementation and power consumption. Fixed-switching frequency hysteretic control has been experimented as well as full sliding-mode control. Transient performances are reduced due to latencies introduced in the switching frequency control. A new analog implementation of the sliding-mode control is presented here with switching frequency control using a particular analog phase-locked loop but preserve transient performances. The dc/dc converter is implemented in CMOS 130 nm. The switching frequency range has been voluntarily limited and excludes the possible integration of passive components. A hybrid demonstrator is presented with peak efficiency of 89% at 2-W output power.

This paper presents the application of fixed frequency (or indirect) Sliding Mode Control (SMC) to the DC-DC Single-Ended Primary Inductor Converter (SEPIC) where the switching frequency is in the range of hundreds of kHz and consequently a FPGA is required. Due to the constraint of FPGA, only the output voltage is measured. As the proposed SMC requires the knowledge of all the states, an extended Kalman observer is introduced to estimate the state vector and the load variation. A multi-bit second-order Δ-Σ modulator is used to effectively achieve 11-bit resolution at high-frequency through only a 8-bit hardware Core Digital Pulse - Width - Modulator (DPWM). Simulation and experimental studies are conducted for a laboratory prototype with switching frequency of 500 kHz. Results proved the performance of the proposed solution.

Taking advantage of FPGA's attractive features, this paper presents an improved digital pulse-width-modulator (DPWM) based sliding-mode controller (SMC) for boost con- verter that effectively alleviates the quantization effects. The dithering Multi-stAge-noise-SHaping (MASH) DPWM is intro- duced exhibiting a better idle tone suppression effect that achieves relatively higher effective number of bit (ENOB). The Linear Feedback Shift Register (LFSR) replaces the cumbersome pseudo-random generator as the dither generation module that is proven to be more effective. The SMC with the proper sliding coefficients aiming at a better dynamic response than the traditional PID controller cooperates with our proposed DPWM. Two individual boards, an analog-to-digital converter (ADC) and a boost converter, connecting to a Virtex-II FPGA platform compose a close-loop test environment. Experimental results verify the switching-mode-power-supply (SMPS) close-loop operation at 1MHz switching frequency with an 11-bit effective DPWM resolution.

International audience; Dans ce papier, une méthode d'identification dynamique pour des systèmes non linéaires linéairement paramétrés est développée. La technique proposée repose sur l'utilisation d'un observateur d'état étendu aux paramètres combinée à une technique des moindres carrés et une approche par variable instrumentale. L'étude de l'effet du bruit sur l'identification montre la consistance et la robustesse de la technique. L'application en expérimentation sur une machine synchrone à aimants permanents confirme les résultats obtenus par simulation.

This paper presents a control scheme for a multilevel multi-cell converter. For this type of converters, load current and capacitor voltages must be jointly controlled. Moreover the real-time constraint is important. This constraint leads us to propose a control based on a simplified state-space model. The model allows predicting the state vector evolution for every converter configuration. The control algorithm directly determines the converter switching state which minimizes a simple cost function. A normalization of the state variables is proposed for the cost function calculation in order to ensure a trade-off between the tracking of the load current and the tracking of the capacitor voltages. The proposed control scheme is detailed and compared with a classical control scheme with simulations. Finally, experimental results are presented to show the effectiveness of the proposed method.

This paper presents the experimental results-based comparison of two predictive current controls (PCC) for permanent magnet synchronous machines (PMSM) drives. The first tested control scheme is based on a model including the inverter and the PMSM and taking into account the discrete nature of the inverter leg states. It predicts the future evolution of the currents for each possible configuration of the inverter. The switching state which minimizes a given cost function is selected. The selected inverter state is applied during the next sampling time. The second tested control scheme uses a model of the PMSM to predict the output voltages which allow to reach desired currents after one modulation period. A new algebraic method is presented to directly compute the duty cycle of each leg. Then a modulator generates the corresponding gate drive pulses of the inverter. These two control schemes are tested with a 1.6 kW PMSM drive at several operating points during steady state and transient operation. A detailed comparison of results is given. Advantages and drawbacks of each method are discussed.

Many research efforts have been dedicated to matrix converters for several years. As major technological issues are now solved, this structure will widespread in industrial applications, in particular with AC motors. Current control is a key issue for AC motor drives, so many control schemes have been proposed. Some of them proposed at first for inverters, were applied to matrix converters. Among algorithms used with inverters, predictive control shows very good performances. In this paper a new control scheme is proposed for a matrix converter- fed permanent magnet synchronous machine. Literature about matrix converter technology and control and about predictive control for inverter-fed AC machines is reviewed. The proposed predictive control principle, the model of the whole machine - converter and the cost-function are detailed. The method offers a trade-off between the quality of motor currents and input power factor. Finally experimental results are reported. The feasibility and the effectiveness of the proposed method is assessed.

This paper presents a predictive current control (PCC) strategy applied to an induction motor (IM) drive. In contrast to conventional vector control, where the inverter is not taken into account by the controller, the PCC considers the state of the inverter as a discrete control variable. Based on a simplified state space model of the IM and the inverter, the proposed control scheme calculates the state vector evolution direction in the d- and q-reference frame for all possible switching states of the inverter. The switching state which minimizes a given cost function is selected and is applied during an adapted duration which is calculated for each computation cycle. As the control requires the knowledge of the instantaneous rotor fluxes and in order to improve the control robustness, a reduced-order extended high gain observer is proposed for estimation of rotor fluxes and inverse rotor time constant. The simulation results validate the observer. The controller is implemented on a 5.5 kW IM drive with a digital-signal-processor (DSP) on a DSpace1103 board. Experimental results show the effectiveness and the performances of the proposed control.

Nos remerciements à IEEE pour l'autorisation de mise à disposition du papier complet. © IEEE Copyright Notice : Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.; International audience; This paper presents a control scheme suitable for systems composed of a continuous process modulated in energy by a power converter with a finite number of topologies. To track the continuous reference values, a topology of the power converter is determined from a criterion based on a process state variable model and taking into account the possible topologies of the power converter. The proposed hybrid control scheme is applied to an electrical motor drive composed of an inverter coupled to a permanent magnet synchronous machine. An evolution which insures a fixed modulation frequency is also proposed. Experimental results validate the feasibility and effectiveness of the proposed schemes.

ISBN: 978-0-7695-3336-0; International audience; In this paper, a model of Embedded DC Motor Ethernet Control System is proposed with analysis on transmission time delay and data packet dropout. Based on this model, the ECS is described as a two-state asynchronous dynamical system with output feedback control and then implemented on a platform which uses a PC as a central controller and an ARM9 kit as a remote controller. As a key part of developments, embedded programs including client program and Linux Module are realized on the kit. The platform is proved to be flexible to run different control algorithms and extensibility to add nodes. The experiments and results demonstrate the validity of the system.

Nos remerciements à IEEE pour l'autorisation de mise à disposition du papier complet. © IEEE Copyright Notice : Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.; International audience; This paper presents the implementation of a hybrid-control strategy applied to a permanent-magnet synchronous-motor (PMSM) drive. Hybrid control is a general approach for control of a switching-based hybrid system (HS). This class of HS includes a continuous process controlled by a discrete controller with a finite number of states. In the case of ac motor drives, in contrast to conventional vector control like proportional-integral control or predictive control, where the inverter is not taken into account by the controller, hybrid control integrates the inverter model and considers the state of the inverter as a control variable. It allows to obtain faster torque dynamics than vector-control algorithms. The hybrid control algorithm requires both computing velocity for real-time implementation and code flexibility for management of low-performance functions and analog-digital interfaces. Codesign appears as a promising methodology for partitioning hybrid-control algorithm between software (flexible) and hardware (velocity) while taking care of overall time constrains. In this paper, the implementation of hybrid-control algorithm for a PMSM drive is performed through a codesign approach on an Excalibur board, embedding a CPU-core (Nios-2 by Altera) inside an APEX20KE200EFC484-2X field-programmable gate array. The partitioning of software and hardware parts is explained. Experimental results show the effectiveness of the implementation. Performances, advantages, and limitations are discussed.

Analogue control of monolithic DC/DC converters is technologically coming to a limit due to high switching frequency and a request for large regulation bandwidth. Digital control is now experimented for low-power low-voltage switch-mode power supply. Digital implementation of analogue solutions does not prove real performances. This paper compares a classical digital controller to a candidate alternative strategy. Sensitivity functions are used to compare controller performances. An off-line approach using fuzzy logic to quantify controller performances and a genetic algorithm to obtain an optimal controller is presented. A so-called RST algorithm optimized with this approach shows better performances.

Les systèmes électrotechniques et plus particulièrement l'association d'un convertisseur statique avec une charge constitue un système dynamique hybride pour lequel la commande est sujette à des contraintes de calcul en temps réel sévères (quelques centaines de microsecondes). La spécificité de ces systèmes hybrides nous a conduits à proposer un modèle simplifié dont la validité est restreinte à l'horizon de décision et une commande pouvant satisfaire à ces contraintes. Diverses stratégies de commande ont été testées et validées sur des démonstrateurs en temps réel sur les applications suivantes : convertisseur multicellulaire, machine synchrone, machine asynchrone et vérin électropneumatique. Electrotechnical systems and specially the association of a static converter to a load constitute a dynamic hybrid system. Real-time constraints are important for the control of these systems (a few hundreds of microseconds). Specificities of these hybrid systems lead us to propose a simplified model with validity limited to the calculation occurrence and control approach which can fulfil these constraints. Different control strategies were established, tested and validated on real-time experimental equipments such as multi-cell converter, synchronous and asynchronous machine and pneumatic cylinder.

International audience; Analogue control of monolithic DC/DC converters is coming to a limit due to high switching frequency and a request for large regulation bandwidth. Digital control is now experimented for low-power low-voltage switch-mode power supply. Digital implementation of analogue solutions does not prove real performances. Other digital controllers have been experimented but applied to discrete converters. This paper compares a classical digital controller to a possible alternative strategy. Sensitivity functions are used to compare controller performances. RST algorithm determined by robust pole assignment shows better performances

International audience; At the present time in the pneumatic field, when a system requires a wide range of force control,servodistributors are currently used. On/off distributors are used when the system is simpler and only a smallrange of force values are required [1]. The work presented here consists of using on/off distributors in a systemwhich requires a wide range of force values. This has been carried out in order to reduce the cost of complexsystems [2-4], because present day distributors are five to ten times cheaper than servodistributors.This paper presents a new control method applied to the electro-pneumatic field. This strategy originates fromthe hybrid control theory recently developed for the control of asynchronous or synchronous electrical motors,[5, 6]. This control strategy is an improvement of the proposed one in [4].Based on both cylinder and distributor models, the hybrid control presented here chooses the best state for eachon/off distributor to reach the desired force value. Hybrid control is based on a state space model for both theenergy modulator and the continuous sub-process. For this model, a control vector, depending on the number ofpossible configurations for the energy modulator, is defined. Two formal approaches have been developed forchoosing a control vector to track the reference state of interest in the state space. Experimental results arepresented and discussed.

As Alternating Current (AC) is well suited for most of the production, transmission, and distribution applications, its massive use is easy to understand. However, for over a century, the benefits of High Voltage Direct Current (HVDC) for long-distance energy transmission are well known. To connect both, AC/DC converters are mandatory, whose nature evolves with technological progress. After the problematic induced by HVDC on AC/DC converters is presented, this manuscript is focused on three topologies: Modular Multilevel Converter (MMC), Alternate Arm Converter (AAC) and Series Bridge Converter (SBC). They are presented, sized, analyzed thoroughly, and compared in quantitative terms, using original key performance indicators. It appears that MMC and SBC are particularly promising. The conventional control method of the MMC is then presented, and its structural properties are highlighted. A first original control law is presented, with similar performances but less complexity than the state-of-the-art. A second control law, non-linear and based on differential flatness theory, is introduced. It allows a very fast power tracking response while ensuring the global exponential stability of the system. These control laws are tested in simulation, using an average model and a detailed model with 180 sub-modules per arm. The last part is dedicated to the SBC. After a modeling step, some results regarding its structural analysis are presented, and an original control law is introduced. The essential role of the transformer for series converters like the SBC is highlighted. Finally, the performance of the proposed control law is assessed in simulation., Le courant alternatif (AC) se prêtant bien à la majorité des problématiques de production, de transport et de distribution de l'électricité, on comprend qu'il soit massivement utilisé. Cependant, depuis plus d'un siècle, les bénéfices du courant continu haute tension (HVDC, pour High Voltage Direct Current) pour les longues distances sont bien connus. Aux interfaces, des convertisseurs AC/DC sont requis, leur composition évoluant au fil des avancées technologiques. Après avoir présenté les spécificités du HVDC et les contraintes qu'il introduit sur les convertisseurs AC/DC, ce manuscrit se focalise sur trois topologies : Modular Multilevel Converter (MMC), Alternate Arm Converter (AAC) et Series Bridge Converter (SBC). Elles sont présentées, dimensionnées et analysées en détail, puis comparées de façon quantitative en utilisant des indicateurs de performance originaux. Il en ressort que le MMC et le SBC sont particulièrement intéressants. La méthode de commande conventionnelle du MMC est ensuite présentée et ses propriétés structurelles sont mises en évidence. Une première loi de commande originale est présentée, avec des performances similaires mais une complexité inférieure à l'état de l'art. La seconde est non linéaire, basée sur la théorie de la platitude différentielle, et permet un suivi de puissance très rapide tout en assurant la stabilité exponentielle globale du système. Ces lois de commande sont évaluées en simulation, avec un modèle moyen et un modèle détaillé intégrant 180 sous-modules par bras. La dernière partie concerne le SBC. Après l'avoir modélisé, des résultats concernant une analyse structurelle de la topologie sont présentés ainsi qu'une loi de commande originale. Le rôle fondamental du transformateur pour les convertisseurs à structure série comme le SBC est souligné. Enfin, les performances de la loi de commande proposée sont testées en simulation.

This thesis is about the use of control allocation theory in the field of electrical engineering and power electronics. It is composed of two parts. The first one is related to control allocation theory and the second one focuses on the control of a particular application: the parallel interconnection of DC/DC converters. The control allocation theory gives tools to design control laws for input redundant systems. However, current definitions about those systems suffer from a lack of uniformity. That is why the first part provides new definitions and characterisations of input redundant systems in order to take input and state constraints into account. Furthermore, the control allocation method named "geometric approach" is enhanced to deal with input constraints while designing the control law. Thanks to those contributions, it is shown in the second part that the system composed of the parallel interconnection of DC/DC converters to a single load is an input redundant system. On this system, some methods of the control allocation theory are used to design a controller with stability and performance guarantees. Simulations and experimentations are given to highlight those development., Cette thèse s'inscrit dans le cadre de l'utilisation de l'allocation de commande dans le domaine de l'électrotechnique et l'électronique de puissance. Elle comporte deux volets, le premier orienté sur la théorie de l'allocation de commande et le second en lien avec la conception de lois de commande pour les convertisseurs en parallèles. L'allocation de commande donne des solutions pour commander des systèmes redondants en entrée. Toutefois les définitions et caractérisations de cette classe de systèmes souffrent d'un manque d'uniformité. C'est pourquoi dans le premier volet, cette thèse présente de nouvelles définitions et caractérisations générales de systèmes linéaires redondants en entrée en prenant en compte les contraintes d'entrée et d'état. Le cadre d'application de la technique d'allocation dite "géométrique" est également étendu pour prendre en compte des contraintes d'entrée. À la lumière de ces contributions, nous montrons dans le second volet que le système de convertisseurs continus en parallèle alimentant une charge unique est redondant en entrée. Sur ce système, des techniques de commande basées sur l'allocation de commande sont utilisées en apportant des preuves de stabilité et de performance. Ces développements sont appuyés par des simulations et des expérimentations.

This thesis is about using control allocation for electrotechnics and power electronics. It is composed of two parts. The first one is linked with control allocation theory whereas the second one is related to designing control law for parallele interconnection of power converters. Control allocation provides methods for controlling input redundant systems. However, definitions and characterizations of input redundant systems suffer from a lack of uniformity. As a result, the first part of this thesis presents new definitions and generalized characterizations of such systems, while taking input and state constraints into account. The application framework of geometric control allocation is also extended to take input constraints into account.Using those theoritical developments, the second part of this thesis show that the parallel interconnection of DC/DC converters is an input redundant system. Control laws, based on control allocation are then used to bring stability and performance conditions. Those contributions are highlighted by simulations and experiments., Cette thèse s’inscrit dans le cadre de l’utilisation de l’allocation de commande dans le domaine de l’électrotechnique et l’électronique de puissance. Elle comporte deux volets, le premier orienté sur la théorie de l’allocation de commande et le second en lien avec la conception de lois de commande pour les convertisseurs en parallèle. L’allocation de commande donne des solutions pour commander des systèmes redondants en entrée. Toutefois les définitions et caractérisations de cette classe de systèmes souffrent d’un manque d’uniformité. C’est pourquoi dans le premier volet, cette thèse présente de nouvelles définitions et caractérisations générales de systèmes linéaires redondants en entrée en prenant en compte les contraintes d’entrée et d’état. Le cadre d’application de la technique d’allocation dite "géométrique" est également étendu pour prendre en compte des contraintes d’entrée. À la lumière de ces contributions, nous montrons dans le second volet que le système de convertisseurs continus en parallèle alimentant une charge unique est redondant en entrée. Sur ce système, des techniques de commande basées sur l’allocation de commande sont utilisées en apportant des preuves de stabilité et de performance. Ces développements sont appuyés par des simulations et des expérimentations.

Current trends in global energy production indicate that renewables will continue to increase their market share due to continuous efficiency improvements and cost reductions. Power converters constitute the interfaces that enable energy transfers in microgrids between sources, storage and loads, playing a fundamental role in their operation. These devices are required to meet particular efficiency, robustness and stability requirements to guarantee a proper operation. The present work is focused on two particular problems present in the operation of power converters: control and observation. These problems are hard to solve because of the nonlinearities and complex behaviors present in power converters. The mathematical model used to represent power converters is the switched system. Based on this model we take elements from subjects like Lyapunov stability, the method of moments, algebraic geometry and direct filtering, and propose novel approaches to control and observation of switched systems. We introduce moment relaxations of switched systems. These representations allow to obtain models where the switching input is mapped to a moment space. This map removes the nonlinearity associated to the switching input and yields a model which is more suitable for performing numerical computations. This is the fundamental idea in the proposed parameter-varying control method. After computing a control signal for the relaxed model, the control signal for the switched system can be recovered. This approach has shown good performance with respect to reference tracking and stability. A data-driven approach for the observation of switched systems is proposed. This method involves the design of a direct filter that computes worst-case bounds on the estimation error. This method is applied to the case of power converters operating in continuous and discontinuous modes. A practical implementation is described, and its performance is compared with other estimation approaches., En matière de production énergétique, les tendances actuelles indiquent que la part de marché mondiale des énergies renouvelables continuera d’augmenter du fait de l’amélioration continue de l’efficacité des équipements et de la réduction de leurs coûts. Les convertisseurs de puissance jouent un rôle fondamental dans le fonctionnement des réseaux électriques car ils permettent les transferts d'énergie entre les sources, les éléments de stockage et les charges. Ces dispositifs doivent répondre à des exigences particulières d'efficacité, de robustesse et de stabilité pour garantir un fonctionnement correct. Le travail présenté dans ce mémoire est centré sur deux problèmes particuliers liés au comportement des convertisseurs de puissance : la commande et l’observation. Ces problèmes sont difficiles à résoudre en raison des non-linéarités et des phénomènes physiques complexes présents dans ces composants. Le modèle mathématique utilisé pour représenter les convertisseurs de puissance est le modèle dit à commutation. Sur la base de ce modèle, nous exploitons des outils abordés dans des problématiques telles que la stabilité de Lyapunov, la méthode des moments, la géométrie algébrique et le filtrage direct. Nous proposons des approches novatrices pour commander et observer ces systèmes commutés. Nous introduisons notamment la notion de moments de relaxation pour les systèmes commutés. Ces représentations permettent d’obtenir des modèles permettant d'établir une cartographie de l’entrée de commutations dans l'espace des moments. Cette carte supprime la non-linéarité associée à l'entrée de commutation et fournit un modèle plus approprié pour effectuer les calculs numériques de la commande. C'est l'idée fondamentale de la méthode de commande à variation de paramètres proposée. Après avoir calculé un signal de commande pour le modèle relaxé, le signal d'entrée pour le système commuté peut alors être déduit. Cette approche a montré de bonnes performances en termes de suivi de référence et de stabilité. Une approche reposant sur les données pour l'observation des systèmes commutés est proposée. Cette méthode implique la synthèse d'un filtre direct qui calcule les limites du cas le plus défavorable sur l'erreur d'estimation. Cette méthode est appliquée au cas des convertisseurs de puissance fonctionnant en mode continu et discontinu. Une mise en oeuvre pratique est effectuée et ses performances sont comparées à d’autres méthodes d’estimation.

The problem of speed sensorless control of induction machine is a well-studied problem in the field of automatic control and electrical engineering. This work addresses the problem of the feasibility of induction machine control over the whole speed range and for any type of induction machine, aiming at being used in a railway application. An intensive study of observability has defined a continuous quantitative measurement of the speed observability, named observability index. An observer was developped to confirme the correspondence between low-observability index areas and poor observation quality. The observability index defined in this thesis is thus a way to predict the observation performance of systems. The main originality of the work presented in this thesis is the use of this observability index in the induction machine speed sensorless drive. By taking advantage of the degree of freedom given by the variation of the flux amplitude, we propose to guarantee the observability index value above a given threshold while respecting the other constraints of the control. The speed observation becomes precise over the entire speed range, and a long time operation is possible on any point, without interfering with torque regulation. Experiments on real-time simulator confirm these results, including when slipping and sliding occurs. The electrical parameter observation is also studied to ensure the accuracy of speed observation. For this, the sensitivity of the speed observation to electrical parameter variations and the observability of these parameters are studied, leading to the definition of the electrical parameter observation strategy. The testings on power bench, and in real application on a regional train, validate the performance of the speed sensorless drive proposed on the entire speed range, and its suitability to the constraints of a railway application., Le problème de la commande de machine asynchrone sans capteur mécanique de vitesse est très étudié dans les domaines de l'automatique et de l'électrotechnique. Ce travail porte sur la faisabilité de ce type de commande sur l'ensemble de la plage de vitesse et pour tout type de machine asynchrone afin d'être utilisée en traction ferroviaire. Une étude poussée de l'observabilité permet de définir une mesure quantitative continue de l'observabilité de la vitesse de la machine, appelée index d'observabilité. La synthèse d'un observateur confirme ainsi la correspondance entre les zones de faible index d'observabilité et celles où l'observation est de mauvaise qualité. L'index d'observabilité défini dans cette thèse permet donc de prédire les performances d'observation d'un système. L'originalité principale des travaux présentés tient dans l'utilisation de cet index d'observabilité dans la commande de la machine asynchrone sans capteur mécanique de vitesse. En tirant profit du degré de liberté qu'offre la variation d'amplitude du flux, nous proposons de garantir un index d'observabilité au-dessus d'un seuil tout en respectant les autres contraintes de la commande. L'observation de la vitesse devient alors précise sur toute la plage de vitesse, et un fonctionnement long est possible en tout point sans perturber le couple réalisé. Des expérimentations sur simulateur numérique viennent confirmer ces résultats, y compris lors de phases de patinage. L'observation des paramètres électriques est également étudiée afin de garantir la précision de l'observation de vitesse. Pour cela, la sensibilité de l'observation de vitesse aux variations de paramètres électriques et l'observabilité de ces paramètres sont étudiées, permettant de définir une nouvelle stratégie d'observation des paramètres électriques. Des essais sur banc de puissance, puis en application réelle sur un train régional, valident les performances de la commande sans capteur de vitesse proposée sur toute la plage de vitesse, et son adéquation aux contraintes d'une application ferroviaire.

Nowadays, the electro-mechanical actuators (EMA) are seen as an alternative to the conventional hydraulic cylinders in a growing number of industrial applications, including the aeronautic sector, thanks to their high efficiency and relatively simple commandability. According to this features, this actuator technology was selected to replace the hydraulic cylinders of the actuation system of a full-electric compact excavator in the ELEXC project. However, some concerns arose from the lack of existing examples with EMA functioning in relatively severe working conditions implying collisions of the actuators with stiff environments. Therefore, this manuscript will first focus on the characterization of the electro-mechanical actuator and of its performances in this specific case of application, enabled by the development of a multi-energy test-rig. Beside the validation of the EMA capacities, a physic law model will be developed, and its parameters tuned according to the experimental results. In a second part of this manuscript, two novel control laws will be proposed, aiming respectively to reduce the negative repercussions of the discrete characteristic of the electrical converter, and to counterpart the lack of natural compliance of the mechanical actuator. They will be both based on the concept of predictive control, but will differ from the conventional implementation of this type of controller. The first proposed topology is a modulator based on a vector Pulse Width Modulation, in which the PWM sequence to be used is defined online by a predictive controller according to a cost function that takes into account the current ripples, the switching losses and the common mode voltages specific to the sequences. Next, a position-based impedance controller, whose impedance parameters are defined online by a predictive controller, is introduced, in order to enable the EMA to penetrate a stiff environment with a relatively low dynamic increase of compression force on the mechanical transmission., Les actionneurs électromécaniques supplantent les solutions d’actionnement hydrauliques concurrentes dans un nombre croissant d’applications industrielles, particulièrement en aéronautique. Ils le doivent à leurs rendements élevés et à une relative simplicité de commande. Fort de ces avantages, cette technologie a été mise en place dans le cadre du projet ELEXC, en vue de remplacer les vérins hydrauliques sur une mini-excavatrice tout-électrique et s’attaquant pour la première fois au secteur du bâtiment et des travaux publics. Cependant, cette intégration s’accompagne de doutes sur la performance et la durabilité de ces actionneurs dans ce cas précis d’utilisation. Pour répondre à ces interrogations, ce manuscrit se focalisera dans un premier volet sur la caractérisation des performances d’une structure spécifique, permise par la mise en place d’un banc de test multi-énergies instrumenté. Cette étude aboutira à la mise en place d’un modèle par lois physiques dont les paramètres auront été identifiés expérimentalement.Deux solutions innovantes vont ensuite être proposées, visant respectivement à réduire les répercussions, notamment énergétiques, de la caractéristique discrète du convertisseur à découpage, et à parer au manque de compliance de l’actionneur tout en conservant ses capacités de creusage. Elles seront toutes deux associées au concept de commande prédictive, mais offriront chacune une spécificité par rapport à l’implémentation standard de ce type de commande tout en conservant ses avantages. La première prendra la forme d’un modulateur fondé sur le principe d’une modulation de largeur d’impulsion vectorielle, dont la séquence sera sélectionnée par un contrôleur prédictif en rapport avec une fonction de coût alliant les ondulations de courant, les pertes par commutation et la tension de mode commun spécifiques aux séquences. La seconde consistera en un contrôleur d’impédance à boucle de position interne, dont les paramètres d’impédance seront variables et définis par un contrôleur prédictif afin de permettre une bonne pénétration dans l’environnement tout en limitant la dynamique et la valeur maximale de l’effort de compression sur la transmission.

Wake-Up Receivers (WU-RX) have been recently proposed as candidates to reduce the communication power budget of wireless networks. Their role is to sense the environment and wake up the main receivers which then handle the bulk data transfer. Existing WU-RXs achieve very high sensitivities for power consumptions below 50uW but severely degrade their performance in the presence of out-of-band blockers. We attempt to tackle this problem by implementing an ultra low power, tunable, intermediate frequency filtering stage. Its specifications are derived from standard WU-RX architectures; it is shown that classic filtering techniques are either not tunable enough or demand a power consumption beyond the total WU-RX budget of 100uW. We thus turn to the use of Continuous Time Digital Signal Processing (CT-DSP) which offers the same level of programmability as standard DSP solutions while providing an excellent scalability of the power consumption with respect to the characteristics of the input signal. A CT-DSP chain can be divided into two parts: the CT-ADC and the CT-DSP itself; the specifications of these two blocks, given the context of this work, are also discussed. The CT-ADC is based on a novel, delta modulator-based architecture which achieves a very low power consumption; its maximum operation frequency was extended by the implementation of a very fast feedback loop. Moreover, the CT nature of the ADC means that it does not do any sampling in time, hence no anti-aliasing filter is required. The proposed ADC requires only 24uW to quantize signals in the [10MHz 50MHz] bandwidth for an SNR between 32dB and 42dB, resulting in a figure of merit of 3-10fJ/conv-step, among the best reported for the selected frequency range. Finally, we present the architecture of the CT-DSP which is divided into two parts: a CT-IIR and a CT-FIR. The CT-IIR is implemented by placing a standard CT-FIR in a feedback loop around the CT-ADC. If designed correctly, the feedback loop can now cancel out certain frequencies from the CT-ADC input (corresponding to those of out-of-band interferers) while boosting the power of the useful signal. The effective amplitude of the CT-ADC input is thus reduced, making it generate a smaller number of tokens, thereby reducing the power consumption of the subsequent CT-FIR by a proportional amount. The CT-DSP consumes around 100uW while achieving more than 40dB of out-of-band rejection; for a bandpass implementation, a 2MHz passband can be shifted over the entire ADC bandwidth., La consommation des systèmes de communication pour l'IoT peut être réduite grâce à un nouveau paradigme de réception radio. La technique consiste à ajouter un récepteur supplémentaire à chaque noeud IoT, appelé Wake Up Radio (WU-RX). Le rôle du WU-RX est de surveiller le canal de communication et de réveiller le récepteur principal (aussi appelé récepteur de données) lors de la réception d'une demande de communication. Une analyse des implémentations des WU-RX existants montre que les systèmes de l'état de l'art sont suffisamment sensibles par rapport aux récepteurs de données classiques mais manquent de robustesse face aux brouilleurs. Pour améliorer cette caractéristique nous proposons un étage de filtrage accordable `a fréquence intermédiaire qui nous permet de scanner toute la bande FI en cherchant le canal utilisé pour la demande de réveil. Ce filtre a été implémenté en utilisant les principes du traitement numérique de données à temps continu et consiste en un CAN suivi par un processeur numérique à temps continu. Le principe de fonctionnement du CAN est basé sur les modulateurs delta, avec une boucle de retour améliorée qui lui permet la quantification des signaux de fréquence plus élevé pour une consommation énergétique plus faible. Par conséquent, il a une plage de fonctionnement entre 10MHz et 50MHz ; pour un SNDR entre 32dB et 42dB et une consommation de 24uW. Cela se traduit par une figure de mérite entre 3fJ/conv-step et 10fJ/conv-step, une des meilleures pour la gamme de fréquences sélectionnée. Le processeur numérique est constitué d'un filtre IIR suivi par un filtre FIR. L'atténuation hors bande apportée par le filtre IIR permet de réduire le taux d'activité vu par le filtre FIR qui, par conséquent, consomme moins d'énergie. Nous avons montré, en simulation, une réduction de la puissance consommée par le filtre FIR d'un facteur entre 2 et 3. Au total, les deux filtres atteignent plus que 40dB de réjection hors bande, avec une bande passante de 2MHz qui peut être délacée sur toute la bande passante du CAN. Dans un pire cas, le système proposé (CAN et processeur numérique) consomme moins de 100uW, cependant la configuration des signaux à l'entrée peut rendre cette consommation plus faible.

This thesis is dedicated to the control and the observation of a class of switched linear systems (SLS). This class contains the systems that can be represented by a port-Hamiltonian model. A lot of works have been studied SLS for several years using an average modeling approach. Recently, various works have shown that hybrid system theory allows to cope with stabilization, controllability, and observability problems of switched linear systems. However, several problems are still open and need more development such as the design of hybrid observers for SLS that have unobservable modes or the control of systems with reduced number of switching inputs and numerous variable states to control. In this work, we are interested in the design of state observers for a particular class of SLS using both the average and the hybrid port-Hamiltonian models. This formalism has the necessary tools to study and establish the stability of the observation errors. At the beginning, a nonlinear observer based on the average modeling is proposed. Next, a hybrid observer is designed for switched linear systems. This observer takes into account the unobservable operating modes of the system. The second point of our work concerns the design of control laws for the considered class of SLS. At first, two Lyapunov-based control laws have been established using either an average model or a hybrid model of the system. A hybrid optimal control based on the maximum principle of Pontryagin and the computation of singular arcs has been also proposed. Finally, a hybrid predictive control based on a discrete model of the system is synthesized. Simulation results and an experimental implementation on a SEPIC converter are given to show the efficiency of the proposed methods. Our motivation to study such a converter is mainly due to its particular topology that includes observable and unobservable subsystems. It is also known to be difficult to be controlled because only one switching input is used to control four state variables., Cette thèse s’intéresse à la commande et l’observation d’une classe de systèmes linéaires à commutations (SLC). La classe considérée regroupe les systèmes pouvant être représentés par un modèle Hamiltonien à ports. Récemment, plusieurs travaux ont utilisé la théorie des systèmes dynamiques hybrides pour traiter les problèmes de stabilité, de commandabilité et d’observabilité des systèmes linéaires à commutations. Cependant, certains verrous scientifiques demeurent et nécessitent d’être levés tels que la synthèse d’observateurs pour des SLC présentant des modes de fonctionnement inobservables ou la commande hybride de systèmes possédant un nombre réduit d’entrées de commutations et un nombre élevé de variables d’état à contrôler. Dans ce travail, nous nous intéressons à la synthèse d’observateurs s’appuyant sur la modélisation moyenne et la modélisation hybride de SLC ayant une topologie Hamiltonienne à ports particulière. Ce formalisme possède les outils nécessaires pour établir des preuves de stabilité des erreurs d’observation. Dans un premier temps, nous proposons un observateur non linéaire reposant sur le modèle moyen de la classe des SLC considérée. Ensuite, nous traitons le problème de synthèse d’un observateur hybride où nous proposons un observateur commuté prenant en compte les modes de fonctionnement inobservables. Le problème de la commande des SLC est abordé par la suite. Au départ, la théorie de Lyapunov est utilisée pour proposer deux lois de commandes : La première est synthétisée à partir du modèle moyen et la deuxième exploite le modèle hybride. Une commande optimale hybride est élaborée en utilisant le principe du maximum de Pontryagin et une approche utilisant la recherche d’arcs singuliers. Finalement, une commande prédictive hybride est établie à partir d’un modèle discrétisé du système. Des résultats de simulation et une mise en œuvre expérimentale sur un convertisseur DC-DC SEPIC sont donnés pour montrer l’efficacité des méthodes proposées. L’étude d’un tel circuit est motivée par sa topologie particulière qui contient à la fois un mode de fonctionnement observable et un mode de fonctionnement inobservable. En outre, il possède une seule entrée de commutations et quatre variables d’état ce qui lui vaut la réputation être difficile à commander.

The issue of sensorless control of permanent magnet synchronous machine (PMSM) has been well studied in the field of automation and electrical engineering. In the following work, we were interested in the peculiar problem of low-speed sensorless control of surface PMSM. The objective is to o er a new method for the position control of an industrial electro-mechanical actuator, considering an aeronautical context (a flap actuator). First, modeling and observability analyses of surface (non-salient) PMSM and interior (salient) PMSM have been studied. A loss of observability can be especially found at low speed for models of non-salient machines. To overcome this loss, we have developped new models, that take into account vibrations caused by a signal injection. A theoretical observability study of these latter models shows that observability is now guaranteed for the whole speed range, including standstill. We thus proposed an estimation method of the position and speed of the PMSM, based on these models, with a dedicated injection signal. Unlike sallient-based methods that are only valid for sailent machines, our approach can be used with all types of PMSM. An Extended Kalman Filter is used to observe the position, speed and eventually the load torque. The proposed approach was validated on a dedicated test bench. Many simulation and experimental tests were performed on an industrial-oriented benchmark. Results showed the performances, robustness and limitations of the proposed observers. We have also studied speed sensorless control of the machine. Indeed, two observer syntheses have been presented. The first one is based on robust synthesis of an LPV observer using a polytopic approach. The second method is an adaptative-gain second-order sliding mode observer. Experimental feasibility of the two methods has been demonstrated., Le problème de la commande sans capteur mécanique de la machine synchrone à aimants permanents (MSAP) est un problème très étudié dans le domaine de l'automatique et de l'électrotechnique. Le travail présenté s'intéresse au problème particulier de la commande sans capteur à basse vitesse des MSAP à pôles lisses. L'objectif est de proposer une méthode permettant de contrôler en position un actionneur électromécanique utilisé dans un contexte aéronautique (ici pour des ailerons). Une étude théorique des deux familles de machines, saillantes et non saillantes, a permis de montrer une différence d'observabilité des modèles de la MSAP selon la vitesse de la machine, avec en particulier une perte d'observabiltié à basse vitesse pour les machines à pôles lisses. Pour pallier cette perte, de nouveaux modèles sont développés, qui prennent en compte des vibrations de la machine, sollicitée par une injection de signaux. Une nouvelle analyse, appliquée à ces modèles, permet de garantir l'observabilité de la machine sur toute la gamme de vitesse, y compris à l'arrêt. Nous avons donc proposé une approche d'observation de la position et de la vitesse basée sur ces nouveaux modèles avec une injection bien choisie. Contrairement aux méthodes classiques basées sur la saillance, cette approche est applicable à tous les types de machine. Elle a été validée sur plusieurs bancs d'essais par l'application d'un observateur de Kalman étendu. De nombreux tests ont été réalisés sur un benchmark dédié aux applications industrielles. Les résultats ont montré les performances, la robustesse et les limites des observateurs proposés. Nous nous sommes également intéressés à l'asservissement en vitesse sans capteur mécanique de la machine, pour lequel nous avons proposé deux méthodes de synthèse d'observateurs. La première est basée sur la synthèse d'un observateur LPV robuste par approche polytopique. La seconde utilise un observateur à modes de glissement d'ordre deux à gains adaptatifs. La faisabilité expérimentale de ces deux observateurs a été démontrée.

The use of SMPS (Switched mode power supply) in embedded systems is continuously increasing. The technological requirements of these systems include simultaneously a very good voltage regulation and a strong compactness of components. SEPIC ( Single-Ended Primary Inductor Converter) is a DC/DC switching converter which possesses several advantages with regard to the other classical converters. Due to the difficulty in control of its 4th-order and non linear property, it is still not well-exploited. The objective of this work is the development of successful strategies of control for a SEPIC converter on one hand and on the other hand the effective implementation of the control algorithm developed for embedded applications (FPGA, ASIC) where the constraints of Silicon surface and the loss reduction factor are important. To do it, two non linear controls and two observers of states and load have been studied: a control and an observer based on the principle of sliding mode, a deadbeat predictive control and an Extended Kalman observer. The implementation of both control laws and the Extended Kalman observer are implemented in FPGA. An 11-bit digital PWM has been developed by combining a 4-bit Δ-Σ modulation, a 4-bit segmented DCM (Digital Clock Management) phase-shift and a 3-bit counter-comparator. All the proposed approaches are experimentally validated and constitute a good base for the integration of embedded switching mode converters, L’utilisation des alimentations à découpage (SMPSs : switched mode power supplies) est à présent largement répandue dans des systèmes embarqués en raison de leur rendement. Les exigences technologiques de ces systèmes nécessitent simultanément une très bonne régulation de tension et une forte compacité des composants. SEPIC (Single-Ended Primary Inductor Converter) est un convertisseur à découpage DC/DC qui possède plusieurs avantages par rapport à d’autres convertisseurs de structure classique. Du fait de son ordre élevé et de sa forte non linéarité, il reste encore peu exploité. L’objectif de ce travail est d’une part le développement des stratégies de commande performantes pour un convertisseur SEPIC et d’autre part l’implémentation efficace des algorithmes de commande développés pour des applications embarquées (FPGA, ASIC) où les contraintes de surface silicium et le facteur de réduction des pertes sont importantes. Pour ce faire, deux commandes non linéaires et deux observateurs augmentés (observateurs d’état et de charge) sont exploités : une commande et un observateur fondés sur le principe de mode de glissement, une commande prédictive et un observateur de Kalman étendu. L’implémentation des deux lois de commande et l’observateur de Kalman étendu sont implémentés sur FPGA. Une modulation de largeur d’impulsion (MLI) numérique à 11-bit de résolution a été développée en associant une technique de modulation Δ-Σ de 4-bit, un DCM (Digital Clock Management) segmenté et déphasé de 4-bit, et un compteur-comparateur de 3-bit. L’ensemble des approches proposées sont validées expérimentalement et constitue une bonne base pour l’intégration des convertisseurs à découpage dans les alimentations embarquées.

Owing to the development of modern semiconductor technology, it is possible to implement a digital controller for low-power high switching frequency DC-DC power converter in FPGA and ASIC. This thesis is intended to propose digital controllers with high performance, low power consumption and simple implementation architecture. Besides existing digital control-laws, such as PID, RST, tri-mode and sliding-mode (SM), a novel digital control-law, direct control with dual-state-variable prediction (DDP control), for the buck converter is proposed based on the principle of predictive control. Compared to traditional current-mode predictive control, the predictions of the inductor current and the output voltage are performed at the same time by adding a control variable to the DPWM signal. DDP control exhibits very high dynamic transient performances under both load variations and reference changes. Experimental results in FPGA verify the performances at switching frequency up to 4MHz. For the boost converter exhibiting more serious nonlinearity, linear PID and nonlinear SM controllers are designed and implemented in FPGA to verify the performances. A digital control requires a DPWM. Sigma-Delta DPWM is therefore a good candidate regarding the implementation complexity and performances. An idle-tone free condition for Sigma-Delta DPWM is considered to reduce the inherent tone-noise under DC-excitation compared to the classic approach. A guideline for Sigma-Delta DPWM helps to satisfy proposed condition. In addition, an 1-1 MASH Sigma-Delta DPWM with a feasible dither generation module is proposed to further restrain the idle-tone effect without deteriorating the closed-loop stability as well as to preserve a reasonable cost in hardware resources. The FPGA-based experimental results verify the performances of proposed DPWM in steady-state and transient-state. Two ASICs in 0.35µm CMOS process are implemented including the tri-mode controller for buck converter and the PID and SM controllers for the buck and boost converters respectively. The lab-scale tests are designed to lead to a power assessment model suggesting feasible applications. For the tri-mode controller, the measured power consumption is only 24.56mW/MHz when the time ratio of stand-by operation mode is 0.7. As specific power optimization strategies in RTL and system-level are applied to the latter chip, the measured power consumptions of the SM controllers for buck converter and boost converter are 4.46mW/MHz and 4.79mW/MHz respectively. The power consumption is foreseen as less than 1mW/MHz when the process scales down to nanometer technologies based on the power-scaling model. Compared to the state-of-the-art analog counterpart, the prototype ICs are proven to achieve comparable or even higher power efficiency for low-to-medium power applications with the benefit of better accuracy and better flexibility., Les technologies avancées de semi-conducteur permettent de mettre en œuvre un contrôleur numérique dédié aux convertisseurs à découpage, de faible puissance et de fréquence de découpage élevée sur FPGA et ASIC. Cette thèse vise à proposer des contrôleurs numériques des performances élevées, de faible consommation énergétique et qui peuvent être implémentés facilement. En plus des contrôleurs numériques existants comme PID, RST, tri-mode et par mode de glissement, un nouveau contrôleur numérique (DDP) pour le convertisseur abaisseur de tension est proposé sur le principe de la commande prédictive: il introduit une nouvelle variable de contrôle qui est la position de la largeur d'impulsion permettant de contrôler de façon simultanée le courant dans l'inductance et la tension de sortie. La solution permet une dynamique très rapide en transitoire, aussi bien pour la variation de la charge que pour les changements de tension de référence. Les résultats expérimentaux sur FPGA vérifient les performances de ce contrôleur jusqu'à la fréquence de découpage de 4MHz. Un contrôleur numérique nécessite une modulation numérique de largeur d'impulsion (DPWM). L'approche Sigma-Delta de la DPWM est un bon candidat en ce qui concerne le compromis entre la complexité et les performances. Un guide de conception d'étage Sigma-Delta pour le DPWM est présenté. Une architecture améliorée de traditionnelles 1-1 MASH Sigma-Delta DPWM est synthétisée sans détérioration de la stabilité en boucle fermée ainsi qu'en préservant un coût raisonnable en ressources matérielles. Les résultats expérimentaux sur FPGA vérifient les performances des DPWM proposées en régimes stationnaire et transitoire. Deux ASICs sont portés en CMOS 0,35µm: le contrôleur en tri-mode pour le convertisseur abaisseur de tension et la commande par mode de glissement pour les convertisseurs abaisseur et élévateur de tension. Les bancs de test sont conçus pour conduire à un modèle d'évaluation de consommation énergétique. Pour le contrôleur en tri-mode, la consommation de puissance mesurée est seulement de 24,56mW/MHz lorsque le ratio de temps en régime de repos (stand-by) est 0,7. Les consommations de puissance de command par mode de glissement pour les convertisseurs abaisseur et élévateur de tension sont respectivement de 4,46mW/MHz et 4,79mW/MHz. En utilisant le modèle de puissance, une consommation de la puissance estimée inférieure à 1mW/MHz est envisageable dans des technologies CMOS plus avancées. Comparé aux contrôlés homologues analogiques de l'état de l'art, les prototypes ASICs illustrent la possibilité d'atteindre un rendement comparable pour les applications de faible et de moyen puissance mais avec l'avantage d'une meilleure précision et une meilleure flexibilité.