La commande sans capteur mécanique de la machine synchrone à aimants permanents est un sujet largement répandu dont les plus grandes difficultés connues, quel que soit l’observateur utilisé, sont celui du démarrage à vitesse nulle et plus largement de la commande aux basses vitesses, et celui du rejet des fortes perturbations du couple. Afin d’y faire obstacle, diverses adaptations des algorithmes d’observateur ont été proposées afin d’améliorer le comportement de la machine en commande sans capteur. Par ailleurs, il est couramment admis que les déchets de tension produit par l’onduleur sont nuisibles à l’observation de la position, les tensions de référence étant légèrement différentes des tensions appliquées aux enroulements de la machine. Quelques propositions apparaissent dans certaines publications notamment en établissant un algorithme de compensation. C’est ainsi que les travaux de cette thèse portent sur cette thématique, celle d’améliorer la commande sans capteur dans les situations d’observabilité difficile en proposant une solution originale afin de faire correspondre au mieux les tensions appliquées à la machine et les tensions de référence utiles à l’observateur. Les résultats montrent que la solution proposée et largement analysée améliore considérablement le comportement de la machine en commande aux basses vitesses et en rejet de perturbation, tant en régime permanent qu’en régime transitoire ; une analyse de Fourier des courants mesurés atteste l’efficacité de la méthode et une analyse des grandeurs observées par la statistique descriptive met en lumière l’intérêt de l’algorithme. Nous montrons ainsi que la solution proposée permet d’observer la vitesse et la position en deçà de la vitesse mécanique de 15 rad/s alors que la commande est instable en deçà de 20 rad/s quand la solution n’est pas mise en œuvre. Un constat similaire apparaît en rejet de perturbation. D’autres résultats montrent que l’observation à plus basse vitesse est entachée d’une perturbation liée à un couple pulsatoire dont l’origine peut être le couple de détente, lequel n’est pas pris en compte par le modèle de la machine. / The sensorless control of the permanent magnet synchronous machine is a subject widely spread. Two great difficulties are known; (i) the start at nil initial speed and more generally the control at very low speed whatever the observer used and (ii) the high torque disturbance rejection. In order to hinder these difficulties numerous modifications of observer algorithms were proposed to improve the performances of the permanent magnet synchronous machine sensorless control. Moreover, we admit commonly that the drop voltages due to the inverter are prejudicial to the position estimated, because the difference between the voltage reference transmitted to the PWM (pulse width modulation) and the motor winding voltage is not negligible at low speed and low load torque. According to the literature, several papers propose some solutions by compensation algorithms and voltage estimator in particular. So, the goal of this thesis is to estimate the winding voltage and to apply the state observer by Extended Kalman Filtering to improve more finely the sensorless control. We propose an original solution to estimate the voltage references which is applied to the observer. Numerous experimental results show the attractive effects in marked contrast to the sensorless control results without estimation of the winding voltages. The results of sensorless control show that the solution proposed which widely analysed improves significantly the estimation errors of the motor running in low speed range and low torque disturbances range. Fourier analyses and statistic data obtained in steady state speed and results during the transient response indicate complementary results and highlight the interest of the estimation algorithm. Our study brings out that the estimation error reduction allows to running the motor at mechanical speed short of 15 rad/s. In the other hand, the system is instable with speed short of 20 rad/s if the voltage references are used by the observer rather than the estimation voltages. The same improvement appears in disturbance rejection. Other results show that the estimated position errors at lower speed increases in spite of the estimation algorithm. In fact, the torque disturbances are dominant at low speed, low load torque and are harmful to control the electromagnetic torque.

Switching from fossil-fuel-powered vehicles to electric vehicles has become an international focus in the pursuit of combatting climate change. Regardless, the adoption of electric vehicles has been slow, in part, due to range anxiety. One solution to mitigating range anxiety is to provide a more accurate state of charge (SOC) and range estimation. SOC estimation of lithium-ion batteries for electric vehicle application is a well-researched topic, yet minimal tools and code exist online for researchers and students alike. To that end, a publicly available Kalman filter-based SOC estimation function is presented. The MATLAB function utilizes a second-order resistor-capacitor equivalent circuit model. It requires the SOC-OCV (open circuit voltage) curve, internal resistance, and equivalent circuit model battery parameters. Users can use an extended Kalman filter (EKF) or adaptive extended Kalman filter (AEKF) algorithm and temperature-dependent battery data. A practical example is illustrated using the LA92 driving cycle of a Turnigy battery at multiple temperatures ranging from -10C to 40C. Current range estimation methods suffer from inaccuracy as factors including temperature, wind, driver behaviour, battery voltage, current, SOC, route/terrain, and much more make it difficult to model accurately. One of the most critical factors in range estimation is the battery. However, most models thus far are represented using equivalent circuit models as they are more widely researched. Another limitation is that any machine learning-based range estimation is typically based on historical driving data that require odometer readings for training. A range estimation algorithm using a machine learning-based voltage estimation model is presented. Specifically, the long short-term memory cell in a recurrent neural network is used for the battery model. The model is trained with two datasets, classic and whole, from the experimental data of four Tesla/Panasonic 2170 battery cells. All network training is completed on SHARCNET, a resource provided by Canada Compute to researchers. The classically trained network achieved an average root mean squared error (RMSE) of 44 mV compared to 34 mV achieved by the network trained on the whole dataset. Based on the whole dataset, all test cases achieve an end range estimation of less than 5 km with an average of 0.29 km. / Thesis / Master of Applied Science (MASc)