Low Voltage Ride-through in DFIG Wind Generators by Controlling the Rotor Current without Crowbars

Among all the different types of electric wind generators, those that are based on doubly fed induction generators, or DFIG technology, are the most vulnerable to grid faults such as voltage sags. This paper proposes a new control strategy for this type of wind generator, that allows these devices to withstand the effects of a voltage sag while following the new requirements imposed by grid operators. This new control strategy makes the use of complementary devices such as crowbars unnecessary, as it greatly reduces the value of currents originated by the fault. This ensures less costly designs for the rotor systems as well as a more economic sizing of the necessary power electronics. The strategy described here uses an electric generator model based on space-phasor theory that provides a direct control over the position of the rotor magnetic flux. Controlling the rotor magnetic flux has a direct influence on the rest of the electrical variables enabling the machine to evolve to a desired work point during the transient imposed by the grid disturbance. Simulation studies have been carried out, as well as test bench trials, in order to prove the viability and functionality of the proposed control strategy.

Plataforma de simulación en Matlab-Simulink de un accionamiento regulado para emular aerogeneradores

En este proyecto se desarrolla un modelo de simulación de un accionamiento controlado que emula el comportamiento de una turbina eólica, el cual se ha llevado a cabo a través del programa para simulación Matlab/Simulink. Su desarrollo se ha estructurado de la siguiente forma: Tras una breve introducción a la energía eólica y a las máquinas eléctricas objeto de estudio en este proyecto, se procede a la caracterización y representación de dichas maquinas dentro de la plataforma de simulación virtual Simulink. Posteriormente se explican posibles estrategias de control de la máquina de inducción, las cuales son aplicadas para la realización de un control de velocidad. Asimismo, se realiza un control vectorial de par de la máquina de inducción de modo que permita un seguimiento efectivo del par de referencia demandado por el usuario, ante distintas condiciones. Finalmente, se añade el modelo de turbina eólica de manera que, definiendo los valores de velocidad de viento, ángulo de paso y velocidad del eje, permite evaluar el par mecánico desarrollado por la turbina. Este modelo se valida comprobando su funcionamiento para diferentes puntos de operación ante diversas condiciones del par de carga. Las condiciones de carga se establecen acoplando al modelo de la turbina, un generador síncrono de imanes permanentes conectado a una carga resistiva. ! II! ABSTRACT In this project, the simulation model of a controlled drive that emulates the behaviour of a wind turbine is developed. It has been carried out through the platform for multidomian simulation called Matlab/Simulink. Its development has been structured as follows: After a brief introduction to the wind energy and the electrical machines studied in this project, these machines are characterized and represented into the virtual simulation platform, Simulink. Subsequently, the possible control strategies for the induction machine are explained and applied in order to carry out a speed control. Additionally, a torque vector control of the induction machine is performed, so as to enable an effective monitoring of the reference torque requested by the user, under different conditions. Finally, the wind turbine model is implemented so as to assess the turbine mechanical torque, after defining the wind speed, the pitch angle and the shaft speed values. This model is validated by testing its functionality for different operating points under various load torques. The load conditions are set up by attaching a permanent magnets synchronous machine, with a resistive load, to the turbine model.

Regularization for sparsity in statistical analysis and machine learning

Pragmatism is the leading motivation of regularization. We can understand regularization as a modification of the maximum-likelihood estimator so that a reasonable answer could be given in an unstable or ill-posed situation. To mention some typical examples, this happens when fitting parametric or non-parametric models with more parameters than data or when estimating large covariance matrices. Regularization is usually used, in addition, to improve the bias-variance tradeoff of an estimation. Then, the definition of regularization is quite general, and, although the introduction of a penalty is probably the most popular type, it is just one out of multiple forms of regularization. In this dissertation, we focus on the applications of regularization for obtaining sparse or parsimonious representations, where only a subset of the inputs is used. A particular form of regularization, L1-regularization, plays a key role for reaching sparsity. Most of the contributions presented here revolve around L1-regularization, although other forms of regularization are explored (also pursuing sparsity in some sense). In addition to present a compact review of L1-regularization and its applications in statistical and machine learning, we devise methodology for regression, supervised classification and structure induction of graphical models. Within the regression paradigm, we focus on kernel smoothing learning, proposing techniques for kernel design that are suitable for high dimensional settings and sparse regression functions. We also present an application of regularized regression techniques for modeling the response of biological neurons. Supervised classification advances deal, on the one hand, with the application of regularization for obtaining a na¨ıve Bayes classifier and, on the other hand, with a novel algorithm for brain-computer interface design that uses group regularization in an efficient manner. Finally, we present a heuristic for inducing structures of Gaussian Bayesian networks using L1-regularization as a filter. El pragmatismo es la principal motivación de la regularización. Podemos entender la regularización como una modificación del estimador de máxima verosimilitud, de tal manera que se pueda dar una respuesta cuando la configuración del problema es inestable. A modo de ejemplo, podemos mencionar el ajuste de modelos paramétricos o no paramétricos cuando hay más parámetros que casos en el conjunto de datos, o la estimación de grandes matrices de covarianzas. Se suele recurrir a la regularización, además, para mejorar el compromiso sesgo-varianza en una estimación. Por tanto, la definición de regularización es muy general y, aunque la introducción de una función de penalización es probablemente el método más popular, éste es sólo uno de entre varias posibilidades. En esta tesis se ha trabajado en aplicaciones de regularización para obtener representaciones dispersas, donde sólo se usa un subconjunto de las entradas. En particular, la regularización L1 juega un papel clave en la búsqueda de dicha dispersión. La mayor parte de las contribuciones presentadas en la tesis giran alrededor de la regularización L1, aunque también se exploran otras formas de regularización (que igualmente persiguen un modelo disperso). Además de presentar una revisión de la regularización L1 y sus aplicaciones en estadística y aprendizaje de máquina, se ha desarrollado metodología para regresión, clasificación supervisada y aprendizaje de estructura en modelos gráficos. Dentro de la regresión, se ha trabajado principalmente en métodos de regresión local, proponiendo técnicas de diseño del kernel que sean adecuadas a configuraciones de alta dimensionalidad y funciones de regresión dispersas. También se presenta una aplicación de las técnicas de regresión regularizada para modelar la respuesta de neuronas reales. Los avances en clasificación supervisada tratan, por una parte, con el uso de regularización para obtener un clasificador naive Bayes y, por otra parte, con el desarrollo de un algoritmo que usa regularización por grupos de una manera eficiente y que se ha aplicado al diseño de interfaces cerebromáquina. Finalmente, se presenta una heurística para inducir la estructura de redes Bayesianas Gaussianas usando regularización L1 a modo de filtro.

Reproductive long-term effects, endocrine response and fatty acid profile of rabbit does fed diets supplemented with n-3 fatty acids

The effect of a diet enriched with polyunsaturated n -3 fatty acids (PUFA) on endocrine, reproductive, and productive responses of rabbit females and the litters has been studied. Nulliparous does ( n = 125) were fed ad libitum from rearing to second weaning two diets supplemented with different fat sources: 7.5 g/kg lard for the control diet (group C; n = 63) or 15 g/kg of a commercial supplement containing a 50% ether extract and 35% of total fatty acids (FAs) as PUFA n -3 (Group P; n = 62). Dietary treatments did not affect apparent digestibility coefficients of nutrients, or reproductive variables of does including milk pro- duction, mortality and average daily gain of kits over two lactations. However, on Day 5 and 7 post-induction of ovulation, progesterone of Group P tended to increase to a greater extent than in does of Group C. Total PUFAs, n -6 and n -3 and eicosapentanoic (EPA) contents were greater in adipose tissues of does in Group P than in Group C. Docosapentaenoic acid (DPA), EPA, and docosahexaenoic acid (DHA) concentrations were greater in peri-ovarian than in scapular fat with abdominal fat being intermediate in concentration. In PUFA sup- plemented does, kit mortality at the second parturition tended to be less than in control does. Also, kits born to does of the PUFA-supplemented group weighed more and were of greater length than from does of control group. In conclusion, effectiveness of dietary intervention on reproductive and performance response is greater in the second parity, which suggests an accumulative long-term beneficial effect of n -3 FA supplementation in reproductive rabbit does