Hamza Chaal

Toward a Generic Torque and Reactive Power Controller for Doubly Fed Machines

A novel shaft-position sensorless algorithm for decoupled control of torque and reactive power (TRPC) of doubly fed machines, such as the classical wound-rotor induction machine (DFIM) and the emerging brushless reluctance machine (BDFRM), has been discussed and experimentally verified in this paper. The underlying control concept is derived from first principles of magnetization and torque production in the machines. For control purposes, only the grid-connected winding measurements and rough knowledge of its resistance value are required. Such a weak parameter dependence makes the TRPC inherently robust, structurally simple, and fast to execute even on low-cost DSPs. A variety of applications are possible including drive and generator systems with limited variable speed ranges (e.g., large pumps and wind turbines), where cost savings of using partially rated power electronics are significant. Two custom-designed and built BDFRM prototypes have served as case studies to evaluate the controller performance by computer simulations and through laboratory experiments.

Practical Implementation of Sensorless Torque and Reactive Power Control of Doubly Fed Machines

A recently proposed shaft position sensorless method for torque and reactive power control of the brushless doubly fed reluctance machine has proven successful in simulation studies. In this paper, its real-time performance has been evaluated on a custom-designed machine prototype. The preliminary experimental results have confirmed the viability of the scheme and represent a serious step toward the development of a generic virtually parameter independent controller for doubly excited machines. A wide range of applications is possible, including drive and generator systems with limited speed ranges (e.g., large pumps and wind turbines), where the cost savings of using a partially rated power electronic converter are significant.

A new sensorless torque and reactive power controller for doubly-fed machines

A novel sensor-less algorithm for decoupled control of torque and reactive power of a class of doubly-fed machines like the conventional wound rotor induction machine (DFIM) or the emerging brushless reluctance machine (BDFRM), has been proposed in this paper. The control concept has been derived from the first principles of torque production and magnetisation of the machines and requires only the grid-connected winding measurements. The sensitivity analysis presented has shown an inherent robustness of the flux estimator to resistance variations making the overall control scheme practically parameter independent and structurally simple and as such suitable for implementation using low cost DSP platforms or micro-controllers. Potential target applications include variable speed drive and generator systems with limited speed ranges (e.g. large pumps and/or wind turbines) where the cost benefits of partially-rated power electronics can be fully exploited. A custom-designed BDFRM prototype has served as a case study to illustrate the good control performance through computer simulations.

Flux observer algorithms for direct torque control of Brushless Doubly-Fed Reluctance machines

Direct Torque Control (DTC) has been extensively researched and applied to most AC machines during the last two decades. Its first application to the Brushless Doubly-Fed Reluctance Machine (BDFRM), a promising cost-effective candidate for drive and generator systems with limited variable speed ranges (such as large pumps or wind turbines), has only been reported a few years ago. However, the original DTC scheme has experienced flux estimation problems and compromised performance under the maximum torque per inverter ampere (MTPIA) conditions. This deficiency at low current and torque levels may be overcome and much higher accuracy achieved by alternative estimation approaches discussed in this paper using Kalman Filter (KF) and/or Sliding Mode Observer (SMO). Computer simulations accounting for real-time constraints (e.g. measurement noise, transducer DC offset etc.) have produced realistic results similar to those one would expect from an experimental setup.

Sliding Mode Observer based direct torque control of a Brushless Doubly-Fed Reluctance Machine

Direct Torque Control (DTC) has been extensively researched and applied during the last two decades. However, it was applied to the Brushless Doubly-Fed Reluctance Machine (BDFRM) for the first time only a few years ago in its basic form inheriting its intrinsic flux estimation problems that propagate throughout the algorithm and hence compromise the DTC performance. In this paper, we propose the use of Sliding Mode Observer (SMO) as an alternative to improve the estimation quality and consequently the control performance of the DTC. The SMO is designed around a nominal model, but is shown to be reliable over the whole operating range of the BDFRM. Moreover, we use a modified robust exact differentiator based on Sliding Mode (SM) techniques to calculate the angular velocity from an angular position encoder. Computer simulations are meticulously designed to take into account real-world physical constraints and thus show illustrative supporting results as expected from an experimental setup.

Second Order Sliding Mode Control of a DC drive with uncertain parameters and load conditions

The Higher Order Sliding Mode (HOSM) is considered as a generalization of standard sliding modes that attenuates the chattering phenomenon. The main problem in implementation of the HOSM is the increasing information demand. In this work we implement two recent non-linear robust control schemes on a DC drive emphasizing their ability to tackle realistic situations in which the drive parameters are subject to variation, in addition to a time varying load constraint. Namely, we consider the Twisting algorithm which requires the first time derivative of the sliding variable to be known. In addition to that, we use the Super Twisting algorithm, which unlike other HOSM algorithms merely requires measurements of the sliding variable. Simulation results are given to support the approach.