Sébastien Mariéthoz

Explicit Model-Predictive Control of a PWM Inverter With an LCL Filter

This paper deals with the control of pulsewidth modulation inverters connected to the grid through resonant LCL filters. It proposes two alternative (piecewise affine) models that account for the switched behavior of the converter. Based on these improved models, an explicit model-predictive control scheme is derived in order to provide a fast response, making it very suitable for applications, such as active filtering, where a large bandwidth is required. A state observer and a grid voltage estimator are used in order to reduce the number of required sensors and to eliminate noise. The control scheme relies only on filtered current measurements and on the DC voltage.

Explicit Model Predictive Control of DC–DC Switched-Mode Power Supplies With Extended Kalman Filtering

This paper presents a sensorless explicit model predictive control scheme for the dc-dc boost converter. No direct inductor current measurement is needed as the coil current is derived either via a static approximation or, for improved accuracy, through an extended Kalman filter. The estimate is used in the chosen optimal control problem formulation which yields the optimal input by intrinsically accounting for duty cycle and current constraints. The optimization problem is explicitly presolved offline so that the online effort is reduced to a simple search in the resulting lookup table. No online optimization is required, greatly facilitating physical implementation and allowing for experimental validation on an integrated dc-dc converter through a fixed-point DSP.

Comparison of Hybrid Control Techniques for Buck and Boost DC-DC Converters

Five recent techniques from hybrid and optimal control are evaluated on two power electronics benchmark problems. The benchmarks involve a number of practically interesting operating scenarios for fixed-frequency synchronous dc-dc converters. The specifications are defined such that good performance can only be obtained if the switched and nonlinear nature of the problem is accounted for during the design phase. A nonlinear action is featured in all methods either intrinsically or as external logic. The designs are evaluated and compared on the same experimental platform. Experiments show that the proposed methods display high performances, while respecting circuit constraints, thus protecting the semiconductor devices. Moreover, the complexity of the controllers is compatible with the high-frequency requirements of the considered application.

New configurations for the three-phase asymmetrical multilevel inverter

This paper deals with cascade multilevel inverters. It focuses on asymmetrical topologies where the cell input voltages are of different values. These hybrid topologies might be used in several applications. Nevertheless, the need of DC-DC converters to supply the cells of reversible multilevel converters increases the cost and losses of such inverters. It limits their application field. Furthermore, the simultaneous-commutation problem, which increases the switching losses for some operating points, reduces the design choice to configurations of lower resolution. Combining in series a 3-phase 6-switch voltage source inverter with single-phase H-bridges, we obtain an asymmetrical hybrid cascaded multilevel inverter with advantageous properties in terms of voltage resolution and energetic efficiency. This paper describes the design and use of such a structure, the way to increase the asymmetrical multilevel inverters resolution.

Policy-Based Reserves for Power Systems

This paper introduces the concept of affine reserve policies for accommodating large, fluctuating renewable in feeds in power systems. The approach uses robust optimization with recourse to determine operating rules for power system entities such as generators and storage units. These rules, or policies, establish several hours in advance how these entities are to respond to errors in the prediction of loads and renewable infeeds once their values are discovered. Affine policies consist of a nominal power schedule plus a series of planned linear modifications that depend on the prediction errors that will become known at future times. We describe how to choose optimal affine policies that respect the power network constraints, namely matching supply and demand, respecting transmission line ratings, and the local operating limits of power system entities, for all realizations of the prediction errors. Crucially, these policies are time-coupled, exploiting the spatial and temporal correlation of these prediction errors. Affine policies are compared with existing reserve operation under standard modeling assumptions, and operating cost reductions are reported for a multi-day benchmark study featuring a poorly-predicted wind infeed. Efficient prices for such “policy-based reserves” are derived, and we propose new reserve products that could be traded on electricity markets.

Open loop and closed loop spectral frequency active filtering

This paper proposes two frequency based methods for parallel active filtering of current harmonics in utility grids. Active filter (AF) reference prediction and AF current control are the two key-points to obtain high performances. This paper deals with the AF reference prediction. The two proposed methods are based on the Fourier series and its implementation on DSP. They belong to two prediction structures: (1) the open-loop structure which predicts the reference currents from the load currents; and (2) the closed-loop structure which predicts the reference currents from the grid currents. Fourier series based methods are very flexible and have several interesting properties. These two methods have been successfully implemented and tested. The latter method cancels permanent errors and results in better performances. Both methods are very flexible.

High-speed online MPC based on a fast gradient method applied to power converter control

Bounding the computational complexity of an online optimization method in a real-time environment with hard time constraints is a challenging problem. This paper investigates a new solution approach based on a fast gradient method in the context of model predictive control (MPC) of power converters. Different from other solution methods that either provide bounds that are far off from the practically observed ones or do not allow for bounding the computational effort at all this method enables easy to compute and meaningful bounds that can further be decreased by means of a pre-conditioning technique.We report an implementation of the fast gradient method on an industrial-type digital signal processor with integer arithmetics and show that worst case runtimes are in the order of tens of μs using less than one kByte of memory while being numerically robust. Moreover, this method also improves the control performance compared to explicit MPC.

Design and control of asymmetrical multi-level inverters

This paper deals with asymmetrical multi-level inverters. The investigated topologies consist of series connected cells with different input voltages. Some combinations of different input voltages have been proposed to improve the converter resolution. This paper has a general approach. It describes the rules to obtain a uniform step with these topologies. It points out the switching problem of some of these asymmetrical solutions, describes the rules to avoid this problem, resulting in a solution where the number of commutations and hence the switching losses may be reduced. For this purpose, it also proposes an adapted optimal control strategy based on a geometrical approach, and a scheme for the control of multi-level inverters.

Asymmetrical Cascade Multilevel Converters With Noninteger or Dynamically Changing DC Voltage Ratios: Concepts and Modulation Techniques

Asymmetrical cascade multilevel inverters offer a high number of voltage levels with a given switch count. For a given topology, the number of levels depends on the configuration of the dc voltage (ratios). This paper deals with the design and control of such converters with noninteger dc voltage ratio, which leads to unevenly distributed space vectors. It describes how to select the dc voltage ratio configurations that yield space vectors as evenly distributed as possible. It describes how to produce an effective PWM modulation that allows generating undistorted current even in the presence of (some) unevenly distributed space vectors.

High-Bandwidth Explicit Model Predictive Control of Electrical Drives

Field-oriented control (FOC) has proven effective for controlling ac drives with good dynamic performance. However, operation at low-switching frequencies and the sensitivity of traditional feedforward loops to system parameters pose severe limitations on the achievable performance and require a tedious tuning procedure. In this paper, we present a systematic cascade explicit model predictive control framework for the FOC of electrical drives, resolving the aforementioned issues while being sufficiently simple to be widely implemented on various ac drive systems. The resulting closed-loop system exhibits high dynamic performance for all operating points, even at low-switching frequencies. We present experiments with a permanent-magnet machine and an induction motor, demonstrating the practical feasibility and the merits of the proposed framework over traditional controller designs for electrical drives.