IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 1
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS1Model Predictive Controlin the Multi-Megawatt RangeThomas J. Besselmann, Sture Van de moortel, Stefan Almér, Pieder Jörg and Hans Joachim FerreauAbstract—The paper at hand presents an application of modelpredictive control to a variable speed drive system operatingin the multi-megawatt range. The variable speed drive systemcomprises a synchronous machine fed by a line commutatedrectifier and a load commutated inverter. The control task is toregulate the DC link current, and hence the machine torque,to ensure the machine speed follows a given reference. Theproposed control approach is model predictive control whereboth the rectifier and inverter firing angles are considered ascontrol inputs. The nonlinear model predictive torque controllerhas been implemented on an embedded system and applied inan industrial-scale pilot plant installation. The experiments showthe successful operation of model predictive control on a plantwith more than 48 MW power.Index Terms—Load commutated inverter, Predictive control,High-power application.I. I NTRODUCTIONModel predictive control (MPC) [1], [2] is by now astandard solution in many applications with relatively slowsystem dynamics and ample computational resources. However, recent advances in optimization algorithms and computational hardware have enabled the use of MPC also inapplications where sampling rates are higher and the controlleris implemented on embedded platforms. In particular, MPChas been successfully applied for control of power convertersand electric machines where the control frequency ranges fromkilohertz to megahertz, see e.g., [3]–[11].The literature on MPC of power electronics is most oftenlimited to either simulations or to laboratory-scale experimental setups. In contrast, the paper at hand reports anapplication of high-speed MPC to an industrial-scale pilotplant installation, with a variable-speed drive providing a gascompressor with power in excess of 48 MW.The focus of most, if not all, previous research on MPCof power electronics has been on voltage source convertertopologies. In the present paper we consider a synchronousmachine fed by a load-commutated current source converter.To the best of our knowledge, MPC has not been applied tothis type of system prior to the presented line of research.Manuscript received May 22, 2015; revised September 10, 2015 andOctober 8, 2015; accepted November 1, 2015.Copyright 2015 IEEE. Personal use of this material is permitted.However, permission to use this material for any other purposes must beobtained from the IEEE by sending a request to pubs-permissions@ieee.orgCorresponding author: T. Besselmann (phone: 41 58 586 74 40, e-mail:thomas.besselmann@ch.abb.com).T. J. Besselmann, S. Almér and H. J. Ferreau are with ABB Corporate Research, Segelhofstrasse 1K, 5405 Baden-Dättwil, Aargau, Switzerland. S. Vande moortel and P. Jörg are with ABB Medium-Voltage Drives, Austrasse, 5300Turgi, Aargau, Switzerland (e-mail:{thomas.besselmann, sture.vandemoortel,stefan.almer, pieder.joerg, joachim.ferreau}@ch.abb.com).(C) IEEE - DOI: 10.1109/TIE.2015.2504346The paper considers a variable speed synchronous machineconnected to the grid via a line commutated rectifier and aload commutated inverter (LCI) [12]. This type of variablespeed solution is often the preferred choice in high powerapplications, ranging from a few megawatts to over a hundredmegawatts, [13], [14]. Such applications include high speedcompressors and rolling mills.For the variable speed system at hand, we consider thedesign of a model predictive torque controller. The MPCconsiders both the rectifier and inverter firing angles as controlinputs and minimizes the deviation of the DC link currentfrom the reference while respecting constraints on the stateand control inputs. By controlling the DC current, the machinetorque is controlled indirectly to its reference.The work presented here is motivated mainly by electricallydriven gas compression plants which are often situated inremote locations and operate under weak grid conditions.Weather phenomena occasionally produce sudden sags of thegrid voltage, which can cause the drive to trip, interrupting oreven aborting the gas compression process. The goal of thisline of research is to design a more agile torque controllerto increase the system robustness to external disturbances.In particular, we want to improve the ability to reliably ridethrough power loss situations due to grid faults and delivertorque during partial loss of grid voltage, [15].Conventional PI-based control approaches typically assigndifferent tasks to the two control inputs (i.e., the rectifier andinverter firing angles). The inverter angle is set to determinethe power factor of the machine, whereas the rectifier angleis used to control the DC link current. The fact that the MPCcontrols the rectifier and inverter angles without pre-assigningtasks to them implies a potential for better disturbance rejection. In particular, in the case of a disturbance of the grid, thePI approach would only adjust the rectifier angle while theMPC would adjust both firing angles.Implementing the model predictive controller requires tosolve a constrained nonlinear, nonconvex optimization problem in real-time. This is a challenging task as our applicationrequires a sampling time of one millisecond and the embeddedcomputing power is limited. Solving nonlinear MPC problemsin such a situation requires both a careful problem formulationand highly efficient, state-of-the-art optimization algorithms.In this paper, we follow the promising approach of autogenerating customized nonlinear MPC algorithms that aretailored to the problem at hand based on a symbolic problemformulation as proposed in [16].Preceding work in this line of research includes [17]. Thepaper at hand differs from the previous publication (a) in
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICSurec,12uinv,1θrThe model describes the average behavior of the switchedsystem. For the sake of simplicity, we neglect phenomena suchas commutation overlap, thyristor recharge time, forced commutation at low speeds, asymmetric grid conditions, inductivevoltage losses at the transformer or intermittent operation atlow DC currents.ωrurec,2uinv,2vfFig. 1. Variable speed drive system comprised of line commutated rectifier,inductive DC link, load commutated inverter and synchronous machine.its theoretical content by reformulating the torque controlproblem into a DC current control problem in order to simplifythe optimization problem at hand; and (b) in its practicalcontent by providing experimental data from tests on anindustrial-scale pilot installation.The paper is organized as follows. In Section II we describethe synchronous machine and the load-commutated inverter.Then a mathematical model of this system is presented inSection III. The developed control solution including theMPC current controller is described in Section IV. Section Vcontains the experimental results. Finally, conclusions aredrawn in Section VI. Note that all quantities in this paperare normalized quantities.II. C URRENT S OURCE C ONVERTERS AND S YNCHRONOUSM ACHINEThe paper considers a variable speed drive system composedof a line commutated rectifier, inductive DC link, load commutated inverter and a synchronous machine, see Fig. 1. Thistype of drive system is suitable for high power applicationsranging from a few megawatts to over a hundred megawatts.Such applications include high speed compressors and rollingmills [18].In the considered configuration the rectifier and inverterconsist of twelve-pulse thyristor bridges, each comprisingtwo six-pulse bridges. However, the proposed control schemecan easily be adapted to other configurations, such as sixpulse bridges and poly-phase synchronous machines as wasconsidered in [17]. A thyristor bridge can operate as rectifieror as inverter, depending on the choice of the firing angle. Inthe context of this paper we will follow the common notationto denote the line side converter as rectifier and the machineside converter as inverter.The control inputs (signals to be manipulated by the controller) are the firing angle α of the line side rectifier and firingangle β of the machine side inverter. Furthermore, the statorvoltage magnitude us is controlled by means of an excitationvoltage vf . The variable to be controlled is the air gap torqueproduced by the synchronous machine.III. P REDICTION M ODELThe model predictive controller is based on the dynamicmodel of the DC link and thyristor bridges developed below.A. Control Input, State and ParametersIn deriving a dynamic model of the system suitable fortorque control, we first decide which system quantities tomodel as states and which to consider as (slowly varying)parameters. Certain quantities are assumed to vary sufficientlyslow to be approximated as constant when regulating thetorque and are therefore considered as parameters. Thesequantities are the line voltage amplitude ul and the machinevoltage amplitude us . The state of the system consists of theDC link current idc .The rotor excitation flux varies considerably slower thanthe DC link current idc . We therefore control the excitationof the machine with a slower outer loop and the design ofthis control loop is not discussed in this paper. The controlvariable (excitation voltage) vf discussed above is thereforenot considered in the sequel. Thus, the control input is therectifier firing angle α and the inverter firing angle β.B. DC Link DynamicsThe DC link current dynamics are described by 1d rdc idc urec,1 urec,2 uinv,1 uinv,2 ,idc dtLdcwhere Ldc , rdc are the inductance and parasitic resistance ofthe DC link inductor and where urec,1 , urec,2 and uinv,1 , uinv,2are the DC voltage of each thyristor bridge of the rectifier andthe inverter, respectively.C. Thyristor Bridge DC VoltageThe DC side voltage of the six-pulse thyristor bridgesin Fig. 1 is a switched waveform which is constructed byswitching between the AC side line-to-line voltages. Theprinciple is illustrated in Fig. 2 where the sinusoids representthe line-to-line voltages on the AC side and where the thicklines illustrate the DC side voltage for a few different valuesof the firing angle which ranges from 0 to 180 degree. Thefiring angle determines the time instant of the switch fromone line-to-line voltage to another and this determines theaverage value of the DC side voltage. For a firing angle of 0degree the thyristor bridge operates identical to a diode bridge,where the instant value of the line-to-line voltages determineswhich diodes are conducting. Larger firing angles representthe time delay of the thyristor bridge switchings compared tothe switchings of a diode bridge.For the purpose of control, we describe the average value ofthe switched DC side voltage as a function of the firing angle.The thyristor bridge DC voltage is approximated by a cosineof the firing angle as illustrated in Fig. 3. The approximationis intuitively clear considering the waveforms in Fig. 2. Aderivation can be found in [19].
3voltage [pu]1current, voltage [pu]IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS0 180 0 10120240360stator curfundamentalstator voltβ10 10120240360480600electrical angle [deg]voltage [pu]130 Fig. 4. AC side approximation: Stator current and its fundamental, and statorvoltage of the synchronous machine. The power factor is determined by theangle β between current and voltage.150 0 10120240360voltage [pu]160 120 0 1012024036090 voltage [pu]10 10120240360electrical angle [deg]Fig. 2. AC and DC side voltages of a six-pulse thyristor bridge over oneperiod of the AC side voltage. The thin lines show the line-to-line voltageson the AC side. The thick lines show the switched voltage of the DC side fordifferent values of the firing angle.voltage [pu]1E. Torque ExpressionThe MPC problem formulation penalizes the deviation ofthe torque from a given reference and we therefore need anexpression for the torque.By scaling the nominal values accordingly, the electricpower at the stator and the mechanical power at the shaft canbe stated as 16090120150180firing angle [deg]Fig. 3. DC side approximation: Approximate relation between AC and DCvoltages of a thyristor bridge. The DC side voltage is approximated by acosine of the firing angle.By applying the same firing angle to both bridges of theconverters and considering the sum of the DC side voltages,and by neglecting the switching and commutation intervals wehaveurec k1 ul cos(α),Pm τe ωr ,respectively. On average, and by ignoring the losses in themachine, the power balance holds, i.e. Pm Pel , such thatinverter mode30D. Thyristor Bridge AC CurrentThe AC side inverter current is illustrated in Fig. 4. Theideal waveform (neglecting commutation time [21]) is piecewise constant. For the purpose of control, the AC currentis approximated by its fundamental component, which isillustrated by the dashed line in Fig. 4.The modulator of the inverter, which takes the firing angle βand controls the switching, places the stator current at theangle β to the stator voltage and thus determines the powerfactor of the machine.Pel us idc cos(β),rectifier mode00We note that the thyristors can be turned on at any time, butthey can only be turned off by reducing the current runningthrough them to zero. Thus, the off-switching of the thyristorsis state dependent. This is neglected in the control model.uinv k1 us cos(β) ,where urec and uinv are the combined DC side voltages of theline side and the machine side thyristor bridges, respectively,k1 is a constant, ul is the amplitude of the AC side voltage ofthe rectifier and us is the amplitude of the stator voltage, [20].τe us idc cos(β)/ωr .Moreover, if the excitation controller adapts the excitation fluxsuch that us ωr , the torque expression simplifies toτe idc cos(β).F. Model SummaryBy defining the line side and machine side power factors asauxiliary control variables uα : cos(α) and uβ : cos(β),the prediction model can be stated asd1( rdc idc ul k1 uα us k1 uβ ) ,(1a)idc dtLdcτe idc uβ ,(1b)with the time-varying parameters ul and us . By further replacing the torque as controlled variable by the current andthe power factor, the prediction model simplifies to a onedimensional linear parameter-varying system.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS4 cω̂rulωr SpeedControlt eReferenceGovernor i dc , α , βMPCusExcitationControlk·kω̂rFig. 5.Overview of the proposed control solution for load commutated inverter-fed synchronous machines.IV. P ROPOSED C ONTROL S OLUTIONA simplified block diagram of the proposed control solutionis shown in Fig. 5. Parts of the solution are state of the artand have been discussed in previous work, see e.g. [19], [22].For brevity we thus focus on the innovative part of the controlsystem, being the reference governor and the model predictivecurrent controller.A. Reference GovernorThe task of the reference governor is to transform the torquereference τe into references i dc , u α and u β for the DC current,the line side and the machine side power factor, respectively.Limitations on the eligible machine side firing angles,βmin β βmax ,correspond to the auxiliary limitationsuβ,min : cos(βmax ) uβ cos(βmin ) : uβ,max .In order to minimize the reactive power in the machine, thepower factor reference is set as large as possible, i.e. uβ,min , if τe ωr 0 , (motoring) uβ uβ,max , if τe ωr 0 . (generating)Respecting some limitations on the permissible DC current0 idc idc,max ,the DC current reference is chosen according to equation (1b)as if τe /u β idc,max , idc,max , τe /uβ , if 0 τe /u β idc,max ,idc 0,if τe /u β 0 .Finally, a reference for the line side firing angle, α can bedetermined by means of the steady-state relation stemmingfrom the prediction model (1a) 1rdc i dc us k1 u β .u α ul k1B. Model Predictive Current ControllerAt each sampling time, the model predictive controllertakes an estimate of the system state as initial condition andminimizes a finite time horizon cost integral subject to thedynamic constraints of the system and constraints on the stateand input. The cost criterion is T Z kTs Tpuα u αuα u α 2dt ,RQ(idc idc ) J : uβ u βuβ u βkTs(2)where k is the sampling instance, Ts is the sampling periodand Tp is the prediction horizon length.Model predictive control allows for the intuitive integrationof constraints on inputs, states and outputs. In the applicationat hand we limit the eligible firing angles and request an upperbound on the DC current,uα,min uα uα,max ,idc idc,max ,uβ,min uβ uβ,max ,(3a)(3b)for some application-dependent bounds. The time-continuousoptimal control problem can thus be stated asmin (2) s.t.uα ,uβ(1a), (3) .(4)In order to solve the optimal control problem (4) numerically, we follow a so-called “direct” approach (see e.g., [23]for a detailled discussion). This means that the optimal controlproblem is first discretized in time to yield a finite-dimensionalnonlinear programming (NLP) problem, which can then betackled by an appropriate optimization algorithm.If the dynamic model would be linear, one would need todiscretize problem (4) only once before the actual runtime ofthe controller. Moreover, in that case the resulting NLP wouldactually be a quadratic programming (QP) problem, such thatthe only computational effort to be performed on-line wouldbe solving a (convex) QP problem. Recent years have seena rapid development of on-line QP solvers that are able tosolve such kind of linearized problems in the milli- or evenmicrosecond range on embedded hardware, see e.g., [24]–[28].
10.50460.5461.510.500246control 818000.58DC current [pu],torque [pu]21810DC current [pu],torque [pu]2speed [pu]speed [pu]0control angle[deg]5line voltagemagnitude [pu]line voltagemagnitude [pu]IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS180900time [sec]time [sec]Fig. 6. Simulation results: Speed, DC current, torque and control anglesduring a voltage dip scenario with conventional PI control. The current andinverter angle are plotted in black. The torque and rectifier angle are plottedin gray.Fig. 7. Simulation results: Speed, DC current, torque and control anglesduring a voltage dip scenario with MPC. The current and inverter angle areplotted in black. The torque and rectifier angle are plotted in gray.Since our system model is nonlinear, we are forced toperform the time-discretizaton of problem (4) on-line at eachsampling instant. For doing so, cost criterion (2) is replacedby a finite sum over a fixed grid of discretization points, i.e. j T j N 1Xu u αuα u αQ(ijdc i dc )2 αjJ disc : Rjuβ u βuβ u βIn order to obtain a highly efficient implementation of thisapproach, we make use of the code generation functionalityof the ACADO Toolkit [16]. This software takes a symbolicformulation of the control problem and allows the user toautomatically generate customized nonlinear MPC algorithmsthat are tailored to the specific problem structure. The resultingC code is self-contained, highly optimized and able to run onembedded computing hardware.In our case, the MPC algorithm was implemented on ABB’scontroller AC 800PEC, which is based on a 32-bit 600 MHzPower PC processor and which also includes an FPGA and a64-bit floating point unit. The entire MPC solution, runningwith a sampling time of 1 ms, consumes only a minor fractionof the computational resources, such that the whole controlsystem can be executed on time.j 0 2 Q(iNdc idc ) ,where the superscript j (or N ) denotes the respective quantityat time tj [kTs , kTs Tp ]. Also the input and state bounds (3)are only imposed at these discretization points. Finally, timediscretization of the dynamic model (1a) is achieved by meansof an on-line integrator (we applied an explicit Runge-Kuttascheme [29] with constant stepsize).Along with this discretization in time, the integrator schemealso computes first-order derivatives of the state trajectory withrespect to the initial state value and the control moves alongthe horizon (so-called sensitivities). We obtain a discretetime linearization of the optimal control problem (4), whichcorresponds to a convex QP problem as in the case of lineardynamics. In order to reduce computational load for solvingthe resulting QP problem, we exploit its sparsity structureby eliminating all state variables from the QP formulation toarrive at a smaller-scale, dense QP problem. This QP problemis then solved by the on-line QP solver qpOASES [30], [31].The procedure just described to solve nonlinear MPCproblems is a sequential quadratic programming (SQP)-typeapproach known as real-time iteration scheme with GaussNewton approximation of the second-order derivatives [32].TABLE ID ESIGN DATA OF THE EXPERIMENTAL SYSTEM .ParameterValueUnitLine voltage, prim. sideLine voltage, sec. sideLine frequencyRated line current, sec. sideDC link inductanceRated DC currentRated stator voltageRated stator currentRated stator frequencyRated shaft powerRated rotational AVAHzMWrpm
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS6speed [rpm]300020001000actref0501015202510152025152025DC current [A]300020001000actref05control angles [deg]0150100rectifierinverter5000510time [min]Fig. 8.Experimental results: Overview of recorded signals during experimental tests on an industrial-scale pilot plant installation.V. E VALUATIONThe suggested model predictive current controller wasapplied to an industrial-scale pilot plant comprising ABB’sMegadrive LCI, a dual-winding synchronous machine and agas compressor. The two-pole synchronous machine has anominal speed of 3600 rpm and a nominal power of 48 MW.The LCI is in a 12/12-pulse configuration as the one shownin Figure 1. The installation was connected via a transformerto the medium voltage grid with a grid voltage of 132 kV. Forpower factor correction and current harmonics compensation,electric filters are used. Table I summarizes the technicalspecifications of the setup.A. HIL SimulationsBefore applying the developed control solution to a mediumvoltage drive, it was tested in a hardware-in-the-loop (HIL)simulation environment. In the following the MPC scheme iscompared to a conventional PI control approach in a scenariowith heavy grid disturbances, where the system describedabove was simulated.At the beginning of the scenario, the drive is operatingwith nominal speed. Then a sequence of instantaneous voltagedips of increasing magnitude occurs. Figures 6 and 7 showthe selected control angles and the resulting DC link currentduring this scenario. The conventional PI controller is reactingonly with the rectifier firing angle, which is not sufficient tosustain the DC link current (and thus the drive torque) duringundervoltage conditions. Moreover there are big overshootswhen the grid voltage returns, ultimately resulting in theactivation of the overcurrent protection, and thus in trippingthe drive.In contrast the MPC solution: the constraint on the DCcurrent (3) ensures that the height of the current peaks arelimited and no trip occurs. Moreover, by actuating both controlangles, the MPC is able to maintain the DC current and thus toprovide some drive torque during undervoltage conditions. Theamount of residual torque however is limited by the availablegrid voltage and by the tolerable height of the current peaks.B. MV Drive EvaluationAfter successful evaluation in HIL simulation, the MPCscheme was tested on a medium-voltage (MV) drive. Anoverview of the recorded signals is shown in Figure 8. Therecorded signals are the speed reference ωr and the actualspeed ωr of the rotor, the DC current reference i dc providedby the reference governor and the actual DC link current idc ,and the control angles α and β at the line side and the machineside of the converter.In the top part of Figure 8, the speed reference ωr andthe actual speed ωr of the rotor are displayed. The MPC isactivated while the machinery is rotating with 2700 rpm. Anumber of speed reference ramps are requested, acceleratingand decelerating the compressor between 2700 rpm and 3750rpm. After around 27 minutes of operation, the drive is stoppedand the compressor slows down to standstill.The middle part of Figure 8 shows the DC current referencei dc provided by the reference governor together with the actual
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS73800phase-to-phasevoltage [kV]speed [rpm]1037503700613.2613.4613.6 000DC current [A]3000DC current [A]02500200010000613.4613.6613.8150control angle [deg]control angle [deg]613.210050613.2613.4613.6613.8time [sec]Fig. 9. Experimental results: Zoom to recorded signals at first speed decrease.For signal legend see Figure 8.DC link current idc . It can be seen that at each speed referenceramp, the DC current reference exhibits step-wise changes.The actual DC current exhibits a large ripple. A certain amountof ripple is unavoidable, due to the topology of the loadcommutated inverter. A part of the DC current ripple is due toaggressive tuning of the MPC. The average of the DC currenthowever is well-aligned with the DC current reference.In the bottom part of Figure 8, the control angles α and β ofthe line side and the machine side of the converter are shown.With the suggested model predictive control solution, smalldeviations of the machine side control angle from a speedand current-dependent upper bound βmax are taking place tosupport the regulation of the DC current. The line side firingangle α shows also a certain ripple for disturbance rejection.More insight into the shape of the signals is provided inFigure 9. Exemplarily a zoom at the first decrease of thereference speed is shown, which was requested shortly after10 minutes of operation. It can be seen how the DC currentfollows the stepwise changes of the current reference. For sucha step it is sufficient to adapt the line-side firing angle, whereasthe machine-side angle stays virtually constant.The waveforms of the DC current and the phase-to-phasevoltages under steady-state conditions are shown in Figure 10.VI. C ONCLUSIONThe paper considered nonlinear model predictive controlfor torque regulation of a synchronous machine supplied bycurrent source converters. By reformulating the torque controlproblem into a DC current control problem, a constrained control problem for a linear parameter-varying system is derived.In contrast to standard PI controllers, the MPC formulation15010050time [sec]Fig. 10.Experimental results: Zoom to waveforms under steady-stateconditions. Line side voltage plotted in gray, machine side voltage plottedin black. For remaining signal legend see Figure 8.does not impose a separate control structure, but uses both therectifier and inverter angles simultaneously to stabilize the DClink current and thereby to control the torque. This increasesthe ability to stabilize the system and reject disturbances.Experimental verification on an industrial-scale pilot plantdemonstrate the viability of the approach.R EFERENCES[1] J. M. Maciejowski, Predictive Control with Constraints. Prentice Hall,2001.[2] J. Rawlings and D. Mayne, Model Predictive Control: Theory andDesign. Nob Hill Pub., 2009.[3] S. Kouro, P. Cortes, R. Vargas, U. Ammann, and J. Rodriguez, “ModelPredictive Control - A Simple and Powerful Method to Control PowerConverters,” IEEE Trans. Ind. Electron., vol. 56, no. 6, pp. 1826–1838,June 2009.[4] S. Bolognani, S. Bolognani, L. Peretti, and M. Zigliotto, “Designand Implementation of Model Predictive Control for Electrical MotorDrives,” IEEE Trans. Ind. Electron., vol. 56, no. 6, pp. 1925–1936, June2009.[5] S. Mariéthoz, A. Domahidi, and M. Morari, “High-Bandwidth ExplicitModel Predictive Control of Electrical Drives,” IEEE Trans. Ind. Appl.,vol. 48, no. 6, pp. 1980–1992, Nov. 2012.[6] J. Scoltock, T. Geyer, and U. Madawala, “Model Predictive DirectPower Control for Grid-Connected NPC Converters,” IEEE Trans. Ind.Electron., vol. 62, no. 9, pp. 5319–5328, Sept. 2015.[7] T. Geyer, G. Papafotiou, and M. Morari, “Model Predictive DirectTorque Control - Part I: Concept, Algorithm, and Analysis,” IEEE Trans.Ind. Electron., vol. 56, no. 6, pp. 1894–1905, June 2009.[8] L. Tarisciotti, P. Zanchetta, A. Watson, S. Bifaretti, and J. Clare,“Modulated Model Predictive Control for a Seven-Level Cascaded HBridge Back-to-Back Converter,” IEEE Trans. Ind. Electron., vol. 61,no. 10, pp. 5375–5383, Oct. 2014.[9] S. Almér, S. Mariéthoz, and M. Morari, “Sampled Data Model PredictiveControl of a Voltage Source Inverter for Reduced Harmonic Distortion,”IEEE Trans. Control Syst. Technol., vol. 21, no. 5, pp. 1907–1915, Sept.2013.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS[10] Y. Xie, R. Ghaemi, J. Sun, and J. Freudenberg, “Model PredictiveControl for a Full Bridge DC/DC Converter,” IEEE Trans. Control Syst.Technol., vol. 20, no. 1, pp. 164–172, Jan. 2012.[11] S. Almér, S. Mariéthoz, and M. Morari, “Dynamic Phasor ModelPredictive Control of Switched Mode Power Converters,” IEEE Trans.Control Syst. Technol., vol. 23, no. 1, pp. 349–356, Jan. 2015.[12] A. B. Plunkett and F. G. Turnbull, “Load Commutated Inverter Synchronous Motor Drive Without a Shaft Position Sensor,” IEEE Trans.Ind. Electron., vol. IA-15, pp. 63–71, 1979.[13] E. Wiechmann, P. Aqueveque, R. Burgos, and J. Rodriguez, “On theEfficiency of Voltage Source and Current Source Inverters for HighPower Drives,” IEEE Trans. Ind. Electron., vol. 55, pp. 1771–1782, Apr.2008.[14] R. Bhatia, H. Krattiger, A. Bonanini, D. Schafer, J. Inge, and G. Sydnor,“Adjustable speed drive with a single 100-MW synchronous motor,”ABB Review, vol. 6, pp. 14–20
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 1 Model Predictive Control in the Multi-Megawatt Range Thomas J. Besselmann, Sture Van de moortel, Stefan Almér, Pieder Jörg and Hans Joachim Ferreau Abstract—The paper at hand presents an application of model predictive control to a variable
IEEE 3 Park Avenue New York, NY 10016-5997 USA 28 December 2012 IEEE Power and Energy Society IEEE Std 81 -2012 (Revision of IEEE Std 81-1983) Authorized licensed use limited to: Australian National University. Downloaded on July 27,2018 at 14:57:43 UTC from IEEE Xplore. Restrictions apply.File Size: 2MBPage Count: 86Explore furtherIEEE 81-2012 - IEEE Guide for Measuring Earth Resistivity .standards.ieee.org81-2012 - IEEE Guide for Measuring Earth Resistivity .ieeexplore.ieee.orgAn Overview Of The IEEE Standard 81 Fall-Of-Potential .www.agiusa.com(PDF) IEEE Std 80-2000 IEEE Guide for Safety in AC .www.academia.eduTesting and Evaluation of Grounding . - IEEE Web Hostingwww.ewh.ieee.orgRecommended to you b
Signal Processing, IEEE Transactions on IEEE Trans. Signal Process. IEEE Trans. Acoust., Speech, Signal Process.*(1975-1990) IEEE Trans. Audio Electroacoust.* (until 1974) Smart Grid, IEEE Transactions on IEEE Trans. Smart Grid Software Engineering, IEEE Transactions on IEEE Trans. Softw. Eng.
1066 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 51, NO. 5, OCTOBER 2004 Stator-Flux-Oriented Vector Control of Synchronous Reluctance Machines With Maximized Efficiency Heath F. Hofmann, Member, IEEE, Seth R. Sanders, Member, IEEE, and Ahmed EL-Antably, Member, IEEE
IEEE TRANSACTIONS ON IMAGE PROCESSING, TO APPEAR 1 Quality-Aware Images Zhou Wang, Member, IEEE, Guixing Wu, Student Member, IEEE, Hamid R. Sheikh, Member, IEEE, Eero P. Simoncelli, Senior Member, IEEE, En-Hui Yang, Senior Member, IEEE, and Alan C. Bovik, Fellow, IEEE Abstract— We propose the concept of quality-aware image, in which certain extracted features of the original (high-
IEEE Robotics and Automation Society IEEE Signal Processing Society IEEE Society on Social Implications of Technology IEEE Solid-State Circuits Society IEEE Systems, Man, and Cybernetics Society . IEEE Communications Standards Magazine IEEE Journal of Electromagnetics, RF and Microwaves in Medicine and Biology IEEE Transactions on Emerging .
Standards IEEE 802.1D-2004 for Spanning Tree Protocol IEEE 802.1p for Class of Service IEEE 802.1Q for VLAN Tagging IEEE 802.1s for Multiple Spanning Tree Protocol IEEE 802.1w for Rapid Spanning Tree Protocol IEEE 802.1X for authentication IEEE 802.3 for 10BaseT IEEE 802.3ab for 1000BaseT(X) IEEE 802.3ad for Port Trunk with LACP IEEE 802.3u for .
EIC, IEEE Transactions on Cloud Computing – Yuanyuan Yang EIC, IEEE Transactions on Cognitive Communications and Networking – Ying-Chang Liang EIC, IEEE Transactions on Molecular, Biological, and Multi-Scale Communications – Chan-Byoung Chae EIC, IEEE Transactions on Signal and Info
IEEE SENSORS JOURNAL, VOL. XX, NO. XX, XXXX 2017 1 Preparation of Papers for IEEE TRANSACTIONS and JOURNALS (February 2017) First A. Author, Fellow, IEEE, Second B. Author, and Third C. Author, Jr., Member, IEEE Abstract—These instructions give you guidelines for preparing papers for IEEE Transactions and Journals. Use this document as a
