Encoderless & Predictive Control of Synchronous Machines

Encoderless & Predictive Control of Synchronous Machines Ralph M. Kennel, Technische Universitaet Muenchen, Germany kennel@ieee.org EMAD E M A D lectrical achines nd rives Laboratories Wuppertal University Germany

Museums District

Institute for Electrical Drive Systems & Power Electronics Prof. Dr.-Ing. Ralph Kennel kennel@ieee.org

Prof. Dr.-Ing. Ralph M. Kennel 1984 Ph.D. at University of Kaiserslautern, Germany 1984 1997 Development of Industrial Servo Drives with Digital Control, Robert BOSCH GmbH, Erbach/Odw., Germany 1997 1999 Advanced Development of Electrical Drives for Automotive Applications Robert BOSCH GmbH, Buehlertal, Germany 1994 1999 Visiting Professor at the University of Newcastle upon Tyne, UK 1999 2008 Professor for Electrical Machines & Drives at Wuppertal University, Germany since 2008 Professor for Electrical Drive Systems & Power Electronics at Technische Universität München, Germany

Main Research Areas 1. Dynamic Drives for Industrial Applications 2. Drive Control (progressive concepts) 3. Hardware-in-the-Loop Systems

Dynamic Drives for Industrial Applications Sensorless Control for Synchronous and Asynchronous Machines this is our main topic #1 Position Feld Position.. Laufer.. Trager Asynchron maschine? Position Feld Position.. Laufer Ultra High Speed Drives (finished)

Drive Control Predictive Control of Inverters and Drives this is our main topic #2 Optimal PWM Adaptive Control of Mechatronic Systems reference commands precalculation of the behaviour for each of the switching states definite number of equivalent circuits without switching elements definite number of switching states comparison between precalculation and reference commands definite number of switching elements next switching state or switching time can be fixed inverter

Hardware-in-the-Loop Systems High Power Current Source Inverters (finished) Multilevel Inverters Serial Connection of Inverter Switches (e.g. IGCTs) Driving High Power IGBTs with Shaped Gate Voltages and Currents this is our main topic #3 Virtual Machine virtual machine

Encoderless & Predictive Control of Synchronous Machines Ralph M. Kennel, Technische Universitaet Muenchen, Germany kennel@ieee.org Page 11

Reasons for Industrial Applications of Drives with encoderless Control: Cost???? Reliability Robustness is encoderless (sensorless) resulting in additional cost??? Page 12

Industrial Drives with Sensorless Control since several years / decades sensorless control is investigated and published on conferences and magazines - acceptance in industry, however, is rather low Why? new ideas and concepts are interesting for industry, only if they do not result in higher cost or higher effort!!! What does that mean for industrial drives with sensorless control? no additional or more powerful processors / controllers no additional hardware or additional sensors (e. g. voltage this sensors) was valid no increased installation effort with respect to parameter from adjustments 2000 to 2010 Page 13

Industrial Drives with Sensorless Control since several years / decades sensorless control is investigated and published on conferences and magazines - acceptance in industry, however, is rather low Why? new ideas and concepts are interesting for industry, only if they do not result in higher cost or higher effort!!! What does that mean for industrial drives with sensorless control? single scheme for wide speed range (no phase over) no additional noise (except usual noise by inverter supply) insensitivity with respect to parameter variations What does industry think today? Page 14

Industrial Drives with Sensorless Control Actual Requirements from Industry there should be a single concept for encoderless control for the complete speed range (from standstill to maximum speed) single scheme for wide speed range (no phase over) in case there is a signal to be injected for speed/position detection no additional noise this should not cause any additional noise - except usual noise caused by inverter supply with standard PWM parameters of electrical machine and/or power elctronics should not impact the performance of encoderless control too much (a certain impact is acceptable) insensitivity with respect to parameter variations Page 15

Sensorless (Encoderless) & Predictive Control encoderless control of synchronous machines fundamental model methods high frequency injection methods tracking of magnetic saliencies / anisotropies predictive control of power electronics basic principles Model Predictive Control (MPC) encoderless predictive control of synchronous machines idea performance and potential practical results conclusions Page 16

Field oriented control of PMSM rotor position needed Page 17

Fundamental model based position estimation when knowing voltage (by reference values) as well as current (by real values) it is possible to estimate rotor speed and rotor position Page 19

Calculation of Speed by Fundamental Model is not Practicable for Very Low Speeds... because the voltage signal becomes very small errors between real voltage and values used for calculation cannot be avoided and become more significant DC components of these errors let the integrators for flux calculation drift away the calculated speed gets more and more incorrect is an encoder/resolver the only feasable solution?? Page 20

fundamental model high frequency injection simple realisation does not work at frequency 0 parameter dependencies current injection voltage injection measuring voltage is high enough additional voltage sensors transient current response Basic Principles of Encoderless Control no additional hardware very short measuring time stationary current response standard microcontroller sufficient very small measuring current Page 22

Resolver injection of a stationary (sinusoidal) high frequency signal sensing of a two-dimensional stationary (sinusoidal) signal response Tamagawa R1 S2 u 1 (γ) 0 π 2π γ Stator u e Rotor u R Stator u 2 R2 S1 Stator u 1 S3 S4 u 2 (γ) 0 π 2π γ Page 23

Stationary Signal Injection Method according to R. Lorenz, S.-K. Sul, R. Kennel, etc. the basic idea is to use the electrical machine itself as a resolver!!! a resolver is nothing else but an electrical machine can we operate the motor itself like a resolver? if the machine itself is a resolver (encoder) is that really an encoderless control??? now we do the same with an electrical AC machine Page 24

Standard industrial servo PMSM distributed stator windings surface mounted PM Page 25

Fundamental Frequency Excitation (in α-β coordinates) Page 26

High Frequency Excitation (in α-β coordinates) Page 27

injection of high frequency voltages fundamental voltage phasor/vector fundamental current phasor/vector injected high frequency voltage phasor/vector high frequency current phasor/vector (response) Page 28

Injection of High Frequency Rotating Phasors rotating voltage phasor u c elliptic current response i c Page 32

Position Information of Salient Rotors in High Frequency Rotating Phasors machine responds on a rotating voltage phasor with an elliptic current response ellipse is correlated with the geometric anisotropy rotor position information is included of the rotor in the high frequency current elliptic current response i c (rotating) Page 33

Injection of High Frequency Alternating (Pulsating) Voltage Phasors composing an alternating (pulsating) voltage phasor by two phasors rotating in opposite direction advantage : no rotational (HF) field no additional torque Page 34

High Frequency Current Response of a Synchronous Machine Page 35

Current Response of a Misoriented System voltage and current show different orientation!! Page 36

Page 37

Page 38

Page 39

Page 40

Page 41

Tracking Scheme for Magnetic Anisotropies i (Fˆ) cd = K sin ( ω c t) l cq i (Fˆ) cq ( ωct)( lcq cd ) ˆ δa = K sin l Page 43

Tracking Scheme for Magnetic Anisotropies Tracking the estimated angle of the rotor flux by controlling i cq to 0 Page 44

Encoderless Control Structure step response of the PLL; PLL is locked after ca. 10-15 ms Page 45

Encoderless Control Structure control structure of an encoderless control with alternating high frequency signal injection the estimated angle can be used for field orientation as well as for speed or position control of synchronous machines Page 46

Sprungantwort der Drehzahlregelung Drive with Speed Control Page 47

Sprungantwort der Lageregelung Step Response of Encoderless Position Control Page 48

Results the tracking control scheme presented here synchronizes on the saturation anisotropy of a synchronous machine. the tracking control scheme is not depending on any machine parameter. the size of additional software for the tracking control is comparable to the software of a rotor model for the field oriented control of an induction machine. the high frequency current signal can be measured together with the fundamental current by the standard current transducers of a standard drive inverter. Page 49

Practical Experience with an Industrial Servo Drive Implementation of a sensorless control into a servo drive of training of a development engineer 2 x 1 week in our laboratory programming of additional software in manufacturer s factory delivery of prototype after ca. 3 months presentation on Hanover Fair in April 2006 Page 50

meanwhile : more industrial applications WEG (Brazil) as mentioned before BAUMÜLLER same experiences as WEG TRÜTZSCHLER successful application in textile machinery YASKAWA successful implementation in one product two more companies who do not want to be mentioned ABM Greiffenberger advertising actively on SPS/IPC/Drives 2010 Page 51

some more experiences Bolognani reported (in 2006?)...... saturation in q direction increases under load difference between l cq and l cd decreases... and vanishes at a certain load (armature reaction) an encoderless tracking of the anisotropy does not work any more this effect appears around 2 to 3 times rated load with IPM motors around 5 to 6 times rated load with SMPM motors Page 52

Accuracy of the Rotor Position Identification under Load Conditions a) without load b) rated load (carrier frequency f c = 2 khz) why is the armature reaction so small??? Page 53

Accuracy of the Rotor Position Identification under Load Conditions... because the usual rotor designs of servo motors (mechanical holes for inertia reduction) do not allow a load depending displacement of the main field why is the armature reaction so small??? Page 54

Typical Cascaded Structure of Drive Control ϕ position controller speed controller current controller power motor inertia gear etc. electronics windings I ω ϕ Page 56

Problems of Linear Algorithms in cascaded control structures speed control must be much faster than position control and current control must be much faster than speed control current control must be very fast to achieve position control with reasonable cycle times in the controlled system (drive, converter, ) however, there is no time constant justifying cycle times of 100 µs or less Page 57

General Structure of a Predictive Controller ϕ prediction and calculation switching state actual machine state power electronics machine and power electronics model motor windings I inertia ω gear etc. ϕ reminds slightly to state control state control, however, is basically a linear control predictive control is not!!! Page 58

Usual Structure of Drive Control why PWM? linearization of the inverter consequences? very high switching frequency DC link PI controller Page 59

Structure of a Direct Control DC link direct controller Page 60

Principle of Predictive Control reference commands precalculation of the behaviour for each of the switching states comparison between precalculation and reference commands next switching state or switching time can be fixed definite number of equivalent circuits without switching elements definite number of switching states definite number of switching elements inverter Page 61

Trajectory Based Predictive Control Strategies system states are forced to follow (pre-)defined natural reference trajectories difference to sliding mode control there the trajectories are not natural Page 62

Example : Trajectory Based Predictive Control Direct Speed Control acc. to Mutschler ω * ω e model and prediction u k u d ω i s u s i = ~ u e / a k+1 k+1 S k S k+1 S k Hy e / a k+3 k+3 a = ω +Hy ω M 3~ e / a k k S k+2 e / a k+2 k+2 e = ω ω ref Page 63

Characteristics of Trajectory Based Predictive Control system states are forced to follow (pre-)defined reference trajectories switching takes place at intersections between different system-trajectories or at (pre-)defined instants switching frequency of the inverter can be fixed to a constant value control behaviour comparable to feedforward control exact knowledge of system parameters is required appropriate for realisation by digital circuits or controllers Page 64

Example : Trajectory Based Predictive Control Direct Self Control (DSC) acc. to Depenbrock Page 65

Hysteresis Based Predictive Control Strategies switching of inverter takes place at the (multi-dimensional) border(s) of a hysteresis area Page 66

Example : Hysteresis Based Predictive Control Predictive Current Control acc. to Holtz Page 67

Example : Hysteresis Based Predictive Control Predictive Current Control acc. to Holtz Page 68

Example : Hysteresis Based Predictive Control Predictive Current Control acc. to Holtz i s * i s i s u sk predict model u k di sk dt = ~ u d u s jim ω s * i s di n dt i s ω M 3~ 0 i s Re Page 69

Example : Hysteresis Based Predictive Control Predictive Current Control acc. to Holtz Page 70

Page 71

Characteristics of Hysteresis Based Predictive Control switching takes place at borders of a hysteresis area a maximum error can be (pre-)defined switching frequency of the inverter is not constant control behaviour comparable to feedback control exact knowledge of system parameters is not required appropriate for realisation by analog circuits Page 72

The Human Behaviour of DMPC DMPC is like playing chess the player calculates in advance all possible moves until a prediction horizon the player chooses the move with the best expectations of success after each opponent s move pre-calculation and optimization is repeated Page 74

Predictive Control History Future Page 75

Predictive Control Overview Page 76

Direct Model Predictive Control System Model / Cost Function Page 77

Direct Model Predictive Control System Model / Cost Function Page 78

Characteristics of Model Based Predictive Control basic ideas are derived from state-space control the past is explicitely considered (mostly by the system state) future control values are pre-calculated and optimized the first of the precalculated control values only model parameters can be estimated on-line until a (pre-)defined horizon is transmitted to the controlled system extension to MIMO-control is possible with little additional effort use of non-linear model is possible for non-linear control systems a lot of calculation power is required Page 79

Model Based Predictive Current Control complete enumeration extensive processing power needed there are 7 (or 8) possiblities for the following switching state the respective system behaviour (current) can be calculated in advance a chess player, so why however, should we does not that really in predictive consider control each possibility???

Model Based Predictive Current Control further prediction, however, is only considered for the candidate sequences staying within the permitted limits so why should we do that in predictive control???

Model Based Predictive Current Control determine those switching possibilities only that are either feasible or point in the proper direction these are candidate sequences feasible pointing in the proper direction

Model Based Predictive Current Control determine those switching possibilities only that are either feasible or point in the proper direction these are candidate sequences not feasible not pointing in the proper direction

Model Based Predictive Current Control for the candidate sequences, further prediction (e. g. by a reduced system model) is performed example : the number of steps after which the first of the two variables the i sα and i isβ leaves the feasible region is the number η

Different Way of Thinking in Model Based Predictive Control 1. model of the controlled system this is no difference to conventional control the better the model, the better the prediction 2. cost function the engineer has to learn to describe what he wants the controlled system really to do!!! 3. stability that s a really good question next question? Page 87

Experimental Results (DMPC) current control comparison : PI control model predictive control Page 88

Experimental Results (DMPC) current control Low a change switching of the frequency cost function (nothing high switching else!!!) frequency dynamic of step response results in is different identical behaviour!! Page 89

Advantages Features of (Longe Range) Predictive Control possibility to use foreknowledge about drive system (system model) inverter limitations and dynamic behaviours are taken into account improved representation of non-linear systems no need for time challenging cascade structure improved dynamic behaviour Disadvantages high processing capability required for industrial use change in teaching engineers necessary stationary accuracy and dynamic behaviour depend on accurracy of model parameters Page 90

Actual Situation in cascaded control structures speed control must be much faster than position control and current control must be much faster than speed control current control must be extremely fast to achieve position control with reasonable cycle times at the time most requirements in industrial applications are satisfied sufficiently there is no strong need for improvement in industry however at a certain time there will be a demand for improvement with respect to a future increase of requirements more investigations should be done Page 91

Discussion predictive control strategies offer the possibility to use foreknowledge about the drive system physical limitations and dynamic behaviour of power electronics non-linear systems are represented better (by non-linear models) no need for time challenging cascaded structures the way of thinking is different are taken into account model of the controlled system cost function with respect to a future increase of requirements more investigations should be done Page 92

Encoderless Control Basic Principle of High Frequency Injection Methods Goal : Detecting anisotropic magnetic characteristic of electrical machine Method : Detecting position dependant inductance of electrical machine Problem : Faraday s Law / Maxwell Equation No. 2 (f = 0) Solution : Injection of a high frequency signal (f 0) Comparison between ideal (= isotropic) complex current and real complex current (obtained by sensors) Difference Signal contains Position Information!

Basic Idea: A Predictive Torque Controller neglecting the saliency in the model causes a prediction error containing the angle information. Overview Predictive Torque Control Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Encoderless Predictive Control Basic Principle Goal : Detecting anisotropic magnetic characteristic of electrical machine Method : Detecting position dependant inductance of electrical machine Problem : Faraday s Law / Maxwell Equation No. 2 (f = 0) Solution : Do not inject an additional high frequency signal!!! there is sufficient excitation (f 0) by the inverter anyway Comparison between ideal (= predicted) current (ideal model) and real current (obtained by sensors) Does Difference Signal contains Position Information?

(Conventional) Predictive Torque Control Basic Structure : Overview Predictive Torque Control Saliency Tracking controller input: reference torque actuating variable: stator voltage Simulation Results Measurements Conclusion current and rotor angle must be measured Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

(Conventional) Predictive Torque Control Current and PM flux linkage from measurements Overview Predictive Torque Control Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

(Conventional) Predictive Torque Control Current and PM flux linkage from measurements 7 voltages vectors from inverter Overview Predictive Torque Control Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

(Conventional) Predictive Torque Control Current and PM flux linkage from measurements 7 voltages vectors from inverter prediction of current and respective torque Overview Predictive Torque Control Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

(Conventional) Predictive Torque Control Current and PM flux linkage from measurements 7 voltages vectors from inverter prediction of current and respective torque Overview Predictive Torque Control Saliency Tracking Selecting optimum of cost function Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

(Conventional) Predictive Torque Control Overview Predictive Torque Control Saliency Tracking Simulation Results Discrete model of the machine Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

(Conventional) Predictive Torque Control Overview Predictive Torque Control Saliency Tracking Simulation Results Discrete model of the machine Measurements Current prediction based on mean inverse inductance (ideal model) Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Saliency Tracking Approach Predicted current progression (ideal model) Overview Predictive Torque Control Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Saliency Tracking Approach Predicted current progression (ideal model) Overview Predictive Torque Control Real current progression Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Saliency Tracking Approach Predicted current progression (ideal model) Overview Predictive Torque Control Real current progression Saliency Tracking Prediction error Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Saliency Tracking Approach Measured prediction error Overview unfortunately depends on switching state! Predictive Torque Control Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Saliency Tracking Approach Measured prediction error Overview Reconstructed prediction error Predictive Torque Control Saliency Tracking has to be corrected by amplitude as well as angle! Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Saliency Tracking Approach Measured prediction error Overview Reconstructed prediction error Predictive Torque Control Saliency Tracking PLL controller input Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Saliency Tracking Approach Saliency tracking scheme (STS) in drive software : Overview Predictive Torque Control Saliency Tracking Measured prediction error Reconstructed prediction error Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Encoderless Predictive Control Performance as HF (high frequency) injection and encoderless predictive schemes use physically identical excitation (f 0) and search for identical magnetic anisotropies (L d L q ) the performance of both schemes is comparable to each other the problems with respect to strange machine designs are identical as well advantage of encoderless predictive scheme : no additional high frequency signal is injected is there some more potential?

Encoderless Predictive Control Potential Yes, there is!!! considering the strange machine designs in the ideal model leaves the position information in the difference signal again is there some more potential?

Simulation Results for PMSM Simulation parameter of PMSM Overview Predictive Torque Control Saliency Tracking Speed controlled encoderless predictive torque control Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Simulation Results for PMSM Speed controlled step response to rated speed very good dynamics in simulation Overview Predictive Torque Control dependency on torque gradients Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Measurements with Reluctance Machine Data of transverse laminated RM Overview Predictive Torque Control Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Measurements with Reluctance Machine Speed controlled step response to 160% rated speed Overview Predictive Torque Control Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Measurements with Reluctance Machine Response to 66% rated torque load step at speed controlled standstill Overview Predictive Torque Control Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Summary Proposed Scheme: Neglects the saliency in PTC equations Prediction error contains angle information Reconstruct Prediction Error using PLL angle Vectorproduct of both is PLL input Benefits: Saliency based: permanent operation at standstill No signal injection: operation at high speed Overview Predictive Torque Control Saliency Tracking Simulation Results Measurements Conclusion Institute for Electrical Drive Systems & Power Electronics Technische Universität München Arcisstr. 21, D-80333 Munich - peter.landsmann@tum.de

Industrial Needs The proposed PTC (Predictive Torque Control) method single scheme for wide speed range (no phase over) The sensorless control scheme presented here works from standstill to maximum speed does not inject any additional signal to the reference - neither fundamental frequency nor high frequency no additional noise (except usual noise by inverter supply) As long as there is a detectable saliency PTC is very robust to variations of the motor parameters? insensitivity with respect to parameter variations Page 127

SLED 2011 www.eee.nott.ac.uk/sled/

www.precede2011.de

Thank you!!!