IN MANY industrial applications, ac machines are preferable

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 46, NO. 1, FEBRUARY 1999 111 Automatic IM Parameter Measurement Under Sensorless Field-Oriented Control Yih-Neng Lin and Chern-Lin Chen, Member, IEEE Abstract A novel approach to automatic induction motor (IM) parameter measurement under sensorless field-oriented control (FOC) is presented. Before startup, the inverter drive automatically performs the dc test, the no-load test, and the lockedrotor test for the driven IM. The only measured values are input current signals. No additional hardware is required in this approach. In order to automatically measure the IM parameters under FOC structure, the concepts of no-load test and dc test should be modified. In addition, because the conventional lockedrotor test requires that the rotor shaft be locked mechanically, this makes automatic measurement impossible. In order to solve this problem, a locked-rotor test at single-phase excitation for sensorless FOC structure is also performed in this paper to facilitate automatic IM parameter measurement. Finally, the test data are then computed to get the IM parameters and the field current command for FOC. The presented method has been tested on a 3-hp inverter-driven IM system. Its effectiveness is illustrated by experimental recordings. Index Terms Automatic measurement, induction motor parameters. I. INTRODUCTION IN MANY industrial applications, ac machines are preferable to dc machines due to their simple and robust construction [1]. In high-performance control areas of ac motors, the field-oriented control (FOC) concept [2] has become a standard tool because it provides linear control characteristics similar to a separately excited dc motor. The FOC normally requires a speed sensor, such as a shaft encoder, to achieve good performance. However, it is inconvenient to mount a speed sensor on the induction motor (IM) rotor shaft in some cases, such as motor drives in hostile environments or highspeed motor drives. Several FOC methods without speed sensors have been proposed [3] [6]. Rapid developments in the sensorless FOC technology may be expected in industrial applications. Sensorless FOC demands accurate IM parameters. Generally, every installation requires specific tuning to obtain the accurate parameters of the motor [7]. However, the IM parameters measured by conventional tests, the dc test, the no-load test, and the locked-rotor test [8], are not directly applicable to sensorless FOC. Thus, the concepts of the noload test and dc test should be modified for the FOC structure. Manuscript received April 14, 1996; revised February 25, 1998. Abstract published on the Internet October 26, 1998. The authors are with the Power Electronics Laboratory, Department of Electrical Engineering, National Taiwan University, Taipei, 10764, Taiwan, R.O.C. Publisher Item Identifier S 0278-0046(99)00495-5. Fig. 1. The per-phase circuit with counter EMF. The field current command can also be obtained in the noload test under sensorless FOC. The conventional locked-rotor test requires that the rotor shaft be locked mechanically. This prohibits automatic measurement by the FOC inverter drive. The mechanical locking of the rotor shaft can be omitted by substituting the three-phase locked-rotor test with a singlephase locked-rotor test [7], [9]. This method is especially suitable for inverter-driven motor systems and is modified in this paper for FOC applications. Using the concepts illustrated above, a fully automatic measurement system for IM parameters is developed. Details of the proposed method will be described in the following sections. No extra hardware is needed in these tests. Only input current signals are measured and sent to the controller. Experiments are performed to verify the effectiveness of the proposed methods. II. SENSORLESS FOC OF IM A. Model of Induction Machine The dynamic - model of the induction machine [10] is used in this paper. Fig. 1 shows the combined per-phase equivalent circuit for the - equivalent circuits with counter EMF. Voltages and fluxes are represented in the complex plane, where represents stator voltage vector and represents rotor flux vector. The magnitude of and can be expressed as and may be derived directly from the conventional dc test, no-load test, and locked-rotor test. State equations of the dynamic - model of the induction machine can be expressed as follows: (1) (2) (3) 0278 0046/99$10.00 1999 IEEE

112 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 46, NO. 1, FEBRUARY 1999 TABLE I SWITCHING TABLE The transformation between the rotationary coordinate and the stationary coordinate system is (8) where (9) Fig. 2. Sensorless FOC. Block diagram. Eight possible switching patterns. where differential operator; electrical speed of the rotor; ; ; ; ; B. Sensorless FOC A simple sensorless FOC with space-vector-based current controller for a pulsewidth modulation (PWM) inverter is shown in Fig. 2. The whole system can be divided into two parts, hardware and software. The hardware part consists of the voltage-source inverter (VSI) circuit and the IM. The software part is composed of FOC, space-vector PWM, and rotor speed estimation algorithm. For a field-oriented controller, it is desirable that where field current; slip frequency; estimation of angular frequency; command of angular frequency. (4) (5) (6) (7) For a space-vector-based current controller, it is desirable that the output current vectors should follow commands and [11]. By comparing the commands and with the feedback currents and, the digital output signals can be obtained. Then, the digital output signals of the comparators are used to determine the state of inverter switches according to Table I. The inverter output voltage vectors are defined according to this switching table. There are eight different voltage vectors in an inverter. All the possible switching patterns of the inverter are shown in Fig. 2. Since the primary voltage can be determined by the switching table and primary current can be detected by sensors, the rotor speed of the IM can be calculated by some speed estimation techniques. III. PARAMETER MEASUREMENT Conventionally, IM parameters are determined by the dc test, the no-load test, and the locked-rotor test. Practically, conventional tests should be performed each time the motor drive is installed on a new motor. There are some disadvantages in obtaining the IM parameters in this way. First of all, the IM parameters determined by conventional tests are not directly applicable to sensorless FOC, since the switching effects on the parameters are not taken into account by the conventional tests. Moreover, a technical person should be present to perform the conventional tests during installation, which is inconvenient in some cases. In order to conquer these problems, an automatic parameter measurement method is desired. In this section, an automatic IM parameter measurement method suitable for sensorless FOC is proposed. In this proposed method, all three conventional tests are modified to meet the automatic requirements. Details of the required modification and the related theories will be described below. A. DC Test In the automatic dc test, the inverter has to play the role of an additional dc power supply for measuring stator winding

LIN AND CHEN: AUTOMATIC IM PARAMETER MEASUREMENT UNDER SENSORLESS FOC 113 stator winding resistance can be found (10) Fig. 3(c) shows the command modifications required for this test. The desired input commands are, and. Under this condition, (11) (12) (c) The terminal current command is a dc signal. In addition, in order to apply the same control signals to phase and phase, digital output signal is forced to zero in the space-vector modulator. In this way, only three vectors,, and, are utilized for estimating stator resistance. The complete flow chart for the modified dc test is shown in Fig. 3(d). No extra measurement is required. In the modified dc test, phase is fed with the current which consists of dc and switching-frequency ripple current. Different test results of the stator winding resistance may appear due to the switching ripple effects which cause a higher temperature of stator windings. To minimize these switching effects and compromise complex computation, three estimated values under different switching frequencies are considered. A second-order polynomial is used. It can be written as (13) (d) where for are the estimated values and for are the different switching frequencies. The coefficient then represents the computed dc stator resistance value which gets rid of the influence of thermal effects. The desired dc stator winding resistance value measured under dc current can thus be obtained. is solved by the Cramer s rule [12] (14) Fig. 3. The modified dc test. Inverter motor connection. Equivalent circuit. (c) Block diagram. (d) Flow chart. resistance. Fig. 3 shows a typical inverter motor connection diagram. Three legs of the inverter are connected to the stator windings and, respectively. In the modified dc test, the same control signals are applied to the phase- and the phase- legs of the inverter. In this way, the phase- leg of the inverter is short to the phase. The inverter output current of phase is sent into the phase- winding and is shared by phase and phase. The current in the phase- winding is adjusted to the rated value, and the voltage between the terminals is computed from control signals. The current in the stator windings is adjusted to heat the windings to the temperature they would have been during normal operation [8]. The equivalent circuit is shown in Fig. 3 and the estimated where for. B. No-Load Test The conventional no-load test provides information about the magnetization inductance and its magnetization current. The test setup for the conventional no-load test is shown in Fig. 4. Two wattmeters, a voltmeter, and three ammeters are used. The IM equivalent circuit under the no-load test is shown in Fig. 4. The equivalent input impedance is, thus, approximated [8] (15) (16) (17) (18)

114 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 46, NO. 1, FEBRUARY 1999 Fig. 4. The conventional no-load test. Testing circuit diagram. Equivalent circuit. Three commands, and are modified in the proposed automatic no-load test, as shown in Fig. 5. Initially,, and. During the test, is gradually decreased to an appropriate no-load value. Before reaches its no-load value, the three-phase inverter output voltage will be unbalanced. Fig. 5 illustrates this idea. The output voltages are unbalanced when is larger than the terminal current. The space-vector-based current controller performs voltage vectors, and in a cycle. Once three-phase balance is achieved, will then be equal to its no-load value. The desirable command for the no-load test is considered as follows: (c) (19) The complete control flow chart for the modified no-load test is shown in Fig. 5(c). C. Locked-Rotor Test For a fully automatic test system, it is necessary to perform a single-phase locked-rotor test. The circuit diagram of this test is shown in Fig. 6. In this test, two of the three motor input terminals are shorted to each other. No torque will be generated under this condition, which implies that the rotor will remain stationary. Hence, no mechanical locking is required. The electric behavior is almost the same as in the case of three-phase excitation [9]. For the single-phase locked-rotor test measurements, the IM equivalent circuit is illustrated in Fig. 6. The magnitude of the equivalent impedance and resistance under excitation can be found (20) (21) Fig. 5. The modified no-load test. Block diagram. Unbalanced output voltages. (c) Flow chart. The equivalent reactance is then obtained (22) For the Y-connection case, the equivalent impedance under test is equal to coil impedance. Hence, (23) (24)

LIN AND CHEN: AUTOMATIC IM PARAMETER MEASUREMENT UNDER SENSORLESS FOC 115 (c) (d) Fig. 7. Results of modified dc test. Estimated resistance values. Estimated resistance values under different switching frequency. Fig. 6. The modified locked-rotor test. Testing circuit diagram. Equivalent circuit. (c) Block diagram. (d) Flow chart. The automatic locked-rotor test can be performed under sensorless FOC. Three commands are set to be, and, as shown in Fig. 6(c). The current command is related to the rated current of the motor. Another current command should be zero (set ) to guarantee the rotor shaft being stationary. The only possible switching patterns of the inverter are only, and. The complete control flow chart for the locked-rotor test is shown in Fig. 6(d). IV. EXPERIMENTS AND RESULTS A. Experimental Setup The test motor is a three-phase Y-connected IM and the related machine data is given in the Appendix. The threephase input signals required for the proposed method are provided by a variable-voltage variable-frequency inverter with microprocessor-based control. B. DC Test The estimated stator resistance of IM is calculated by averaging the samples of voltage commands and measured currents. The results are graphically described by Fig. 7. After 500 samplings, the estimated value of stator resistance approaches to a stable one. The experiments are performed under three different switching frequencies. Fig. 7 shows the test results. In each test, the total counts of output voltage vectors are limited to 100 000. In Fig. 7, the horizontal arrows represent status change transients of output voltage vectors and the number beside it stands for the total counts of status changes. During these transients, the turn-on and off delay time of switching devices will affect the average dc voltage, but it is obvious that the effect caused by turn-on and off delay time during transient is opposite the effect during transient. Therefore, the effects of turn-on and off delay times of the switching devices can be eliminated, since the counts of opposite transient pairs and are the same. Thus, the average dc voltage will not be affected. Finally, the computed dc stator resistance is approximated to 2.58 by (13) to reduce the thermal effects. C. No-Load Test Recordings of the no-load test are shown in Fig. 8. After 300 samplings, the average values of and are graphically described by Fig. 8 and (c), respectively.

116 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 46, NO. 1, FEBRUARY 1999 (c) Fig. 8. Data of modified no-load test. Adjusting values of Id 3. Average values of Id 3. (c) Average values of L ls + L m. Fig. 9. Results of modified locked-rotor test. Equivalent resistance under different input frequency. Leakage inductance under different input frequency. D. Locked-Rotor Test The single-phase locked-rotor test is performed at different input fundamental frequencies (15, 30, and 60 Hz) and different switching frequencies (10, 15, and 20 khz). An input fundamental current will be set as the rated current which is the normal operating current of the IM. The computed equivalent resistance and leakage inductance are graphically illustrated by Fig. 9 and, respectively. From Fig. 9, the equivalent resistance is proportional to the switching frequency under the same input fundamental frequency, but the leakage inductance decreases as the input fundamental frequency increases. This frequency-dependent characteristic is mainly caused by iron losses effects. To minimize iron losses effects, the results of the higher input fundamental frequency are selected as the equivalent resistance and leakage inductance. Fig. 10. Speed control characteristics. E. Speed Characteristics The IM parameters obtained by the conventional test in the Appendix and the automatic test ( mh, mh, and A) are applied to a sensorless FOC. The sensorless FOC used here is an adaptive flux observer proposed by Kubota, et al. [6]. Detailed formulation of the testing sensorless FOC system are listed in the Appendix. The switching frequency is set as 20 khz. Fig. 10 shows the speed control characteristics of the sensorless FOC system using parameters obtained by conventional and automatic test, respectively. From Fig. 10, it is obvious that the speed control characteristics will be improved using the parameters obtained by the proposed method.

LIN AND CHEN: AUTOMATIC IM PARAMETER MEASUREMENT UNDER SENSORLESS FOC 117 V. CONCLUSION A novel automatic IM parameter measurement scheme suitable for sensorless FOC has been proposed in this paper. The proposed method can automatically perform all the tests required and obtain all the parameters of the IM. Remarkable advantages are achieved utilizing the proposed method. 1) All the parameters obtained by the proposed method are derived under the FOC scheme, thus, the parameters are more suitable to the FOC scheme than parameters obtained by the conventional tests. Experimental results show that speed error will be reduced to about 30% under full-load conditions and reduced to about 50% under light-load conditions using parameters obtained by the proposed method. 2) The proposed method will perform all the tests required automatically; no manual disconnection/connection of electric wires or mechanical lock-rotor action is required, which is more convenient for installation. 3) No extra hardware is required by this approach. Only minor modification of the FOC software is needed. Since this method is performed before FOC starts, the execution time of the original controller will, thus, not be affected. 4) Field current command for FOC can be determined by the modified no load test. Experiments were also carried out to verify the effectiveness of the proposed method. All the experimental results show that the parameters obtained by the proposed method can greatly improve the controller performance. It is clear that the proposed method will offer an attractive choice in many industrial applications. is the leakage coefficient, is the rotor time constant,. The state observer is written as the following: and (27) where signifies the estimated values and is the observer gain matrix. The motor speed estimation is given by (28) where and are the arbitrary positive gain. A. Induction Machine Data rated frequency 60 Hz; rated voltage 220 V; rated current 8.6 A; rated power 2.2 kw. APPENDIX B. IM Parameters Obtained by Conventional Tests mh. mh C. Adaptive Flux Observer The state equation of an IM is given by (25) (26) REFERENCES [1] E. Y. Y. Ho and P. C. Sen, A microcontroller-based induction motor drive system using variable structure strategy with decoupling, IEEE Trans. Ind. Electron., vol. 37, pp. 227 235, June 1990. [2] F. Blaschke, The principle of field orientation as applied to the new transvector closed-loop control system for rotating-field machines, Siemens Rev., vol. 34, pp. 217 220, May 1972. [3] T. Ohtani, N. Takada, and K. Tanaka, Vector control of induction motor without shaft encoder, in Conf. Rec. 1989 IEEE-IAS Annu. Meeting, pp. 500 507. [4] C. Schauder, Adaptive speed identification for vector control of induction motor without rotational transducers, in Conf. Rec. 1989 IEEE-IAS Annu. Meeting, pp. 493 499. [5] X. Xu and D. W. Novotny, Implementation of direct stator flux orientation control on a versatile DSP based system, in Conf. Rec. 1990 IEEE-IAS Annu. Meeting, pp. 404 409. [6] H. Kubota, K. Matsuse, and T. Nakano, DSP-based speed daptive flux observer of induction motor, IEEE Trans. Ind. Applicat., vol. 29, pp. 344 348, Mar./Apr. 1993. [7] A. Gastli and N. Matsui, Stator flux controlled V=f PWM inverter with identification of IM parameters, IEEE Trans. Ind. Electron., vol. 39, pp. 334 340, Aug. 1992. [8] S. J. Chapman, Electric Machinery Fundamentals, 2nd ed. New York: McGraw-Hill, 1991, pp. 583 616. [9] N. R. Klaes, Parameter identification of an induction machine with regard to dependencies on saturation, IEEE Trans. Ind. Applicat., vol. 29, pp. 1135 1140, Nov./Dec. 1993. [10] B. K. Bose, Power Electronics and AC Drives. Englewood Cliffs, NJ: Prentice-Hall, 1986. [11] M. P. Kazmierkowski, M. A. Dzieniakowski, and W. Sulkowski, Novel space vector based current controllers for PWM-inverters, IEEE Trans. Power Electron., vol. 6, pp. 158 165, Feb. 1991. [12] P. V. O Neil, Advanced Engineering Mathematics. Belmont, CA: Wadsworth, 1983.

118 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 46, NO. 1, FEBRUARY 1999 Yih-Neng Lin was born in Taiwan, R.O.C., in 1966. He received the M.S. degree in electrical engineering in 1992 from National Taiwan University, Taipei, Taiwan, R.O.C., where he is currently working towards the Ph.D. degree in the Department of Electrical Engineering. His current research interests are highperformance induction motor drives and power electronics interface with utility power systems. Chern-Lin Chen (S 86 M 90) was born in 1962 in Taipei, Taiwan, R.O.C. He received the B.S. and Ph.D. degrees in electrical engineering from National Taiwan University, Taipei, Taiwan, R.O.C., in 1984 and 1987, respectively. Since 1987, he has been with the Department of Electrical Engineering National Taiwan University, where he is currently a Professor. His current research interests lie in the areas of analysis, design, and application of power electronics converters and motor drives.