资源描述
1 Chapter 9Kouhei Ohnishi, Nobuyuki Matsui, Yoichi HoriEstimation, Identification, and Sensorless Control in AC Drives 2 AC drives became more and more economical and popular.n Since the early years of the 20th century, electric motors for variable speed drives have been widely applied in large-capacity applicationsn the steel industry n the automobile industryn In the early stage, DC motors were widely used for adjustable speed control n Since the late 1960s, AC motors have been replacing DC motors in a wide area of industry applications n Since AC drives required more complicated controllers in the beginning, they were not so economically feasiblen Allied to advances both in digital control technology and power semiconductor devices, AC drives became more and more economical and popular. n In almost all areas, DC drives are now replaceable with AC drives 3 n However, there still exist some areas which are not suitable for AC drive applications. n One of these is the area which requires precise torque control. For instance, injection machines.n AC motors sometimes generate torque error or torque pulsation due to some parameter variations. n To overcome such problems, more sophisticated techniques are necessary in the controller. n These techniques are based on estimation or identification of motor parameters n Employ the recent developments in digital control, including high-speed digital signal processors (DSPs) and parallel processing To overcome torque problems. 4 9.2 PARAMETER ESTIMATION IN AC DRIVES 9.2.1. Parameter Identification In Brushless Motors 9.2.2. Parameter Identification In Induction Motors 5 n The control scheme of brushless motors with trapezoidal flux distributions (BLDM, brushless DC motor) is relatively simple. Usually it does not need parameter identification. n For brushless motors with sinusoidal flux distribution (PMSM, permanent magnet synchronous motor), generally identification of the parameters is necessary for precise control . (3 electric parameters) nArmature resistance, narmature inductance, nand EMF coefficient.Parameters of Brushless Motors 6 n Two effective approaches are presented here:n self-tuning regulator (STR) which has a tuning ability to make output-error zero inside the controller; n model reference adaptive system (MRAS) which has a referred model in the controller. n Direct applications of STR and MRAS to parameter identification do not always lead to successful results, because of the limitation of the processing time of the controller CPU. n Since identification should be performed in parallel with current and speed control, it is essential to reduce the processing time for identification by a simple algorithm. 7 STR-Based Parameter Identifiern At steady-state, the PMSM has the simple equivalent circuit just like a DC motor. Terminal voltage, line current, and armature resistance are measured to identify the circuit parameters.n Figure 9-1 shows an experimental evaluation of the influence of such parameter variations in armature current error at steady state. n A current-regulated voltage source inverter supplies almost sinusoidal current. n In the figure, the ordinate is the current control error due to the parameter variation 8 parametercontroller parametermotornthe parameter variation coefficient is defined as 9 10nsince the estimation of the armature inductance uses division and the relatively small d-axis voltage, the error is a little larger compared to that of the EMF coefficient. 11 MRAS-Based Parameter Identifier n Figure 9-4 displays a MRAS-based identification approach, which includes a voltage-based motor model as a reference model. The input of the identifier is a current difference between model and actual motor. The current difference is decomposed into two elements, from which the armature inductance and the EMF coefficient are identified. 12 n the experimental results of identification of armature inductance and EMF coefficient for a tested motor. 13 the experimental estimation error 14 Application of Parameter Identification to Torque Controln An interesting application example for the parameter identification of the brushless motor is a torque sensorless torque control. n Fig(a), the conventional current-based torque control system.n The torque reference is divided by a torque constant (=EMF coefficient in SI unit) to generate a current reference. 15 16 Parameters of Induction Motorsn Basically induction motor in steady state is represented by the equivalent circuit in Figure 9-7. The classical no-load test, locked rotor test, and electrical quantity measurement test give identified parameters in Figure 9-7. 17 n A sample of flowchart is shown in Figure 9-8 4. 18 9.3 Sensorless Drives of AC Motors 19 sensorless drives of AC motorsn basically,the vector-controlled AC motors require speed or position sensorsn these sensors bring several disadvantages drive cost reliability machine size noise immunity n it is necessary to achieve the precise control of torque and speed without using position and speed sensors(so-called sensorless drives of AC motors) 20 9.3.1 Sensorless Drives of Brushless Motorsn As stated, there are two kinds of brushless motors: the motor with a trapezoidal flux distribution and that with a sinusoidal flux distribution. The approaches to sensorless drive of the brushless motor vary, depending on the rotor flux distribution. n the motor with a trapezoidal flux distribution n The first one provides an attractive candidate, because two of the three stator windings are excited at a time. n As a result, the unexcited winding can be used as a sensor 10, 11; that is, the speed EMF induced in the unexcited winding is used to determine the rotor position and speed.n the motor with a sinusoidal flux distribution excites three windings at a time and the sensorless control algorithm becomes complicated. 21nSince the actual rotor position is not known without a position sensor, the aim is to make the angular difference between the fictitious and actual rotor positions converge to zero. the motor with a sinusoidal flux distributionnthe d-q axis corresponds to an actual rotor position nthe - axis is a fictitious rotor position 22 two approaches n Both are the estimation of the angular difference by using the detected state variables and the estimated state variables which are obtained from a motor model in the controller. n The approaches differ according to the motor model, i.e.n Voltage model-based drive 12 n Current model-based drive 13n These two are basically the model-based control and generally require on-line identification of the motor parameters if higher performance is required. However, it is interesting to note that the second method has robust control characteristics against the motor parameter variation. 23 sincosc EcR PL LL R PLv i Kv i 01 rM r EM R PL LL R PLv i Kv i the voltage model-based sensorless drivethe voltage equation isthe voltage equation under the ideal condition that the fictitious and actual axes are coincident is 24n the angular difference can be made to converge to zero by the following rule n if vr0 , then decreasesn if vr 0 , then increases(for clockwise rotation) sinM E EK K Taking a difference between -axis voltage assuming is small cc 25n (whereT is a sampling period and is the motor speed of the model), then sin1 cosc Eci R iLP KLi L iR sincosM E E MM K T K TL Li i ii i i M K i dt ( )c M K i dt the current model-based sensorless driveThe current model isthe current difference is M 26 n torque-speed characteristics under a current model-based algorithm. nThe motor rating is 1.2 (kW), 6-poles, 1200 (rpm), 98 (kgf cm). The maximum speed is 1500 (rpm), the minimum speed is 60 (rpm), and a steady-state maximum speed error is within 0.4%. 27 9.3.2 Sensorless Drives of Vector Controlled Induction Motorsn the basic idea is estimation of speed by using applied voltage, line current, and frequency Slip frequency control approach field orientation control approach 28 Slip frequency control approach sdsqrrs iiLR sr 29 field orientation control approach not only the speed but also the rotor flux are simultaneously estimated for the sensorless drive in a wide speed range. The stator equation is used for correction of adjustable current reference model. The speed is estimated by a kind of error of flux components which is derived from Popovs stability criteria. 30nThe rotor flux observer based on the motor voltage model estimates the rotor flux by using the stator current transformed to the stationary coordinates ( - axis) systemnthe vector control is carried out on the basis of the estimated rotor flux nthe motor speed is directly estimated through MRAS by using the flux and the stator current nthe estimated speed is used to adjust the motor model 31 Figure 9-16nan example of the sensorless control characteristics of 2.2 (kW), 4-poles induction motor. nThe maximum speed is 2400 (rpm) under the field-weakening control and the minimum speed is 15 (rpm). nIt is noted here that the speed control accuracy is 0.4% under the tuned condition; however, it is 1.4% under the detuned condition. nLike the conventional vector control with sensors, the identification of the rotor resistance is important and difficult. 32n End of Chap9
展开阅读全文