1 Introduction
The frequency conversion speed regulation transformation of medium and high voltage large capacity motors is the focus of national energy conservation and emission reduction. The main power circuit of the medium-voltage frequency converter generally uses the multi-level inverter topology to achieve the pressure resistance of the power device, reduce DV / DT, improve harmonics, etc. [1]. Among them, the multi-level inverter of the H-bridge-level structure is most mature in the field of inverter speed regulation of the medium and high voltage motor, and the application is the most widely used.
At present, in the product of medium and high pressure inverters, motor speed control policies are multi-V / F control or vector control (also known as magnetic field directional control), and Direct Torque Control, DTC) research and application There is fewer, and the difficulty is relatively difficult. One of the main reasons lies in the number of switched tubes in the multi-level topology, which causes the switch to the switch to the switch to the scale. In addition, the traditional DTC uses a hysteresis comparator, the inverter switching frequency is not fixed, it is difficult to digitally realize, and the multi-level waveform is difficult, current, and torque pulsation are large.
Implementing a high-performance speed regulatory strategy such as a DTC requires detecting the speed of the motor, but the speed sensor has increased the complexity, cost and maintenance requirements of the system, and reduces reliability and robustness.
This paper is directed to the characteristics of the cascading multiple levels, and introduces the wrong time-sampling space vector modulation method and the speed sensor technology to the level of the plurality of levels, solving the switch vector in the traditional DTC application in the multi-level field. The problem of complex, waveform quality is not good, torque pulsation is equal, with high DC voltage utilization, power unit uses equalization, harmonic content, simple method, easy digital implementation, etc. This method was simulated using MATLAB / Simulink.
2, the difficulty of direct torque control of multi-level
The traditional direct torque control uses the-swipe control of the magnetic chain and torque, according to their change, and the switch state of the voltage space vector is directly selected from the spatial position of the stator magnetic chain, and the fast torque response is obtained. However, its actual torque is pulsating in the upper and lower limits of the hysteresis comparator, and the switching frequency is not fixed. An improved solution is to combine the spatial vector modulation (SVM) method with DTC, close-loop PI adjustment to the torque, replace the switching vector table with the voltage space vector modulation module, generating the switch state of the PWM wave control the inverter, which can be made The switching frequency is constant, and the torque ripple is also significantly reduced.
However, in the multi-level field, the number of basic spatial vessels in the inverter is numerous, and the number of multi-level inverters per phase level is coupled for each phase, and the number of basic spatial vessels is (2N + 1). 3. The high-voltage inverter of each phase of each phase is more than 343, and the number of high voltage inverters per phase 6 units has reached 2197. Such a wide variety of basic spatial vectors make the space vector selection algorithm very complicated. In addition, the selection of spatial vectors should consider the power unit switch load balancing, which puts higher requirements for the algorithm. Therefore, in the case where the number of levels is large, the spatial vector algorithm achieves difficulties, and it is difficult to meet real-time control requirements.
In order to overcome the above problems, the Sampe-Time-Staggered Space Vector Modulation, STS-STAGERED SPACTOR MODULATION, the STS-SVM) policy can greatly reduce the complexity of the spatial vector selection in the cascading multiple level, and can achieve the switch load Automatic equalization, high performance efficiency, easy to achieve high performance real-time control such as non-velocity sensor DTC.
3, sampling SVM strategy when wrong
The sampling space vector method was first applied in a combined converter structure as shown in FIG. The converter was composed of an inverter unit of n 3-phase 6 switch tube, and the output was coupled by a transformer. The basic idea of STS-SVM is the sampling of the reference voltage in each converter unit, and the sampling period is TS, and the sampling time of the adjacent unit is configured. Ts / N. Thus, the number of basic spatial vectors equivalent to the system greatly increases, and the resulting output voltage has a plurality of levels, and the phase voltage level is 2N + 1.
Figure 1 Combined converter topology
Combined converters and 3-phase H-bridge-level multi-level inverters have a conversion equivalent relationship in the topology. The two-stage multi-level converters in Fig. 2 (a) can be equivalent to the primary 3 phase H bridge structure in Fig. 2 (b), and equivalent switch tubes are the same in two pictures. The number indicates that the switch tube of the left bridge arm in the first H bridge and the switch tube of the right bridge arm can be equivalent to two 3-phase 6 bridge arm units of the multi-level transformer. This allows the STS-SVM modulation method to be used for the primary 3-phase H bridge. The specific method is to obtain a driving signal of the switch tube in Fig. 2 (a) with the STS-SVM method, and the same number of the same numbered in Fig. 2 (b) is drove. Since N = 2 in the two-stage converter, the sampling time of the two units is configured to open TS / 2. Transfigured to Fig. 2 (b), which is equivalent to the same amplitude and frequency modulation ratio of 6 switching tubes of the first-stage 3-phase H bridge inverter left bridge arm and the right bridge arm. Level spatial vector modulation, and to make the sampling time of the two reference voltages to TS / 2.
(A) two-stage multi-level converter
(B) 1 three-phase H bridge inverter
Figure 2 equivalent relationship between the two-stage combined converter and the primary three-phase H bridge inverter
According to the above thoughts, the inverter of the N-stage H bridge is equivalent to a combined converter of 2N units, and the sampling time of the adjacent two-stage H-bridge units should be configured with each other. / 2N.
The implementation method of applying the wrong time sampling modulation strategy is applied in the N-stage H-bridge-level inverter. As long as the driving signals of three h-bridge arms in a certain level H bridge are obtained according to the traditional two-level spatial vector algorithm, the drive signals of the other switching tubes in the system can be obtained by the corresponding delay. The two level spatial vector algorithms are performed in the main controller, and the delay can be implemented by adding hardware units outside the primary controller. This greatly reduces the burden on the main controller, and can adapt to the requirements of fast real-time control.
In STS-SVM, the arrangement of the overall output voltage vector of the system is automatically completed, and each bridge arm trigger waveform derived from the two-level space vector algorithm itself has symmetrical and equalization, so the overall switch load is also equal.
4, STS-SVM no speed sensor DTC system
The STS-SVM control algorithm is simple, the switch load balance is simple to achieve complex speed sensor direct torque control.
Figure 3 is an overall structure of a cascaded multi-level no velocity sensor DTC control system based on STS-SVM. In the figure, the speed regulator, torque regulator, and magnetic chain regulators are proportional integral adjustment, and the torque regulator needs to use a limit before the PI adjustment, so as not to prevent excessive torque error causes an overcurrent impact. The system has always used the STS-SVM module to generate the switch state of the PWM wave control inverter, and abandon complex switch vector gauge. In addition, since the hysteresis comparison is not used, the sampling frequency of the system is fixed, it is easier for digital implementation.
Figure 3 Structural Structure of STS-SVM-free sensor DTC system
4.1 STS-SVM modulated multi-level inverter
Here, the multi-level inverter is a three-stage H-bridge-level topology shown in Figure 4 (a). The generation of the driving signal in the STS-SVM model is obtained by the modulation wave derived by the two level spatial vector algorithm to compare the triangular waves corresponding to each switch tube, as shown in FIG. 4 (b). There is a certain phase relationship of each triangular carrier, so that the mutual disorder of the sampling cycle is applied.
(A) Main circuit of the three-level level multiple-level inverter
(B) STS-SVM drive signal generating unit
Figure 4 Class Union Multi-level main circuit and PWM generating unit
4.2 Magnetic Charm and Torque Observation
The estimation of the stator magnetic chain can generally be divided into three models, namely the U-I model, the I-N model, and the U-N model. Where the magnetic chain expression in the U-I model is
(1)
Among them, US, IS, RS are stator magnetic chains, voltages, current values, and stator resistance values. It can be seen that the UI model observed stator magnetic chain does not need to speed information, the only motor parameter required is the stator resistor RS, so it is very suitable Apply here.
Direct torque control requires measurement of electromagnetic torque TE as feedback, generally using calculation. Electromagnetic torque has a variety of different expressions, and an electromagnetic torque can be obtained using any two parameters in a stator current, a rotor current, a stator magnetic chain, a rotor magnetic chain. In direct torque control, the following formula is usually calculated TE:
(2)
Among them, PN is the extremely log of the motor.
The model of the magnetic chain and torque observation is established in MATLAB / Simulink, as shown in Figure 5.
Figure 5 Stator magnetic chain and torque observation model
4.3 Speed Estimation
The model reference adaptive system (MRAS) is more convenient, and has a complete insemination of the change in rotor resistance, the effect of motor parameter changes on the speed estimation is also small. The voltage model of the rotor magnetic chain is independent of the motor speed, and the current model of the rotor magnetic chain is related to the motor speed, so the voltage model of the rotor magnetic chain is selected as the reference model, and the current model of the rotor magnetic chain is selected as the adjustable model.
Since the stator magnetic chain has been estimated in the magnetic chain observation, the reference model can be represented by a stator magnetic chain:
Where Tr = LR / R is the rotor time constant, RR is a rotor resistance, which is a rotor angle velocity.
Figure 6 MRAS scheme for using rotor magnetic chain estimation speed
Figure 7 Speed Estimation Model
5 Simulation results and analysis
The simulation model of the entire system is established in MATLAB / SIMULINK. Among them, the motor model adopts the software self-contained two-stage three-phase induction motor model. The parameter is: rated power PN = 3730W, rated line voltage un = 380V, rated frequency fn = 50Hz, rotor resistor RR = 1.083Ω, stator resistor RS = 1.115Ω, stator, rotor inductance LS = lr = 0.2097H, fixed rotor mutual inductance lm = 0.2037 h, rotation inertia J = 0.02kggm2. The inverter The DC power supply voltage per stage is 104V, the sampling period TS = 952 μs.
A 6 ngm load is added at 0.3s. The simulation waveform of each variable is shown in Figure 8.
(A) Identify speed and actual speed
(B) Torque dynamics
(C) Stator magnetic chain (D) current
(E) phase voltage
(F) line voltage
Figure 8 System simulation waveform
It can be seen from the simulation waveform that the motor is started 0.2s, the system basically enters the steady state; the identification speed can estimate and track the actual speed; the phase voltage output 7 level; the line voltage output 13 level; the current waveform is good; Time flies are relatively small; during startup, the stator magnetic chain can quickly reach a given value, and remain circular; the torque is rapidly reaching the limit value (23 ngm) at startup, and then gradually falls to the empty load. Stable value, dynamic response characteristics are good when loading.
6 Conclusion
In this paper, the sampling space vector modulation method, the model reference adaptive method is combined with the direct torque control, and the direct torque control of the speedless sensor of the cascading multi-level frequency converter is realized, and there is a simple implementation, high reliability, torque. Multi-point, such as pulsation, has better practical value. The modeling method of each part is given, and this method is verified by simulation.
Editor in charge: GT, read full text
Our other product:
