Sensorless control of wind turbine cage induction generator based on phase-locked loop

Doktorska disertacija je rezultat istraživanja u području upravljanja kaveznim asinkronim vjetrogeneratorom bez mjernog člana brzine vrtnje koje je provedeno u suradnji s hrvatskim industrijskim sektorom. Zbog svoje pouzdanosti, robusnosti, niske cijene i niskih troškova održavanja, u posljednje vrijeme kavezni asinkroni vjetrogeneratori postaju sve češći izbor u vjetroagregatima s promjenjivom brzinom vrtnje. Dodatno povećanje pouzdanosti i robusnosti takvog koncepta vjetroagregata postiže se izostavljanjem mjernog člana brzine vrtnje. Međutim, upravljanje kaveznim asinkronim vjetrogeneratorom bez mjernog člana brzine vrtnje ima određene specifičnosti koje su još uvijek nedovoljno istražene poput magnetizacije u vrtnji, estimacije magnetskog toka te upravljanja u području slabljenja magnetskog toka. Magnetizacija u vrtnji kaveznog asinkronog stroja bez poznavanja njegove brzine vrtnje predstavlja izazov jer za uspješnu magnetizaciju frekvencija napajanja mora odgovarati brzini vrtnje stroja. U okviru istraživanja za ovu disertaciju razvijena je nova metoda za magnetizaciju u vrtnji kaveznog asinkronog vjetrogeneratora bez mjernog člana brzine vrtnje temeljena na fazno zatvorenoj petlji koja omogućava uspješnu magnetizaciju vjetrogeneratora u cijelom području rada za metode za estimaciju magnetskog toka temeljene na naponskom modelu. Razvijena metoda je implementirana u digitalni sustav upravljanja i eksperimentalno verificirana, a dobiveni rezultati su pokazali da se može primijeniti u praksi. Budući da kavezni asinkroni vjetrogeneratori bez mjernog člana brzine vrtnje ne rade na brzinama bliskim nuli i zahtijevaju magnetizaciju u vrtnji, postojeće metode za estimaciju magnetskog toka nisu prilagođene za vjetrogeneratore. Stoga je u okviru ove disertacije razvijena metoda za estimaciju magnetskog toka kaveznog asinkronog vjetrogeneratora bez mjernog člana brzine vrtnje temeljena na fazno zatvorenoj petlji koja može magnetizirati vjetrogenerator u vrtnji. Razvijena metoda je također implementirana u digitalni sustav upravljanja, a eksperimentalni rezultati su pokazali odlične performanse u stacionarnom stanju i dinamičkim uvjetima. Kavezni asinkroni vjetrogeneratori rade i u području slabljenja magnetskog toka u kojem se referentna vrijednost toka mijenja s brzinom. Zbog zasićenja u željezu, promjenom referentne vrijednosti magnetskog toka mijenja se i glavni induktivitet. Za razliku od motorskog režima rada, korekcija glavnog induktiviteta kod vjetrogeneratora je nužna kako bi se osiguralo da stroj razvije zadani moment. U ovom radu predložena je nova metoda za korekciju glavnog induktiviteta kaveznoga asinkronog vjetrogeneratora u području slabljenja magnetskog toka. Eksperimentalni rezultati pokazali su dobre performanse razvijene metode u cijelom području rada.Squirrel cage induction generator (SCIG) with full-scale power converter has been receiving increasing attention in the past few years for variable-speed wind energy conversion systems (WECSs). Well-known advantages of SCIG, such as robustness, reliability, low price, low maintenance requirements and costs, coupled with consistent price reduction in power electronics, made the SCIG with full-scale power converter an attractive choice for wind power generation. The reliability and robustness of WECS with SCIG are additionally increased by omitting the speed sensor of the SCIG. However, sensorless control of SCIGs used in wind power applications has certain issues which have gained insufficient attention, such as magnetization at non-zero speeds, rotor flux estimation and torque control in field-weakening range. This thesis provides a new method for magnetization of SCIG at non-zero speeds, a new flux estimation method adjusted for SCIGs used in wind power applications and a new method for the correction of main inductance of SCIG in field-weakening region which ensures the accurate torque control. Magnetization of SCIGs used in wind power applications is a demanding task because these generators are magnetized at non-zero speeds. If the generator speed is unknown, the required converter output frequency is also unknown and it is difficult to achieve a successful magnetization of SCIG when the existing flux estimation methods are used. Although this subject matter is very attractive, little has been published about it. However, commercially available speed sensorless drives contain a flying start function and are thus able to determine the rotor speed of a non-magnetized induction machine during the start-up. This thesis analyzes the magnetization process of SCIGs at non-zero speeds and proposes a new magnetization method, which is based on a phase-locked loop (PLL) and ensures successful magnetization of the generator when voltage-based flux estimation methods are used. The proposed method is experimentally verified by using two different voltage-based flux estimation methods on two different test rigs. The experimental tests are carried out on a 560 kW SCIG in the Laboratory for electrical machines at the University of Zagreb, Faculty of Electrical Engineering and Computing and on a 1900 kW squirrel-cage induction generator at the test station of the company Končar Generators and Motors Inc. The experimental results show that the proposed method is applicable in practice. Since the SCIGs used in wind power applications do not operate at low speeds and require successful magnetization at non-zero speeds, the existing sensorless model based flux estimation methods have to be modified in order to achieve successful magnetization. This thesis deals with the issues associated with flux estimation methods and proposes a new rotor flux estimation method which can magnetize sensorless vector controlled SCIG at non-zero speeds and is, therefore, suitable for use in wind power applications. The proposed method is rather simple to implement, uses minimal processor time and memory and can be used with wind generation systems of high nominal power. The method is based on a PLL and the orthogonality between the rotor flux space vector and its back electromotive force (EMF) vector. Rotor flux is estimated from stator voltage equations without integrating the back EMF components in stationary reference frame and the well-known difficulties with the implementation of pure integrators are thus avoided and the accuracy of the estimated flux angle does not depend on the implementation of the integrator. The proposed method is implemented into digital control system and its steady-state and dynamic performance are experimentally verified. Experimental tests are conducted on the aforementioned test rigs and the results show excellent performance and confirm that the method is highly appropriate for wind power applications. Squirrel cage induction generators used in wind power applications also operate in field-weakening region. Control of induction machine in field weakening region implies varying the rotor flux setpoint in accordance with the speed of rotation. Because of magnetic saturation, with the change of the rotor flux setpoint, the main inductance is also changed. In order to ensure that the SCIG used in wind power application develops the required torque in field-weakening region, the correction of the generator's main inductance has to be implemented. This thesis proposes a new method for the correction of the main inductance of the SCIG in field-weakening region. The proposed method is experimentally tested on the aforementioned test rigs. The experimental results show a satisfactory performance of the developed method in the whole field-weakening region. The thesis is organized into six chapters. Chapter 1. - Introduction contains the review of the research area related to the thesis. Also, the motivation is described and the research objectives are defined. The original contributions of the thesis are listed. The outline of the thesis with a short summary of the contents is presented. In Chapter 2. - Vector control of squirrel cage induction machines, the basic concept of rotor field oriented vector control is explained. The mathematical models of the squirrel cage induction machine in stationary and rotating reference frames used in the thesis are derived. Additionally, the control of variable-speed, pitch-regulated wind turbine is briefly described and sensorless control of wind turbine cage induction generator is discussed. Finally, the two test rigs used for experimental verification of the proposed methods are described. Both test rigs include a 1.5 MW back-to-back converter KONČAR KONvert W1500 which is used for the implementation of the proposed methods. The back-to-back converter KONČAR KONvert W1500 is controlled by a digital control system based on the fixed-point DSP ADSP-21992. The first test rig is located at the Laboratory for electrical machines at the University of Zagreb, Faculty of Electrical Engineering and Computing, Department of Electrical Machines, Drives and Automation. The test rig includes a 375 kW permanent magnet synchronous motor (PMSM) used as a wind turbine emulator and fed by frequency converter ABB ACS800. The PMSM is mechanically coupled with a 560 kW SCIG via a torque measurement sensor. The SCIG is connected to the grid via a back-to-back converter KONČAR KONvert W1500. The generated energy is returned to the grid via the back-to-back converter, which makes this laboratory system energy efficient and compliant with the most demanding grid requirements. The second test rig is located at the test station of the company Končar Generators and Motors Inc., Zagreb. The test rig includes two mechanically coupled 1900 kW squirrel cage induction machines, one of which is connected to a back-to-back frequency converter KONČAR KONvert W1500 and used as a generator and the other one is connected to a 3400 kVA synchronous generator and used as a wind turbine emulator. Chapter 3. - Magnetization of cage induction generator at non-zero speeds by using phase-locked loop describes the new magnetization method for sensorless vector controlled wind turbine cage induction generators. At the beginning of the chapter, an overview of the existing magnetization methods is given and their advantages and drawbacks are discussed. Further on, a detailed investigation into the magnetization process of the SCIG, performed during the development of the proposed magnetization method, is presented. Voltage-based flux estimation methods with low pass filter and low pass filter and adaptive compensation are explained in details. Furthermore, the basics of phase-locked loops and their operating principles are given and a linearized model of a phase-locked loop for three-phase systems is derived. Finally, the developed method is described in detail and the results of its experimental verification are analyzed. The performance of the proposed method is experimentally verified in the whole operating range of the SCIG. Experimental tests are conducted on a test rig at the Laboratory for electrical machines at the University of Zagreb. Experimental results of the magnetization process of the SCIG at two different speeds, 300 rpm and 1200 rpm, respectively, are presented. Also, the impact of the proposed magnetization method on the dynamic performance of the system after the completion of the magnetization process is investigated and the experimental results for a torque command at 1200 rpm after the magnetization process has finished are presented. Additionally, parameter sensitivity analysis of the proposed magnetization technique is performed and experimental results are also presented. The analyzed parameters are magnetizing inductance (Lm), stator resistance (Rs), rotor resistance (Rr), stator leakage inductance (Lσs) and rotor leakage inductance (Lσr). The sensitivity to the variation of each parameter is experimentally tested with three different values of parameters: 120%, 100% and 80% of their rated value. The experimental results of the adaptive compensation method at the rotor speed of 1200 rpm are presented for different parameter values. Chapter 4. - Flux estimation of cage induction generator by using a phase-locked loop describes the new method for rotor flux estimation of sensorless vector controlled wind turbine cage induction generators. The importance of flux estimation is shown through the analysis of impact of the flux estimation error on performance of vector controlled induction machines. An overview of the existing flux estimation methods for squirrel-cage induction machines is given and implementation issues associated with integration of back-EMF in voltage-based estimators are discussed in detail. The requirements on sensorless flux estimation methods for SCIGs used in wind power applications are defined and a new approach to the integration of back-EMF is presented. Finally, the developed method is described and the results of its experimental verification with detailed discussions are given. The performed experimental tests can be classified into three groups: magnetization tests, steady state performance tests and dynamic performance tests. All tests are performed on the aforementioned test rigs. Experimental results for the magnetisation process of the SCIG at 500 rpm and 1200 rpm are presented. The performance of the proposed method in the steady state is tested in the whole operating range of the SCIG under various loads. In the thesis, the experimental results for the no load steady state operation of the SCIG at 500 rpm and experimental results for steady state operation of the SCIG at 1492 rpm under 80% of the rated load are given. The dynamic performance of the proposed method is also investigated by conducting experimental tests of the speed and SCIG torque reference changes. Experimental results for speed change from 500 rpm to 800 rpm under 30% of rated load and experimental results for ramped torque reference at 800 rpm are given. The experimental results show excellent performance and confirm that the method is highly appropriate for wind power applications. Chapter 5. - Correction of the main inductance value within control system of cage induction generator in field-weakening range describes the new method for correction of the main inductance value within control system of vector controlled SCIG in field-weakening range. The chapter contains an overview of the existing approaches for field-weakening control and an overview of the existing methods for identification of the magnetizing curve and main inductance correction in field-weakening range. The proposed method is described in detail and the validity of the proposed method is tested experimentally and confirmed. The tests are performed on the laboratory test rig in the Laboratory for electrical machines at the University of Zagreb. The experimental results for steady state and dynamic performance in field-weakening range are presented and discussed. Chapter 6. - Conclusion contains the main conclusions of the thesis as well as the summary of the original scientific contributions. It also provides guidelines for future work. The original contributions of the thesis are given as follows: 1. Magnetization method based on a phase-locked loop for sensorless vector controlled cage induction generators used in wind power applications. The original contribution is a new method for magnetization of a sensorless vector controlled wind turbine cage induction generator which enables successful magnetization of wind turbine cage induction generators at non-zero speeds without using signal injection techniques. The proposed method is based on a phase-locked loop and can be used with the existing voltage-based flux estimation methods. The proposed method is experimentally verified on two different laboratory test rigs. 2. Rotor flux estimation method based on a phase-locked loop for sensorless vector controlled cage induction generators used in wind power applications. The original contribution is a new method for rotor flux estimation of sensorless vector controlled wind turbine cage induction generator which is based on a phase-locked loop and the orthogonality between rotor flux space vector and back-electromotive force space vector. The proposed method estimates the rotor flux space vector without using pure integrators and low-pass filters and enables successful magnetization of sensorless vector controlled wind turbine cage induction generators in a whole operating range. The proposed method is experimentally verified on two different laboratory test rigs. 3. Method for correction of the main inductance of wind turbine cage induction generators in field-weakening range. The original contribution is a new method for correction of the main inductance value in the control system of a cage induction generator in field-weakening range. The proposed method is used with a voltage controller and ensures that the wind turbine cage induction generator develops maximum torque in field-weakening range. The proposed method is experimentally verified