INVESTIGATION ON SENSORLESS PMSM DRIVE

Abstract

Direct current, induction machine and synchronous machines are three basic electrical machines which serve daily requirements, small household equipments to large industrial applications, year after year. Electric motors are the largest consumer of electric energy including domestic and commercial applications which is 46% of global energy consumption according to the IEA statistic. The application demand of electric motors is increasing rapidly with the technological advancement. Due to the rising demands of electric drive, researchers are continuing their efforts to develop new kinds of machines as the Brushless dc (BLDC) Machine, Switched Reluctance Machine (SRM), Permanent Magnet Hysteresis Machine and Permanent Magnet Synchronous Machine (PMSM). After developing these new types of special machines, researchers are working on the control of these motors to optimize the design performance and cost. These developmental activities are now in a revolutionary stage due to the recent development in semiconductor and microprocessor technologies. The separately excited dc motors are been used, for many decades, extensively for variable speed drives and high performance drives, as the separately excited dc motor can be controlled in a simple way which is attributed to the decoupled nature of its field and armature. However, the dc motor has some disadvantages, which include limited range of speed of operation, lack of overload capability, robustness, and frequent maintenance as well as high cost due to brush-gear, and commutators. These drawbacks of dc motors have motivated researchers to develop high-performance variable speed drive for ac motors like induction and synchronous motors, where robustness and maintenance free operations are the main concern. Among the ac motors, induction motor has been widely used in the industries due to some of their beneficial features like low cost, good efficiency, reliability and ruggedness. However, it has some limitations like; it always runs at a lagging power factor. Another limitation is that due to slip-power loss the IM drive system is not highly efficient. The above mentioned limitations diverted attention of researchers towards the synchronous motors for high performance variable speed drives. The advantages of synchronous motors are as; it always runs at synchronous speed, its control is less complex as compared to IM. It also removes the slip power loss. The wire wound excited synchronous motors have some drawbacks such as the prerequisite of the extra supply, slip ring and brush gears for the field excitation. Keeping in view limitations of the conventional wirewound synchronous motors, more recently different kinds of special motors have been developed. Among them, the permanent magnet (PM) motor is becoming popular due to some of its advantageous features, which include high torque to current ratio, high power to weight ratio, higher efficiency, and robustness. The elimination of excitation winding reduces the cost and power loss. The advantages of PMSM over dc motor are compact size, less ii maintenance and over induction motor are high efficiency, small size and wide range of power factor. The popularity of PMSM comes from its enviable features such as; high efficiency, high torque-inertia ratio, high torque-volume ratio, high flux density, high power factor, lower maintenance requirements, ruggedness, compact size. An extensive review is carried out which starts with the basic motor modelling and covers open loop control, speed control and current control in vector controlled drive, performance analysis, possible inverters & control techniques used including dynamic performance improvement, torque ripple minimization and different sensorless control of the PMSM drive. In high performance drive current controller plays a vital role as it directly affects the quality of current fed to motor and indirectly affects the performance of motor in terms of efficiency, dynamic response etc. In vector controlled PMSM drive the proportional integral (PI) controller is widely used due to its simple execution. However, tuning of gains of PI controller is a challenging task, when there is a change in system parameter or change in load torque or speed command values. The design of PI controller is usually based on the mathematical model of plant. However, even if the plant model is known, building an accurate mathematical model is very difficult task due to the problem of parameter variation. Control of PM motor with fast dynamic response, good speed regulation and high efficiency necessitates information of the rotor position to implement the vector control. Rapid development of microprocessors and controllers (μC) and DSP has facilitated the vector control to become a common technique for PMSM drives. Vector control has been widely used and is a successful technique for PMSM because of its excellent torque response with minimum stator current uses. Implementation of vector control algorithm needs to use stationary to synchronous rotational reference frame transformation to regulate the corresponding current component. The reference frame transformation requires rotor position information. In present work performance of PMSM is investigated and evaluated for different speed-torque control algorithms and sensorless techniques. Mathematical modeling of PMSM with saliency and without saliency in different reference frames is presented. The design of an accurate control system requires the mathematical model of actual system being controlled. The d-q model for PMSM is developed in MTALAB. This model consists of only dynamic equations of PMSM representing the actual PMSM in terms of parameters and load performance. A close loop operation is performed with modelled PMSM, PI controller as speed controller and sinusoidal pulse width modulated (SPWM) inverter. With this PMSM drive shows the regain capability of speed with change in load. The iii controller parameters are tuned to get the fast dynamic response in speed. The simulation results show the effectiveness of controller and high dynamic performance of the drive. In comparison to other speed controllers a fuzzy logic controller is a non-linear adaptive controller. The fuzzy logic controller uses a time varying gains for speed control according to system responses raised by different control rules. Moreover, the FLC is based on the experience and intuition of a human operator. Fuzzy logic can be considered as mathematical theory with multi-valued judgment, probability speculation, and artificial intelligence to simulate the human approach to solve the problem by using an approximate reasoning to share different data sets for decision making. A fuzzy logic controller (FLC) with input and output scaling factor is proposed as speed controller in PMSM drive. Fuzzy speed controller (FSC) is proposed to overcome the demerits of conventional PI speed controller which produces current reference for current controllers. These scaling factors are tuned to achieve the desired performance of drive. The value of these scaling factors depends on motor parameter, load torque, reference speed, and fuzzy speed controller FSC parameter. As for as the FSC parameters are concerned, these constitute range of membership functions, type of membership functions, implication methods, rules and number of rules. The usefulness of proposed method is verified by computer simulation and experimental results. The performance of proposed controller is investigated for different operating conditions. In order to conquer the coupling effect and the sluggish response with scalar control and to achieve the high performance the vector control is implemented. Two-axis mathematical modeling is presented for PMSM. The basic idea of the vector control is to decompose the three-phase stator current into a magnetic flux-generating component and a torque-generating component. After decomposition both currents components are separately controlled like dc machine. The vector controlled PMSM drive has two loops; outer speed control loop, and inner current control loop. The Inner loop is having two- current controllers (d-q), which play an essential role as they directly affects the quality of current fed to the motor and indirectly affects the performance of drive in terms of efficiency and dynamic response. In order to overcome the problems associated with PI controllers, fuzzy based current controllers are employed for motor control which eliminates the controller parameter dependency on the system’s mathematical model and load disturbances. Use of fuzzy logic algorithm, to reduce the torque ripples, has been proposed, it refines the voltage vectors. Use of space vector modulation (SVM) significantly reduces the torque and flux ripples. A fuzzy current controller (FCC) is proposed, which generates the reference signals for the inverter, to improve the quality of voltage and current fed to the motor, resulting in high performance of drive. iv The designed vector controlled PMSM drive has three fuzzy logic controllers (FLC); one FSC and other two FCCs. This complete fuzzy logic based vector controller is termed here as Fuzzy Vector Controller (FVC). The limitations reported in literature are addressed through a FVC is proposed for PMSM drive and its performance is investigated. Further, the effects of load variation and reference commands on the performance of drive are extensively studied. The enhanced performance is achieved through designing and tuning of FLC for each controller separately. Tuning is done by observing the values at input and output of individual controller using PI controller by keeping the interval of membership function ([-15 15] for FSC, and [-2 2] for FCC) in mind. This methodology resolves the problem of nonlinearity and parameter deviations of PMSM drive. Moreover, it achieves high dynamic performance and good speed regulation and torque control with superior steady-state characteristics. Simulation is done in MATLAB/Simulink to prove the efficacy of proposed FVC as compared with PI controllers based vector control, and performance of drive under various operating condition is demonstrated. Further, experimental results obtained from laboratory prototype validate the efficacy & robustness of the proposed controller. Generally the rotor position information is obtained from a optical encoder, revolver or Hall Effect sensors for the implementation of vector control. Though, it is enviable to abolish these position sensors from PMSM to reduce overall cost of drive and hardware complexity, inertia, maintenance requirements and to increase the robustness and reliability, and to have noise immunity. Back-emf based methods offer adequate performance in the higher speed range, but at low or zero speed the magnitude of back-emf becomes negligible and difficult to measure. This makes speed estimation at low speed very difficult and this method is highly sensitive to motor parameters. A high frequency signal injection is used to extract the rotor position is reliable at zero speed but there is adverse effects of injected signal on motor dynamics and necessity of extra hardware. Principles for sensorless control of electrical machines are generally based on the either advance observers or special characteristic of motor, e.g. the saliency. However, it is difficult to find general methods that are possible or suitable to apply to various drives. The estimation is possible by constructing a state observer based on motor electrical and mechanical equations. Then stability of the observer is key point position estimation. Adaptive control seems to be the most promising one of various modern control strategies. The MRAC approach is capable of compensating the variations of the system parameters, such as inertia and torque constant with reduced computation. The estimators based on the Model Reference Adaptive System (MRAS) provide the desired state from two different models, one is used as reference model and another one is as adjustable model. The error between reference model and adjustable model is used for estimation of the unknown v parameter (speed in this case). In MRAS only adjustable model is dependent on speed parameter, while, reference model is not dependent on speed. The error signal is fed into adaptation mechanism, which provides the estimated quantity and is used to tune the adjustable model. This method is simple and requires less computation. Among the existing sensorless schemes, the sliding mode has been recognized as a prospective estimation methodology for the drives. Sliding mode observer (SMO) has remarkable advantages of robustness against external disturbances and less sensitivity to motor parameter deviations. This method is an algorithm that uses the sign of the error on the machine current to update the value of the estimated speed and position stored in the controller. The major advantage of SMO is no requirement of extra electronics. Furthermore, the variation of motor parameters has little impact on the results accuracy. Increase in the mechanical robustness of drive and to reduce the cost of drive, eradication of the position sensor is encouraged. Sensors are not reliable in explosive environment viz. chemical industries and may cause the EMI problem. A speed estimator for PMSM drive has been proposed. In the algorithm the PMSM used as reference model. The adaptation mechanism uses a Fuzzy controller to process the error between reference model and adjustable model. The estimation method used is independent of stator winding resistance, computationally less complex, free from integrator problem because back-emf estimation is not used and provides stable operation of drive system. Thus, the performance at zero and low speed is also good. The proposed estimation algorithm is implemented in MATLAB. Simulations results show the validity of proposed fuzzy logic based MRAS and verified through experimental results. Further, sliding mode observer based position estimation for PMSM without saliency is presented. An equivalent control in the feedback is applied to extend the operating range and improve the estimation performance. In comparison to conventional SMO, the proposed algorithm gives the flexibility in selecting observer parameters for a wide speed range operation. The observer convergence in the high speed range is guaranteed, and the estimation error is reduced with proper selection of feedback gain. The chattering problem existing due to large switching gain at low-speed is reduced. The proposed SMO with equivalent control verified through simulation and laboratory results. In the experimentation dSPACE version DS1104 is used for prototyping. Three phase inverter, used here, is an intelligent power module (PEC16DSM01) make Vi Microsystems. This rapid control prototyping (RCP) consists of software and hardware. The hardware is consisting of DS1104 R&D controller board with digital and analogue, input-output facility. All signals in CPL1104 can be monitored by status LEDs. The software consists of real time interface (RTI) blocks which connect the simulink controller to the real-time hardware. vi Moreover, Control Desk is used to conduct experiment, adjust the controller parameters, and to visualize the desired signals involved in the experimentation. The motor used here is with inbuilt resolver, which provides the sine and cosine of rotor position at its two output windings. These sine and cosine signals are given to dSPACE through ADC channels. Then these signals are used to calculate rotor position in degrees. A square wave signal of 3 kHz, with -5V to +5V amplitude generated from dSPACE, and given to primary winding of resolver through DAC channel, these sine and cosine signals are generated at resolver two secondary windings. The voltage sensor using isolation amplifier AD202JN is used to sense motor terminal voltages. The input voltage to dSPACE is reduced in the range of ±10 V. The voltage sensors are calibrated to convert ±100 V to ±1 V. The current sensors are inbuilt with power module. These sensed voltages and currents are fed to software controller in dSPACE through ADC channel for further processing and calculations. The controller part of the drive system is implemented in MATLAB using RTI block sets, and appropriate signals are generated in real time.