PERFORMANCE INVESTIGATION OF SWITCHED RELUCTANCE MOTOR DRIVE

Abstract

The Switched Reluctance Motor (SRM) offers significant benefits over conventional AC and DC motors which include higher efficiency over wide speed range operation and under partial loading. Even though it exhibits high starting torque, high power density, high torque to weight ratio, smaller size and weight, low material cost and maintenance free operation, its acceptance for industrial as well as domestic applications is limited. It is due to higher torque ripple for which SRM is known. It also produces acoustic noise as well as it requires costly position sensing arrangement to be attached with the motor for control. However, the recent quest of the energy efficiency has promoted the research and development of the cost effective efficient SRM drive. In the present work the performance of the SRM drive and the suitability of the same for the various applications are investigated. Extensive simulation is carried out to evaluate the performance of the SRM drive and experimentally verified with the help of prototype model. The magnetic non-linearity of the SRM increases the complexity of modelling and controller design. Different effective modelling techniques for the SRM have been proposed by the researchers in the past. In the present work few vital non-linear modelling techniques have been exemplified and compared. The embedded control function is designed in MATLAB simulink environment to model a SRM based on analytical modelling technique. The concept of the physical modelling technique of the system is introduced which offers many advantages over conventional modelling techniques. The main advantage of the physical modelling is its ability to consider the effect of change in physical parameters of the system. In the present work the physical modelling technique is proposed to model the non-linearity of the switched reluctance motor. The variable reluctance characteristic of the motor is utilized to derive the non-linear model of the switched reluctance motor. The mathematical relation amongst the phase reluctance, angular rotor position and geometrical parameters of the motor is derived. The results show that the proposed modelling technique which considers magnetic non-linearity of the motor offers acceptable solution. It also requires only few geometrical parameters of the motor and information of the winding turns per phase. Furthermore, the 3-Dimensional (3D) visualization effect is added to the physical model of the motor with the help of virtual reality tool box of the MATLAB which extends the capabilities of the simulink software into the world of virtual reality graphics. The high grade real-time simulation of the complete SRM drive is developed with the help of Opal's RT-Lab technology, which eliminates the issue of model latency present in CPU based simulation due to complex model and current overshoot. The drive is comprised i of asymmetric bridge converter, hysteresis current controller and a proportional and integral (PI) speed controller. The effect of commutation angle on the torque-speed characteristics of the switched reluctance motor drive is investigated with the help of real-time experimental studies. There are applications where the issue of torque ripples of the switched reluctance motor is a major concern e.g. robotic actuators and electric power steering. Efforts are being made into both the motor design and control strategy to reduce the ripples in the torque. The fundamental electronic approach of torque ripple minimization is the current profiling which stores the non-linear relation of torque-current-angle as a lookup table. Then the rotor angle is being used as an index to generate the reference current. Accordingly, the phase current is being controlled by means of hysteresis current controller to maintain the torque to the reference value. Effort have been made by the researchers in the past to replace the lookup table by the mathematical relation for representing non-linear relation of the torque, current and angle. It employs the fuzzy logic controllers, artificial neural networks or the Fourier series based methods. Another vital approach of the torque ripple minimization is the use of torque-sharing-function (TSF) in which the phase current applied is controlled such that, the overall torque becomes constant. More or less all the torque ripple reduction techniques increase the complexity and cost of the drive. However the simplest solution to the torque ripple issue is to use of motor having higher number of phases or the higher stator/rotor pole configuration. The low-cost switched reluctance motor drive is developed in the present work. In the developed model multi-phase excitation scheme is used to reduce the torque ripples. The proposed switched reluctance motor drive includes 8/6 pole four phase motor, one-switch per phase type converter and low-cost position sensing arrangement. The low-frequency PWM controller has been proposed and proven to be effective in noise reduction compared to hysteresis current controller. It also reduces the converter losses and makes the drive immune to electromagnetic interference. The conventional proportional and integral (PI) speed controller can be used for the switched reluctance motor drive, if transfer function of the non-linear switched reluctance motor is known. Due to non-linearity of SRM, it is difficult to derive the transfer function and hence, the design a controller with proper gains to achieve the desired stability. The ac small signal modelling technique is proposed for linearization of the SRM model. Accordingly, the closed-loop speed PI controller is designed for PWM current controlled SRM drive. Performance of the proposed SRM drive is investigated with the development of prototype SRM, converter and controller. The proposed SRM drive shows lower cost-versus-performance ratio and ensures higher efficiency of the drive even at partial loading. The torque ripple performance of the drive is investigated for the varying speed, load torque and PWM frequency. The investigation shows that the torque ripples are ii quite low for the higher speed application as well as for the applications having heavy load inertia. It also shows that the selection of higher PWM frequency is advantageous in reduction of torque ripples for the light load applications; however it increases the converter losses. The requirement of position feedback necessitates the mechanical position sensors to be attached with the motor which increases the overall cost and reduces the reliability of the operation. Many sensorless control techniques are reported in literature to eliminate the requirement of mechanical position sensors, but none of them is found suitable for the wide speed control range without having any constraint on the performance of the drive. It includes the waveform monitoring, state observer, active probing, modulated signal injection and flux-linkages based rotor position estimation techniques. In the present work a technique based on flux-linkage characteristics is investigated in detail with the help of simulation studies. In the present work the fixed turn-off method of rotor position estimation is proposed and investigated for speed control which is low-cost and simple solution for the general purpose drive. The commutation instant of the phase is estimated by monitoring the phase voltage and current, while it is not intended to derive continuous rotor position estimation. The method is suitable where the fixed angle control is sufficient but the harsh environmental condition restricts the use of mechanical sensors for reliability issue. To get rid of rotor position sensing, an ANN based sensorless rotor position estimation technique is proposed where the ANN is used to map the magnetic non-linearity of the motor. The neural network is designed to approximate the continuous rotor position form the flux-linkage and current. The feed-forward neural network structure is used to design a network while the Levenberg-Marquardt technique is used as training algorithm. The proposed method employs only one hidden layer and the training process is accomplished offline. The performance of the ANN is investigated to approximate the non-linear magnetic characteristics as well as the performance of the ANN based sensorless drive is investigated which shows the higher accuracy of the estimation. However the implementation of the ANN based method is computation intensive, time consuming and also pre-requires information of the motor magnetic characteristic. An analytical sensorless method is proposed which reduces the computation burden and requirement of the storage space. It requires the information of the motor's magnetic characteristics at only two rotor positions that are aligned position and mid position. The simple mathematical expression is derived to estimate the continuous rotor position estimation form the phase voltage and phase current. The performance of the proposed sensorless scheme is investigated which shows the good accuracy of estimation. iii The sensorless techniques based on magnetic characteristics require prior knowledge about motor magnetic characteristics which not only increase the time and cost of implementation, but also restrict the operation of the drive to the particular motor. Even the magnetic properties of the two motors from the same manufacturer and having same specifications could not be identical. The sensorless rotor position estimation techniques that do not require any prior knowledge of the motor magnetic characteristics includes chopping current waveform method, regenerative current method and current gradient sensorless method (CGSM). The design and performance issues of the CGSM are investigated to incorporate with the low-frequency PWM controlled SRM drive. Result shows that the use of phase-lock-loop (PLL) and non-synchronized PWM increases the complexity of implementation and reduces the reliability of the CGSM. It is also observed that the response of the system is unstable to the step change in load-torque. A Modified-CGSM (MCGSM) technique is proposed which eliminates the requirement of the PLL and improves the stability of the system. Prototype setup of the sensorless SRM drive is developed in the laboratory and its performance is investigated under open-loop as well as closed-loop condition. The study is carried out to exemplify the features and advantages of the SRM drive for a specific application. The switched reluctance motor is found favourable in the application of battery operated vehicle system because of higher efficiency, high starting torque, light weight, low maintenance and capability of electric braking. Furthermore the issue of the torque ripple is also not predominant due to the higher inertia loading. The performance of the low-frequency PWM controlled SRM drive in a low power in-wheel electric vehicle is investigated in detail. The feasibility of the electric braking with the fixed angle controller is explored. The investigation summarized that, the SRM offers an energy efficient solution for the wide range of applications. However, the complete success of the SRM drive needs proper choice of application, converter, control scheme and position sensing arrangement.