EFFECTIVE MODEL PREDICTIVE CURRENT CONTROL SCHEMES FOR PERMANENT MAGNET SYNCHRONOUS MOTOR DRIVE

Abstract

Out of the total electrical energy generated worldwide, 46% is consumed by electrical machines leading to the emission of 6040 megatons of carbon dioxide gas. Therefore, it is essential to use motors with higher efficiency for the conservation of energy, protection of the environment, and sustainable development. Permanent Magnet Synchronous Motor (PMSM) is favorable for industrial and transportation applications due to its compact size, higher torque to weight ratio, efficiency, and reliability. In recent years, with the progression in digital signal processing technology, an effective and advanced control strategy like Model Predictive Control (MPC) which is computationally exhaustive has received wider attention. It offers the feasibility to use multiple constraints, multiple objectives, and multiple variables while maintaining its simplicity and intuitiveness. The most prominent MPC schemes are Model Predictive Torque Control (MPTC) and Model Predictive Current Control (MPCC). The MPTC scheme has the objective function defined using the torque and flux variables. However, in MPCC, the cost function is defined using stator currents as control variables. This eliminates the problem of tedious tuning of the weighting factor. The MPCC scheme offers excellent dynamic performance through the indirect control of torque and flux variables using the stator currents. However, the main challenges observed in the MPCC scheme are larger ripples in torque and flux ripples under steady-state conditions. To overcome this, the application of two or more voltage vectors is widely researched. However, the main challenge for the real-time application of multiple voltage vector-based control schemes is the increased computational burden. Thus, the wider application of the MPC scheme is still under the radar due to its higher computational complexity. This thesis proposes control strategies to improve the steady-state performance of the MPCC-controlled PMSM drive and addresses the limitation of increased computational complexity. iii To improve the steady-state torque and flux performance of the MPCC controlled PMSM drive, a dual voltage vector concept is implemented. The cause of increased ripples in torque and flux is identified to be the application of a single voltage vector for the entire control period irrespective of the magnitude of the error between control variables. Thus, to control the magnitude of the optimum voltage vector, a null vector is added to it. The duration for which the optimum vector is applied is determined based on the deadbeat principle. The duty ratio calculation used is simple and less sensitive to parameter variations. The application of dual voltage vectors undeniably improves the performance of the drive at the expense of increased computational time. Thus, a low complex dual voltage vector application scheme is evaluated in this research, which reduces the number of voltage vectors used for prediction, cost function evaluation, and optimization to three. The voltage vector preselection does not require additional determination of sector or reference vectors. This reduces the computational time and with the application of an active and null voltage vector, the steady-state drive performance is improved. The duration of application of the active vector is determined using the rms ripple minimization technique. In certain operating conditions where the error magnitude is large, it is required to apply more than two voltage vectors in a control period. Thus, multiple voltage vector application strategies are investigated using the virtual voltage vector concept. However, with the augmentation of the control set using the virtual voltage vectors, there would be a catalyzed increase in the computational burden. To limit the increase in the computational complexity, a voltage vector preselection scheme is employed which effectively locates the optimum voltage vector that minimizes the error. The duration of application of voltage vectors is determined using the average error minimization technique. The application of two or more voltage vectors in a sample time can provide a better steady-state response with an increase in computational complexity. To address this, a simplified voltage vector selection-based MPCC is proposed which directly evaluates the optimum voltage vector without prediction, cost function evaluation, and optimization. The multiple voltage vector application is then achieved by determining the position of the first optimum voltage vector using a modified position determination approach. The proposed MPCC improves the iv steady-state performance with reduced computational complexity. The duration of active voltage vectors is directly evaluated using the cost function ratios. This reduces the parameter sensitivity of the calculation. The proposed MPCC schemes are developed using MATLAB/Simulink. The real-time implementation of the control schemes is achieved using the dSPACE 1104 control platform. The effectiveness of the control techniques is evaluated using comprehensive comparisons with the existing scheme