Performance Evaluation Of Space Vector Modulation (SVM) For Multilevel Inverters

• Auteur principal: Sanusi, Syamim
• Format: Thèse
• Langue: anglais; anglais
• Publié: 2016
• Sujets: T Technology (General); TK Electrical engineering. Electronics Nuclear engineering
• Accès en ligne: http://eprints.utem.edu.my/id/eprint/18383/; https://plh.utem.edu.my/cgi-bin/koha/opac-detail.pl?biblionumber=100225

The Space Vector Modulation (SVM) technique has gained wide acceptance for many AC drive applications, due to a higher DC bus voltage utilization (higher output voltage compared with the Sinusoidal Pulse Width Modulation (SPWM)), lower harmonic distortions and easy digital realization. In recent years, the SVM technique was extensively adopted in multilevel inverters since it offers greater numbers of switching vectors for obtaining further improvements of AC drive performances. However, the use of multilevel inverters associated with SVM increases the complexity of control algorithm (or computational burden), in obtaining proper switching sequences and vectors. The complexity of SVM computation causes a microcontroller or digital signal processor(DSP) to execute the computation at a larger sampling time. This consequently may produce errors in computation and hence degrades the control performances of AC motor drives. This thesis reports the performance evaluation of SVM for two-level of VSI, threelevel and five-level of Cascaded H-Bridge Multilevel Inverter (CHMI) and analyse indepth the accuracy performances of SVM computation and the performance evaluation in variable speed drive systems (i.e. Direct Torque Control (DTC) using SVM). The SVM modulator is implemented using a hybrid controller approach, i.e. with combination between the DS1104 Controller Board and FPGA. In such way, the computational burden can be minimized as the SVM tasks are distributed into two parts, in which every part is executed by a single controller. This allows the generation of switching gates performed by FPGA at the minimum sampling time