Modulated Model Predictive Control with Optimized Overmodulation

Cristian F. Garcia, Cesar A. Silva, Jose R. Rodriguez, Pericle Zanchetta, Shafiq A. Odhano

Research output: Contribution to journalArticle

14 Citations (Scopus)


IEEE Finite Set Model Predictive Control (FS-MPC) has many advantages, such as a fast dynamic response and an intuitive implementation. For these reasons, it has been thoroughly researched during the last decade. However, the wave form produced by FS-MPC has a switching component whose spread spectrum remains a major disadvantage of the strategy. This paper discusses a modulated model predictive control that guarantees a spectrum switching frequency in the linear modulation range and extends its optimized response to the overmodulation region. Due to the equivalent high gain of the predictive control, and to the limit on the voltage actuation of the power converter, it is expected that the actuation voltage will enter the overmodulation region during large reference changes or in response to load impacts. An optimized overmodulation strategy that converges towards FS-MPC’s response for large tracking errors is proposed for this situation. This technique seamlessly combines PWM’s good steady-state switching performance with FS-MPC’s high dynamic response during large transients. The constant switching frequency is achieved by incorporating modulation of the predicted current vectors in the model predictive control of the currents in a similar fashion as conventional Space-Vector Pulse Width Modulation (SV-PWM) is used to synthesize an arbitrary voltage reference. Experimental results showing the proposed strategy’s good steady-state switching performance, its FS-MPC-like transient response and the seamless transition between modes of operation are presented for a permanent magnet synchronous machine drive.
Original languageEnglish
JournalIEEE Journal of Emerging and Selected Topics in Power Electronics
Publication statusPublished - 19 Apr 2018

Fingerprint Dive into the research topics of 'Modulated Model Predictive Control with Optimized Overmodulation'. Together they form a unique fingerprint.

  • Cite this