Adaptive Model Predictive Controller to Reduce Switching Losses for a Capacitor-Less D-STATCOM

In the post-industrial digital-economy, motors powering manufacturing have been replaced by computers powering ecommerce - with significant financial losses attributable to poor power quality (PQ). At the same time, increasing penetration of distributed generation and constant power loads complicate the job for the electric utilities to deliver high-quality electrical power. Traditionally, voltage source converters provide PQ compensation; the backbone of these converters, however, are large electrolytic capacitors (e-caps), which have well-known failure rates. A D-STATCOM that does not use e-caps was shown to increase the converter service life and increase the reliability of the power system; finite control set model predictive control (MPC) was shown to achieve high fidelity tracking for multi-objective cost functions. However, high fidelity (to achieve low total harmonic distortion) results in high switching frequency, which increases losses and device stresses, and reduces the overall converter reliability. This paper improves the capacitorless D-STATCOM by investigating an adaptive MPC that trades-off high-fidelity performance and switching frequency, subject to IEEE 519 THD standards, to achieve good-enough performance. The method is verified experimentally in a 7.5 kVA D-STATCOM hardware prototype. The results show a reduction in switching frequency by over 30% compared with the MPC methods that prioritize high-fidelity alone.