Finite Set MPC Algorithm for Achieving Thermal Redistribution in a Neutral-Point-Clamped Converter

Abstract

The three level neutral-point clamped (3L-NPC) topology is one of the most widely used multilevel topologies in the low and medium voltage applications and also one of the most commercialized topologies. Although it offers many benefits compared to the conventional two level topology, it suffers from a considerable unequal loss distribution among the inner and outer switches and the clamping diodes. To solve this problem, we are proposing a control algorithm based on the finite control set model predictive control (FCS-MPC) that can provide a more balanced stress distribution. For implementing the proposed control algorithm no additional measurements are required nor thermal models of the semiconductor devices. The algorithm benefits are even more noticeable during low voltage ride through (LVRT) scenarios when the output voltage level of the converter is low and the current amplitude is high. Obtained simulation results confirm the positive effects on the thermal redistribution and also the junction temperatures of the most stressed devices are reduced. Effects of the algorithm are also verified on an experimental setup .

FS-MPC Based Thermal Stress Balancing and Reliability Analysis for NPC Converters

“…In Table 1, eight thermal stress distribution methods that can be applied for multilevel converters were selected for a comparison [9], [13], [16], [22], [24]- [29]. The methods are divided into three categories based on the control method: space vector control [24], [25], carrier-based PWM [9], [26]- [28] and model predictive based control [13], [22], [29], [30]. For all the methods the requirements for implementation and limitations are provided.…”

“…The first two solutions are not very practical as they require expensive high-bandwidth temperature measurement equipment or additional circuity [31]. Furthermore, some of the mentioned solutions have a limited application [22], [24], [26].…”

“…It can also be observed in Table 1, that it will be necessary to find a tradeoff between the system performance and the effectiveness of the thermal stress distribution. For predictive control methods in [13], [22], [29], [30], this is done by adjusting the weighting factors in the cost function. However, the multi objective cost function with four weighting factors presents a very complex optimization problem.…”

“…On the other hand, some methods like [15] require a powerful control platform. From the MPC methods, [22] has the simplest implementation and only two weighting factors. The tuning of the weighting factors has to be adjusted to the operating conditions of the converters using one of the advanced weighting factor tuning methods [32].…”

“…This work presents a wider investigation of the FS-MPC strategies for pure Si and hybrid NPC converters, presented by the authors in conference papers [21], [22]. Moreover, the FS-MPC algorithm is applied for the first time to an h-NPC structure, and its advantage is demonstrated.…”