A Sphere Decoding Algorithm for Multistep Sequential Model Predictive Control

TL;DR

This paper introduces a sphere decoding algorithm tailored for multistep sequential model predictive control in power electronics, enabling efficient solution of mixed-integer optimization problems for complex control tasks.

Contribution

A novel sphere decoding algorithm is proposed to solve the mixed-integer subproblem in sequential model predictive control, improving computational efficiency in power electronics applications.

Findings

The proposed method effectively solves the subproblem with mixed-integer constraints.

Numerical simulations demonstrate improved control performance over existing methods.

The approach is validated on a wind turbine system with promising results.

Abstract

This paper investigates the combination of two model predictive control concepts, sequential model predictive control and long-horizon model predictive control for power electronics. To achieve sequential model predictive control, the optimization problem is split into two subproblems: The first one summarizes all control goals which linearly depend on the system inputs. Sequential model predictive control generally requires to obtain more than one solution for the first subproblem. Due to the mixed-integer nature of finite control set model predictive control power electronics a special sphere decoder is therefore proposed within the paper. The second subproblem consists of all those control goals which depend nonlinearly on the system inputs and is solved by an exhaustive search. The effectiveness of the proposed method is validated via numerical simulations at different scenarios on a three-level neutral point clamped permanent magnet synchronous generator wind turbine system and compared to otherlong-horizon model predictive control methods