基于球形解码的三电平PWM整流器多步长模型预测控制

Multi-step Model Predictive Control of Three-level PWM Rectifier Based on Spherical Decoding

PWM整流器作为"绿色电能变换装置",广泛应用于有源电力滤波、电气传动及新能源发电等领域。文章针对三电平PWM整流器传统的单步长模型预测控制易陷入局部最优的问题,以及传统的多步长模型预测控制存在计算量大的缺陷,提出一种基于球形解码的多步长模型预测控制策略。该算法先将多步长模型预测控制的价值函数转化为最小二乘问题,然后采用球形解码算法对最小二乘问题进行求解,快速解出多步长内的最优开关序列,从而减少寻优算法的计算量,同时降低系统的平均开关频率,并在算法中对整流器直流侧中点电位进行约束,保证中点电位的平衡。最后,通过仿真和实验平台验证了基于球形解码的多步长模型预测控制的实用性和有效性。

As a green electric energy conversion device,PWM rectifier is used widely in the fields of active power filtering,electric drive and new energy power generation. Aiming at the problem that traditional single step model predictive control(MPC)of three-level PWM rectifier is easy to fall into local optimum and the traditional multi-step MPC has the defect of large amount of calculation,a kind of multi-step MPC strategy based on spherical decoding is proposed. First,the value function of the multi-step MPC is transformed into the least square problem,and then the spherical decoding algorithm is used to solve the least square problem to solve quickly the optimal switching sequence in the multi-step range. This reduces greatly the calculation amount of the optimization algorithm and,at the same time,reduces the average switching frequency of the system. Also,the neutral point potential at the DC side of the rectifier is constrained in the algorithm so to keep the balance of the midpoint potential. Finally,the practicability and effectiveness of multi-step MPC based on spherical decoding are verified by the simulation and experimental platform.