【摘要】 针对多电平二极管钳位逆变器的传统SVM算法需要进行大量三角函数运算或查表操作、因而运算效率低的不足,引入基于Kohonen竞争性神经网络的分类算法,提出了一种通用的快速多电平二极管钳位逆变器SVM算法.新算法并不需要对神经网络进行训练,在整个实现过程中不再需要任何三角函数计算或查表操作,而只需要进行简单的加减乘除运算,因而能显著简化算法的实现.对三电平和五电平的二极管钳位逆变器进行时域仿真,结果验证了文中数学分析的正确性和所提算法的可行性,同时表明该算法是一种对多电平二极管钳位逆变器通用的算法,将它应用于不同电平数的逆变器时不需要进行任何修改.更多还原

【Abstract】 To overcome the shortcomings of the conventional SVM for diode clamped multilevel inverter,a classification algorithm based on Kohonen’s competitive NN was propsed.Although the algorithm is an NN-based one,it does not need a training stage.The proposed SVM switching strategy eliminates the time consuming operations,so the whole procedure for the implementation of the algorithm is carried out with simple mathematical operations.The validity of mathematical analysis and the feasibility of the proposed algorithm were verified with simulation studies for both a three-level DCI and a five-level DCI.The simulation studies also verify that the proposed algorithm is a generalized one for an n-level DCI and does not need any modification when the number of levels increases.更多还原