中文摘要：

针对SVPWM计算复杂等问题,根据双三角形合成原理,采用ANN实现光伏并网逆变器的SVPWM。在分析推导两种模式下开关矢量的占空比计算公式的基础上,合理构造欠调制模式子网络和过调制模式子网络,通过对各个神经网络单独训练,实现由欠调制状态到过调制状态的平滑过渡。并通过限定轨迹双调制模式法,克服了SVPWM过调制的非线性问题。MATLAB仿真实验结果表明,ANN-SVPWM简便、高效,能实现输出电压连续控制,有效改善三相光伏并网逆变器输出性能。

英文摘要：

To solve questions of SVPWM,a new controller based on artificial neural network（ANN） scheme for PV grid-connected inverter is presented.The duty-cycle of three-phase in this two modulation modes is computed based on double input line-to-line voltage composition,and sub-networks of under-modulation mode and over-modulation mode are built,and every ANN controller is trained lonely and the extending smoothly from under-modulation mode up to over-modulation mode is finished.The un-linearity questions of over-modulation are solved by using theory of double-mode modulation based on limited trajectories.The simulation model has been developed based on MATLAB,together with the neural network toolbox.The simulation results show that ANN-SVPWM has features of simplicity,high efficiency,and the continuous controller both under-modulation period and over-modulation period,so the excellent control effects of photovoltaic grid-connected inverter can be gotten.