单相逆变电源的输出电压波形质量是衡量其性能的重要指标之一.本文提出了一种正弦逆变波形的神经网络内膜控制算法,建立两个BP神经网络预估器,一个作为单相逆变器的内部模型,预测实际的波形输出;一个根据预测误差建立内模控制器,在线修正和补偿内部模型使之最大程度的匹配单相逆变器.仿真和实验结果表明,该算法克服了系统中存在的不确定性,有效的提高了系统的逆变波形质量和负载适应性.%The quality of output waveform is one important factor for the single-phase inverter. This paper proposed the algorithm based on neural network internal model theory, which is used for an output sine waveform control.Based on this control algorithm, two back propagation estimate neural networks were established. One is the internal model of single-phase inverter, which is used to estimate the actual output waveform. The other one is used to make the internal model to fit the actual single-phase inverter due to the estimate error. We finished the simulation and experiment, where the algorithm was proved that it could improve the output waveform quality and load compatibility.
展开▼
