电网不平衡情况下基于神经网络并网逆变器同步技术研究

Research on grid synchronization of grid-connected inverter based on neural network under unbalanced voltage conditions

为解决常规锁相环技术在电网电压不平衡情况下难以对电网电压频率和相位进行有效检测的问题,提出一种电网不平衡情况下基于神经网络的并网逆变器同步算法。首先,在两相静止坐标系下推导电网电压状态方程,并基于此建立神经网络;然后,利用网络输出电压矢量和实际电压矢量误差进行在线调整权值,并利用权值调整计算在线辨识电网电压频率、相位和幅值,从而可以构建电网电压的正负序分量。仿真和实验结果表明:该方法能在电网不平衡情况下快速有效在线自适应辨识电网电压频率和相位,提取电网电压正负序分量,具有较强的鲁棒性。

To solve the problem of conventional phase locked loop technique under the condition of unbalanced power grid voltage, a grid inverter synchronous technology based on neural network under unbalanced power grid case is developed. At first, grid voltage state equation was derived in the two-phase sta- tionary coordinates, and a neural network was built based on the state equation ; the biases of output volt- age vector and the actual voltage vector were used to adjust the neural network weight online, and thus to find out the amplitude, frequency and the phase of the grid voltage, which can construct the positive and negative components of grid voltage. Simulation and experimental results show that the method can be online adaptive to identify the frequency and the phase of grid voltage quickly and efficiently in the case of unbalanced power grid, detect the positive and negative sequences of grid voltage, and has strong robustness.