论文部分内容阅读
针对扰动观察法易在最大功率点附近振荡的问题,研究基于在线自调整神经网络和扰动观察法结合的最大功率跟踪策略。首先提出可在线调整结构的快速资源优化网络,然后以温度、光伏阵列电压和功率为网络输入,以最大功率点电压为输出,建立MPPT神经网络模型。若当前工作状态与神经网络模型记忆模式距离较近,则以神经网络输出U_N为U_(MPP)。否则,启用扰动观察法在U_N基础上精确定位最大功率点。同时将此运行状态数据作为样本训练调整网络结构,增加记忆模式,提高网络的输出精度。仿真实验证明本方法可避免系统在最大功率点的振荡。
Aiming at the problem that the perturbation observation is easy to oscillate near the maximum power point, the maximum power tracking strategy based on online self-tuning neural network and disturbance observation is studied. Firstly, a fast resource optimization network with online structure adjustment is proposed. Then MPPT neural network model is established by inputting temperature, PV array voltage and power as network inputs and outputting maximum power point voltage as output. If the current working state is close to the neural network model memory mode, the neural network output U_N is U_ (MPP). Otherwise, enable disturbance observation to precisely locate the maximum power point based on U_N. At the same time, this running status data is used as a sample training to adjust the network structure, increase the memory mode, and improve the output precision of the network. Simulation results show that this method can avoid the system oscillation at the maximum power point.