基于AEPSO-BPNN的光伏阵列多场景参数辨识

针对光伏阵列内部机理较为复杂、参数难以快速准确辨识的问题，提出了一种自适应进化粒子群算法优化BP神经网络（AEPSO-BPNN）的模型建立和参数辨识方法。通过引入自适应、进化和重构等改进策略，可以提高粒子群算法的收敛性能，并将其对BP神经网络的初始权值和阈值进行优化，使神经网络算法在迭代后期不易陷入局部最优解，以提高参数辨识的精确度和速度。根据光伏阵列的实测输出电流和理论计算电流的差值，并考虑环境变化对内部参数的影响，构造均方根误差函数作为算法的适应度函数，从而将复杂的多参数辨识问题转化为带约束条件的非线性多变量最优化问题。最后采用多场景法，验证算法在不同光照强度和温度下的适用性和效果，并与其他算法进行对比，仿真结果表明该算法在误差、收敛速度和运行时间上有较大优势。

Multi-scene Parameter Identification of Photovoltaic Array Based on AEPSO-BPNN

In view of the complex internal mechanism and the difficulty in fast and accurate parameter identification of photovoltaic array， this paper proposes an adaptive evolutionary particle swarm optimization algorithm to optimize BP neural network model establishment and parameter identification method. By introducing adaptive， evolutionary and reconstruction strategies， the convergence performance of particle swarm optimization algorithm can be improved， and the initial weights and thresholds of BP neural network can be optimized， so that the neural network algorithm do not easily fall into the local optimal solution in the later iteration period，thereby improving the accuracy and speed of the parameter identification. According to the difference between the actual measured output current and the theoretical calculated current of the photovoltaic array， considering the effect of environment change on internal parameters， the root mean square error function is constructed， as the fitness function of the algorithm， thus complex multi-parameter identification problem is transformed into a nonlinear multi-variable optimization problem with constraints. Finally， multi-scene method is used to verify the applicability and effect of the algorithm under different illumination intensity and temperature， and the simulation results show that the algorithm has the decided advantages in error，convergence speed and running time by comparing the results from other algorithms.