基于混合天鹰优化器的风力发电预测

Wind Power Forecasting Based on Mixed Aquila Optimizer

摘要:

为了解决天鹰优化器（AO,Aquila Optimization）集中在全局搜索导致的局部寻优能力略差、依赖初始种群质量和易陷入局部最优的问题，提出一种多策略混合的天鹰优化器（MAO,Mixed Aquila Optimizer）；该算法利用改进的Hooke-jeeves优化基本天鹰优化器的初始化种群质量；引入模拟退火概率对易陷入局部最优解进行改进；自适应权重提高前期全局搜索效率，延缓后期局部搜索速度，避免在正解附近徘徊；选取12个基准测试函数进行实验，并将MAO应用于风力发电预测模型优化；实验结果表明，对于单峰函数、多峰函数和固定维函数，MAO比AO等对比函数具有更快的收敛速度和更高的精度；在春夏秋冬数据集上进行仿真实验，对比其他模型1月和10月预测精度提高了15%，4月和8月的预测曲线更加平滑；证实了MAO对于提高风电预测的精度和速度的可行性和实用性。

Abstract:

In order to solve the problem that the aquila optimizer algorithm concentrates on global search resulting in slightly poor local optimization ability, relies on the quality of the initial population and is prone to fall into local optimum, a multi-strategy mixed aquila optimizer is proposed.The algorithm uses improved Hooke-jeeves to optimize the initialized population quality of the basic aquila optimizer.The introduction of simulated annealing probability improves the easy to fall into the local optimal solution and adaptive weighting improves the efficiency of the global search in the early stage and slows down the local search in the late stage to avoid hovering around the positive solution.Twelve benchmark test functions are selected for experiments and MAO is applied to wind power prediction model optimization.The experimental results show that for single-peak, multi-peak and fixed-dimension functions,MAO has faster convergence speed and higher accuracy than comparative functions such as AO.Simulation experiments on spring, summer, fall and winter datasets,compared with other models,the prediction accuracy in January and October is improved by 15%,and the prediction curves in April and August are smoother.It confirms the feasibility and practicability of MAO for improving the accuracy and speed of wind power prediction.