基于遗传算法改进的BP神经网络的近地气候数据雷电预测模型  

Improred BP Neural Network with Genetic Algorithm for LightningPrediction Based on Near-Ground Meteorological Data

作　　者：仲致诚 郑建勇[1,2] ZHONG Zhicheng;ZHENG Jianyong(School of Electrical Engineering,Suzhou Research Institute,Suzhou 215125,China;School of Electrical Engineering,Southeast University,Nanjing 210018,China)

机构地区：[1]东南大学苏州研究院,江苏苏州215125 [2]东南大学电气工程学院,江苏南京210018

出　　处：《徐州工程学院学报（自然科学版）》2023年第2期71-76,共6页Journal of Xuzhou Institute of Technology(Natural Sciences Edition)

基　　金：江苏省国际科技合作项目(BZ2021012)。

摘　　要：BP神经网络应用广泛,学习能力出众,但由于其训练速度较慢、容易受到极限的影响等原因,无法满足过于复杂的求解需求.相比之下,遗传算法可以有效地提高求解的复杂程度,并且可以通过对权重、阈值的调整来提升算法的准确性与可靠性.因此,文章提出了一种基于近地气候数据雷电预测模型,研究了网络参数对预测效果的影响.实验证明,该模型对比其他模型具有更好的预测能力和泛化能力.BP neural network has a wide range of applications and outstanding learning ability.However,it cannot meet the demand for solving problems that are too complex due to its slow training speed,susceptibility to local extremes and other factors.In contrast,genetic algorithms can effectively increase the complexity of problem solving and improve the accuracy and reliability of the algorithm by adjusting the weights and thresholds.Therefore,this paper proposes a lightning prediction model based on near-ground meteorological data.The experiment proves that this model has better prediction and generalization ability compared to other models.

关 键 词：遗传算法 雷电预测 神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] P427[自动化与计算机技术—控制科学与工程]