摘要
随着全国碳排放权交易市场的启动,碳价的预测对碳市场参与者的风险管理具有重要意义。针对BP神经网络收敛速度慢、容易陷入局部极值的弊端,结合混沌的遍历性,构建基于混沌粒子群(CPSO)算法优化BP神经网络的碳价预测模型:利用Elastic Net方法降维,筛选出碳价的主要影响因素;再用CPSO优化BP神经网络的初始权值和阈值训练模型并预测碳价,结果表明:CPSO-BP碳价预测模型的精度和稳定性明显优于传统BP神经网络、粒子群优化的BP神经网络以及果蝇算法优化的BP神经网络。
With the start of the national carbon emissions trading market,the prediction of carbon emission price is of great significance to the risk management of the carbon financial market participants.BP neural network for forecasting has low speed of convergence and easily falling into the local state.Combined with the advanced searching ability of chaos,aprediction model of BP neural network based on chaotic particle swarm optimization（CPSO）algorithm is constructed：firstly,reducing the dimensions by Elastic net and screening out the main influencing factors of the carbon price;then the initial weights and thresholds of BP neural network are optimized by CPSO to train the model and predict the carbon price.The results show that the CPSO-BP model can improve the accuracy and stability of traditional BP neural network,and overcomes the easily falling into the local state of traditional PSO algorithm.
作者
蒋锋
彭紫君
JIANG Feng;PENG Zi jun(School of Statistics and Mathematics, Zhongnan University of Economics and Law,Wuhan 430073, Chin)
出处
《统计与信息论坛》
CSSCI
北大核心
2018年第5期93-98,共6页
Journal of Statistics and Information
基金
国家自然科学基金面上项目<具Levy噪声的状态切换随机时滞系统控制及在基因调控网络中的应用>(61773401)
国家自然科学基金青年项目<非随机化响应技术中logistic回归及相关问题的研究>(11601524)
湖北省人文社会科学研究项目<湖北县域金融生态评价指标体系的构建与实证研究>(17G024)
关键词
碳排放权
ELASTIC
NET
混沌粒子群算法
BP神经网络
果蝇算法
carbon emissions
Elastic net
chaotic particle swarm optimization
BP neural network
fruit fly optimization algorithm
