基于受限玻尔兹曼机和粗糙集的风速区间概率预测模型  

INTERVAL PROBABILITY PREDICTION MODEL FOR WIND SPEED BASED ON RESTRICTED BOLTZMANN MACHINE AND ROUGH SET

作　　者：于晓要 李娜[1] Yu Xiaoyao;Li Na(Shangqiu Institute of Technology,Shangqiu 476000,Henan,China)

机构地区：[1]商丘工学院,河南商丘476000

出　　处：《计算机应用与软件》2023年第3期157-166,240,共11页Computer Applications and Software

摘　　要：针对风速的不确定性、时变和非线性特征,提出一种用于风速预测的基于受限玻尔兹曼机和粗糙集理论的区间概率分布学习(Interval Probability Distribution Learning, IPDL)模型。该模型包含一组区间隐藏变量,利用Gibbs抽样和对比散度来获取风速的概率分布,结合模糊Ⅱ型推理系统(Fuzzy Type Ⅱ Inference System, FT2IS),设计一个有监督回归的实值区间深度置信网络(Interval Deep Belief Network, IDBN)。算例结果表明,该方法结合了IPDL和FT2IS的鲁棒性,风速预测性能较好。In view of the uncertainty,time-varying and nonlinear characteristics of wind speed,this paper proposes an interval probability distribution learning(IPDL)model based on restricted Boltzmann machine and rough set theory for wind speed prediction.The model contained a set of interval potential variables.It used Gibbs sampling and contrastive divergence to obtain the probability distribution of wind speed.Combined with the fuzzy type II information system(FT2IS),we designed an interval deep belief network(IDBN)with supervised regression.The example results show that the proposed method combines the robustness of IPDL and FT2IS,and its wind speed prediction performance is better.

关 键 词：受限玻尔兹曼机 粗糙集理论 风速预测 区间概率分布学习 人工神经网络 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术] TM76[电气工程—电力系统及自动化]