摘要
传统的信息梯度算法虽具有计算量小的优点,但收敛速度较慢。为加速算法收敛,提出一种多误差信息梯度算法。该算法将信息梯度中使用的标量误差更换为多个误差组成的误差向量,使得收敛速度大为加快;同时,为获得更准确的误差概率密度函数,引入最优高斯核宽度。数值仿真和案例建模表明了算法的有效性。
Although the traditional information gradient algorithm costs small computation,its convergences is more slowly.To accelerate the algorithm,a multi-error information gradient algorithm is proposed.In this algorithm,the scalar error used in the information gradient is replaced by an error vector,which is composed of multi-errors.The convergence speed is accelerated greatly;At the same time,to obtain a more accurate probability density function of error,an optimal Gaussian kernel width is introduced.Numerical simulation and case study are shown the effectiveness of the algorithm.
作者
邱奕武
景绍学
QIU Yiwu;JING Shaoxue(Operation Department,Jiangsu Huaneng Huaiyin Power Generation Co.,Ltd.,Huai'an 223301;School of Physics and Electronic Electrical Engineering,Huaiyin Normal University,Huai'an 223300)
出处
《办公自动化》
2021年第7期10-12,25,共4页
Office Informatization
关键词
参数辨识
高斯核
核宽
信息梯度
多误差
收敛速度
parameter identification
gaussion kernel
kernel width
informatiion gradient
multi-error
convergence speed
