摘要
传统的格拉斯曼流形状态估计是将状态空间模型置于格拉斯曼流行上,在后验跟踪模型的基础上递推估计。但是,该方法将观测模型和噪声都限制在格拉斯曼流形上会产生较大误差。针对上述问题,提出一种基于格拉斯曼流形的粒子滤波算法(Grass-Mann Manifolds-Paticle Filter,GM-PF),利用粒子滤波算法估计流形上的隐马尔科夫过程。仿真实例表明,该方法显著提高了流形上的隐马尔可夫过程的估计精度,且当噪声逐渐增大时,粒子滤波算法表现出良好的抗干扰能力和跟踪性能。
The traditional Grass-Mann manifold state estimation is to put the state space model on the Grass-Mann epidemic,and implement recursive estimation based on the posterior tracking model.However,this method restricts the observation model and noise to the Grass-Mann manifold,which will produce larger errors.Aiming at the above problem,a particle filter algorithm(Grass-Mann Manifolds-Particle Filter,GM-PF)based on Grass-Mann manifold is proposed.It uses a particle filter algorithm to estimate the hidden Markov process on the manifold.Simulation results indicate that this method significantly improves the estimation accuracy of the hidden Markov process on the manifold,and when the noise gradually increases,the particle filter algorithm shows good anti-interference ability and tracking performance.
作者
赵成
苏岭东
马祥林
ZHAO Cheng;SU Ling-dong;MA Xiang-lin(Hangzhou BONUAP Intelligent Technology Co.,Ltd.,Hangzhou Zhejiang 310012,China;Power Dispatch and the Control Center,State Grid Xuzhou Power Supply Company,Xuzhou Jiangsu 221000,China;Changzhou Zhike Automation Technology Co.,Ltd.,Changzhou Jiangsu 213001,China)
出处
《通信技术》
2020年第3期606-610,共5页
Communications Technology
关键词
格拉斯曼流形
粒子滤波
隐马尔科夫模型
随机过程
Grass-Mann Manifolds-Particle Filter
particle filter
hidden Markov model
stochastic process
