摘要
为提高GPS接收机的定位性能,首先对采样粒子数目进行研究,发现并不是采样粒子数目越多粒子滤波(Particle Filter,PF)的滤波效果就越好。然后针对PF算法中存在的粒子退化现象,研究PF算法与扩展卡尔曼滤波算法(the Extended Kalman Filter,EKF)处理加入了高斯噪声干扰的非线性模型,仿真并分析得到,PF在处理高斯非线性模型的时候滤波效果要优于EKF算法,试验中发现PF算法在粒子数目较大的时候滤波效果远远偏离真实值,试想通过EKF算法计算取得的均值和方差来引导PF算法进行下一步采样,以此建立较好的重要性密度函数。试验表明经过扩展卡尔曼滤波改进的粒子滤波算法相比PF算法更加精确,平缓性更好。
In order to improve the positioning performance of GPS receivers, the number of sampling particles is first studied. It is found that the more the number of sampling particles is, the better the filtering effect of particle filter（PF） is. Then for the particle degeneration phenomenon in the PF algorithm, the PF algorithm and the Extended Kalman Filter（EKF） processing are added to the nonlinear model of Gaussian noise interference. Simulation and analysis show that PF is dealing with Gaussian nonlinearity.The filtering effect of the model is better than that of the EKF algorithm. In the experiment, the PF algorithm found that the filtering effect far deviates from the true value when the number of particles is large. Imagine that the EKF algorithm calculates the mean and variance obtained to guide the PF algorithm for the next sampling. In order to establish a better importance density function.Experiments show that the improved particle filter algorithm with extended Kalman filter is more accurate and smoother than the PF algorithm.
作者
梁鹏
耿建平
Liang Peng;Geng Jianping(School of Elec tron ic Engineering and Automation,Guilin University of Electronic Technology,Guangxi,Guilin,541004,China;Changsha Haige Beidou Information Technology Co.,Ltd.,Changsha,410000,China)
出处
《仪器仪表用户》
2018年第8期32-35,共4页
Instrumentation
