CAJ | 학술논문

基于贝叶斯滤波原理，介绍了粒子滤波（Particle Filter，PF）的基本思想和具体算法实现步骤。针对非高斯噪声对水下信号目标跟踪的影响，分别对符合高斯分布、韦伯分布和伽马分布的随机噪声序列，在噪声均值和方差相同的条件下，对比分析了扩展卡尔曼滤波（Extended Kaman Filter，EKF）算法和PF算法的估计精度。仿真结果表明，在非线性非高斯环境下EKF算法跟踪性能严重下降，而PF算法能继续保持较好的跟踪精度，证明PF算法在非线性非高斯系统中的有效性。
기우패협사려파원리，개소료입자려파（Particle Filter，PF）적기본사상화구체산법실현보취。침대비고사조성대수하신호목표근종적영향，분별대부합고사분포、위백분포화가마분포적수궤조성서렬，재조성균치화방차상동적조건하，대비분석료확전잡이만려파（Extended Kaman Filter，EKF）산법화PF산법적고계정도。방진결과표명，재비선성비고사배경하EKF산법근종성능엄중하강，이PF산법능계속보지교호적근종정도，증명PF산법재비선성비고사계통중적유효성。
Based on the principle of Bayesian filtering theory,the basic idea and algorithm description of Particle Filter (PF)are introduced. The estimation accuracy of Extended Kaman Filter (EKF)and PF in simulation experiments are compared and analyzed for random noise sequence with different distributions,including Gaussian distribution,Weibull distribution,and Gamma distribution, which had equal mean value and equal variance. The experimental results demonstrated that EKF algorithm’s performance degrades severely in the circumstance of nonlinear and non - Gaussian system model,while the PF also has good tracking accuracy,and confirms the effectiveness of the PF in the nonlinear and non- Gaussian system.