模糊化模型概率的IMM-SUPF机动面目标跟踪
|
摘要
为了提高跟踪系统对水面机动目标的跟踪能力,本文将水面目标建模为椭圆形面目标,提出一种模糊化模型概率的交互多模型(interacting multiple model,IMM)强无迹粒子滤波算法。首先,利用现代高分辨率雷达获得的面目标扩展测量,给出了基于面目标的跟踪测量方程。其次,将强无迹粒子滤波(strong unscented particle filter,SUPF)算法引入到IMM中得到IMM-SUPF。该SUPF算法利用强跟踪无迹卡尔曼滤波(strong tracking unscented Kalman filter,STUKF)产生粒子建议分布。由于STUKF采用渐消因子调整UKF的状态模型协方差和观测模型协方差的比例,使得建议分布更符合真实状态的后验概率分布,从而提高了IMM算法中子模型滤波器的估计精度。最后,基于模糊隶属度函数对粒子的模型概率进行模糊化,从而在提高真实模型滤波器中粒子模型概率的同时,减小非匹配模型滤波器中粒子模型概率,进而提高IMM算法的估计融合精度。Monte-Carlo仿真实验表明,相比于传统的基于质点目标的IMM-UPF算法,文中所提的基于面目标的IMM算法跟踪精度更高,且所提算法的误差超调量更小,收敛更快。此外,所提面目标IMM算法的跟踪精度也要高于面目标IMM-UPF算法。不同于传统的质点目标IMM算法,文中将水面目标建模为椭圆形面目标,并利用面目标扩展测量信息设计了模糊化模型概率的IMM-SUPF算法。该算法进一步提高了跟踪系统对水面机动目标的跟踪能力。
In order to improve the tracking ability of the surface maneuvering target,the surface target was modeled as an elliptical area,and an interacting multiple model(IMM) strong unscented particle filter based on fuzzed model probability was proposed.Firstly,a measurement equation of the elliptical area target was derived by using the extended area target measurements of modern radars.Then,the IMM-SUPF was obtained by introducing the strong unscented particle filter(SUPF) into IMM.Laterly,the strong tracking unscented Kalman filter(STUKF) was utilized to generate the proposal distribution in SUPF,and a fading factor was provided in STUKF for adjusting the proportion of state-model covariance and measurement-model covariance,making the proposal distribution more consistent with the posterior probability distribution of true state and the estimation precision of the sub-filters for IMM algorithm higher.At last,the model probabilities of particles were fuzzed by the proposed fuzzy membership function.In this way,the model probabilities of particles for filter with true target model was higher while the model probabilities of particles for the incompatible filters was lower.Accordingly,the fusion accuracy of the IMM algorithm was improved.Monte-Carlo simulation results demonstrated that the tracking precision of IMM algorithm for the area target was higher than that of IMM algorithm for the particle target,and the proposed IMM algorithm had faster convergence speed and slighter overshoot than the traditional one.Moreover,the developed IMM algorithm provided higher precision than the IMM-UPF for the area target.By modeling the surface target as an elliptical area and designing the fuzzed model probability based IMM-SUPF,the proposed IMM algorithm achieved better estimation performance than the traditional IMM algorithm for the particle target.
In order to improve the tracking ability of the surface maneuvering target,the surface target was modeled as an elliptical area,and an interacting multiple model(IMM) strong unscented particle filter based on fuzzed model probability was proposed.Firstly,a measurement equation of the elliptical area target was derived by using the extended area target measurements of modern radars.Then,the IMM-SUPF was obtained by introducing the strong unscented particle filter(SUPF) into IMM.Laterly,the strong tracking unscented Kalman filter(STUKF) was utilized to generate the proposal distribution in SUPF,and a fading factor was provided in STUKF for adjusting the proportion of state-model covariance and measurement-model covariance,making the proposal distribution more consistent with the posterior probability distribution of true state and the estimation precision of the sub-filters for IMM algorithm higher.At last,the model probabilities of particles were fuzzed by the proposed fuzzy membership function.In this way,the model probabilities of particles for filter with true target model was higher while the model probabilities of particles for the incompatible filters was lower.Accordingly,the fusion accuracy of the IMM algorithm was improved.Monte-Carlo simulation results demonstrated that the tracking precision of IMM algorithm for the area target was higher than that of IMM algorithm for the particle target,and the proposed IMM algorithm had faster convergence speed and slighter overshoot than the traditional one.Moreover,the developed IMM algorithm provided higher precision than the IMM-UPF for the area target.By modeling the surface target as an elliptical area and designing the fuzzed model probability based IMM-SUPF,the proposed IMM algorithm achieved better estimation performance than the traditional IMM algorithm for the particle target.
引文
[1]Angelova D,Mihaylova L.Extended object tracking using Monte Carlo methods[J].IEEE Transactions on Signal Processing,2008,56(2):825-832.
[2]Zhong Z W,Meng H D,Wang X Q.A comparison of posterior Cramer-Rao bounds for point and extended target tracking[J].IEEE Signal Processing Letters,2010,17(10):819-822.
[3]Meng H D,Zhong Z W,Wang X Q.Performance bounds of extended target tracking in cluttered environments[J].Science China Information Sciences,2013,56(7):1-12.
[4]Shi J,Li Y Y,Qi G Q,et al.Extended target tracking filter with intermittent observations[J].IET Signal Processing,2016,10(6):592-602.
[5]何友,修建娟,关欣,等.雷达数据处理及应用[M].3版.北京:电子工业出版社,2013:191-193.
[6]Foo P H.Combining the interacting multiple model method with particle filters for manoeuvring target tracking with a multistatic radar system[J].IET Radar,Sonar&Navigation,2011,5(7):697-706.
[7]Wu P L,Cai Y D,Wang B B.Satellite bearings-only tracking using extended Kalman particle filter[J].Infrared and Laser Engineering,2011,40(10):2008-2013.
[8]Guo Yue,Liu Xinxue,Wang Caihong.A study on IMMUPF of tracking ballistic missile at free-flight phase in mixed coordinate system[J].Journal of Ballistics,2015,27(1):12-17.[郭跃,刘新学,王才红.混合坐标系下跟踪自由段弹道导弹的IMM-UPF算法研究[J].弹道学报,2015,27(1):12-17.]
[9]Guo Zhi,Dong Chunyun,Cai Yuanli,et al.Time-varying transition probability based IMM-SRCKF algorithm for maneuvering target tracking[J].Systems Engineering and Electronics,2015,37(1):24-30.[郭志,董春云,蔡远利,等.时变转移概率IMM-SRCKF机动目标跟踪算法[J].系统工程与电子技术,2015,37(1):24-30.]
[10]Wang Xiaoxu,Zhao Lin,Xia Quanxi,et al.Strong tracking filter based on unscented transformation[J].Control and Decision,2010,25(7):1063-1068.[王小旭,赵琳,夏全喜,等.基于Unscented变换的强跟踪滤波器[J].控制与决策,2010,25(7):1063-1068.]
[11]He X,Wang Z,Wang X,et al.Networked strong tracking filtering with multiple packet dropouts:Algorithms and applications[J].IEEE Transactions on Industrial Electronics,2014,61(3):1454-1463.
[12]Speekenbrink M.A tutorial on particle filters[J].Journal of Mathematical Psychology,2016,73:140-152.
[2]Zhong Z W,Meng H D,Wang X Q.A comparison of posterior Cramer-Rao bounds for point and extended target tracking[J].IEEE Signal Processing Letters,2010,17(10):819-822.
[3]Meng H D,Zhong Z W,Wang X Q.Performance bounds of extended target tracking in cluttered environments[J].Science China Information Sciences,2013,56(7):1-12.
[4]Shi J,Li Y Y,Qi G Q,et al.Extended target tracking filter with intermittent observations[J].IET Signal Processing,2016,10(6):592-602.
[5]何友,修建娟,关欣,等.雷达数据处理及应用[M].3版.北京:电子工业出版社,2013:191-193.
[6]Foo P H.Combining the interacting multiple model method with particle filters for manoeuvring target tracking with a multistatic radar system[J].IET Radar,Sonar&Navigation,2011,5(7):697-706.
[7]Wu P L,Cai Y D,Wang B B.Satellite bearings-only tracking using extended Kalman particle filter[J].Infrared and Laser Engineering,2011,40(10):2008-2013.
[8]Guo Yue,Liu Xinxue,Wang Caihong.A study on IMMUPF of tracking ballistic missile at free-flight phase in mixed coordinate system[J].Journal of Ballistics,2015,27(1):12-17.[郭跃,刘新学,王才红.混合坐标系下跟踪自由段弹道导弹的IMM-UPF算法研究[J].弹道学报,2015,27(1):12-17.]
[9]Guo Zhi,Dong Chunyun,Cai Yuanli,et al.Time-varying transition probability based IMM-SRCKF algorithm for maneuvering target tracking[J].Systems Engineering and Electronics,2015,37(1):24-30.[郭志,董春云,蔡远利,等.时变转移概率IMM-SRCKF机动目标跟踪算法[J].系统工程与电子技术,2015,37(1):24-30.]
[10]Wang Xiaoxu,Zhao Lin,Xia Quanxi,et al.Strong tracking filter based on unscented transformation[J].Control and Decision,2010,25(7):1063-1068.[王小旭,赵琳,夏全喜,等.基于Unscented变换的强跟踪滤波器[J].控制与决策,2010,25(7):1063-1068.]
[11]He X,Wang Z,Wang X,et al.Networked strong tracking filtering with multiple packet dropouts:Algorithms and applications[J].IEEE Transactions on Industrial Electronics,2014,61(3):1454-1463.
[12]Speekenbrink M.A tutorial on particle filters[J].Journal of Mathematical Psychology,2016,73:140-152.