基于深度学习的联合等式状态约束辨识与递推滤波

doi:10.12382/bgxb.2024.0578

摘要/Abstract

摘要：

针对等式约束跟踪系统中多约束并存且当前约束信息不确定下状态估计问题,提出基于深度学习的联合等式状态约束辨识与递推滤波算法。利用门控循环单元构建约束判别网络,借助雷达量测实现当前时刻等式状态约束的在线辨识;在递推滤波框架下基于级联门控循环单元构建增益学习网络,实现概率模型与数据学习联合驱动的目标状态自适应估计;通过滤波投影联合约束判别网络辨识的约束信息和增益学习网络输出的状态估计,获得满足当前时刻等式状态约束的高精度目标状态估计。典型多道路目标跟踪实验结果表明:新算法相比卡尔曼滤波、交互式多模型(基于不同运动模型构建模式集/不同等式状态约束构建模式集)和KalmanNet等算法,在不同量测噪声水平下具有更高的估计精度和更好的鲁棒性。

关键词: 目标跟踪, 等式约束, 线性随机系统, 状态估计, 深度学习

Abstract:

The state estimation problem of equality constraint-based target tracking where multiple constraints coexist and the current constraint information is unknown is presented.A joint state equality constraint identification and recursive filtering based on deep learning algorithm is proposed.The gated recurrent units are used to construct a constraint discriminant network,and the current state constraint is identified online by using the radar measurement.A filtering gain learning network based on cascaded gated recurrent units is built in the framework of recursive filtering to adaptively estimate the target state with the help of joint probabilistic modeling and data learning.The final high-precision filtered estimate which meets the real state equality constraint at current sampling instant is obtained based on filtering projection,which combines the state estimate obtained by the gain learning network with the state equality constraint identified by the constraint discriminant network.Experimental results of multi-target tracking example demonstrate that the proposed algorithm outperforms Kalman filtering,interacting multiple model (based on different motion modes/different state equality constraints) and KalmanNet algorithm in terms of estimation accuracy and robustness,with different levels of measurement noises.

Key words: target tracking, equality constraint, linear stochastic system, state estimation, deep learning

中图分类号:

TN953

秦月梅, 陈重, 杨衍波, 李淑英. 基于深度学习的联合等式状态约束辨识与递推滤波[J]. 兵工学报, 2025, 46(6): 240578-.

QIN Yuemei, CHEN Zhong, YANG Yanbo, LI Shuying. Joint State Equality Constraint Identification and Recursive Filtering Based on Deep Learning[J]. Acta Armamentarii, 2025, 46(6): 240578-.

图/表 18

图1 DCIF的整体框图

Fig.1 Overall block diagram of DCIF

表1 CDNet网络参数

Table 1 Network parameters of CDNet

网络	输入层	隐含层	输出层
Input	n	16n²	8n²
GRU	8n²	4n
Output	4n	8n²	l

图2 GLNet内部DNN结构图

Fig.2 Internal DNN structure diagram of GLNet

表2 GLNet内部DNN网络参数

Table 2 Internal DNN network parameters of GLNet

网络	输入层	隐含层	输出层
MLP1	2m²	16m²	2m²
MLP2	m	4m	16m
$P^k \| k - 1$ GRU	2m²+16m	m²
MLP3	m²+n²	8(m²+n²)	m²+n²
MLP4	n	4n	16n
$S^k \| k - 1$ GRU	m²+n²+16n	n²
MLP5	m²+n²	16(m²+n²)	mn
MLP6	mn+m²+n²	16(m²+n²)	2m²

表2 GLNet内部DNN网络参数

Table 2 Internal DNN network parameters of GLNet

网络	输入层	隐含层	输出层
MLP1	2m²	16m²	2m²
MLP2	m	4m	16m
$P^k \| k - 1$ GRU	2m²+16m	m²
MLP3	m²+n²	8(m²+n²)	m²+n²
MLP4	n	4n	16n
$S^k \| k - 1$ GRU	m²+n²+16n	n²
MLP5	m²+n²	16(m²+n²)	mn
MLP6	mn+m²+n²	16(m²+n²)	2m²

图3 DCIF算法步骤

Fig.3 Flowchart of DCIF algorithm

图4 多道路约束下车辆运动场景

Fig.4 Vehicle motion scenario under multiple road constraints

图5 CDNet损失曲线

Fig.5 Loss curves of CDNet

图6 GLNet损失曲线

Fig.6 Loss curves of GLNet

图7 CDNet辨识测试集的准确率曲线

Fig.7 Accuracy curve of CDNet identifying test set

图8 ξ轴位置的1000次蒙特卡洛仿真

Fig.8 1000 Monte Carlo simulations of ξ axis position

图9 η轴位置的1000次蒙特卡洛仿真

Fig.9 1000 Monte Carlo simulations of η axis position

图10 ξ轴速度的1000次蒙特卡洛仿真

Fig.10 1000 Monte Carlo simulations of ξ axis velocity

图11 η轴速度的1000次蒙特卡洛仿真

Fig.11 1000 Monte Carlo simulations of η axis velocity

表3 对比算法的ARMSE

Table 3 ARMSEs of comparative algorithms

对比算法	ξ	η	$ξ ·$	$η ·$
KF	8.45	9.24	3.08	3.61
IMM	8.48	8.47	3.79	4.04
IMM-I	12.17	12.19	3.04	3.22
IMM-P	11.94	12.21	3.09	3.29
KNet	7.73	7.91	2.84	3.18
GLNet	7.23	7.08	2.66	2.83
CD-CF	6.41	5.95	2.40	2.46
CFEC	6.20	5.70	2.21	2.04
DCIF	6.13	5.66	2.32	2.37

表3 对比算法的ARMSE

Table 3 ARMSEs of comparative algorithms

对比算法	ξ	η	$ξ ·$	$η ·$
KF	8.45	9.24	3.08	3.61
IMM	8.48	8.47	3.79	4.04
IMM-I	12.17	12.19	3.04	3.22
IMM-P	11.94	12.21	3.09	3.29
KNet	7.73	7.91	2.84	3.18
GLNet	7.23	7.08	2.66	2.83
CD-CF	6.41	5.95	2.40	2.46
CFEC	6.20	5.70	2.21	2.04
DCIF	6.13	5.66	2.32	2.37

图12 不同r在ξ轴位置的ARMSE

Fig.12 ARMSEs at different r positions along ξ axis

图13 不同r在η轴位置的ARMSE

Fig.13 ARMSEs at different r positions along η axis

图14 不同r在ξ轴速度的ARMSE

VFig.14 ARMSEs at different r velocities along ξ axis

图15 不同r在η轴速度的ARMSE

Fig.15 ARMSEs at different r velocities along η axis

参考文献 34

[1]	韩静茹, 吴炎烜, 杨喆, 等. 针对机动目标的防空导弹引战配合建模与仿真[J]. 兵工学报, 2017, 38(增刊1):200-204.
	HAN J R, WU Y X, YANG Z, et al. Modeling and simulation of fuze-warhead matching of air-defense missile for mobile target[J]. Acta Armamentarii, 2017, 38(S1):200-204. (in Chinese)
[2]	闫文旭, 兰华, 王增福, 等. 基于变分贝叶斯的星载雷达非线性滤波[J]. 航空学报, 2020, 41(增刊2):724395.
	YAN W X, LAN H, WANG Z F, et al. Nonlinear filtering for spaceborne radars based on variational bayes[J]. Acta Aeronautica et Astronautica Sinica, 2020, 41(S2):724395. (in Chinese) doi: 10.7527/S1000-6893.2020.24395
[3]	YANG Y B, LIU Z G, QIN Y M, et al. Novel pseudo-linear Kalman filtering for 3D angle-only tracking in the presence of observer’s location errors[J]. Automatica, 2023, 155:111114.
[4]	YANG Y B, LIU Z G, QIN Y M, et al. Iteratively distributed instrumental variable-based pseudo-linear information filter for angle-only tracking[J]. ISA Transactions, 2023, 138:359-372.
[5]	张昀普, 单甘霖. 道路约束下多传感器协同地面目标跟踪的管理方法[J]. 兵工学报, 2022, 43(3):542-555. doi: 10.12382/bgxb.2021.0122
	ZHANG Y P, SHAN G L. Multi-sensor cooperative management for ground target tracking under road constraints[J]. Acta Armamentarii, 2022, 43(3):542-555. (in Chinese) doi: 10.12382/bgxb.2021.0122
[6]	熊光明, 罗震, 孙冬, 等. 基于红外相机和毫米波雷达融合的烟雾遮挡无人驾驶车辆目标检测与跟踪[J]. 兵工学报, 2024, 45(3):893-906. doi: 10.12382/bgxb.2022.0602
	XIONG G M, LUO Z, SUN D, et al. Object detection and tracking for unmanned vehicles based on fusion of infrared camera and MMW radar in smoke-obscured environment[J]. Acta Armamentarii, 2024, 45(3):893-906. (in Chinese) doi: 10.12382/bgxb.2022.0602
[7]	何友, 修建娟, 刘瑜, 等. 雷达数据处理及应用[M]. 第4版. 北京: 电子工业出版社, 2022:402-404.
	HE Y, XIU J J, LIU Y, et al. Radar data processing with application[M]. 4th ed. Beijing: Publishing House of Electronics Industry, 2022:402-404. (in Chinese)
[8]	YANG Y B, LIU Z G, QIN Y M, et al. Maximum correntropy two-filter smoothing for nonlinear systems with non-Gaussian noises[J]. EEE Transactions on Systems,Man,and Cybernetics:Systems, 2024, 54:3796-3808.
[9]	杨家铭, 潘悦, 王强, 等. 水下弱目标跟踪的深度学习方法研究[J]. 兵工学报, 2024, 45(2):385-394. doi: 10.12382/bgxb.2022.0842
	YANG J M, PAN Y, WANG Q, et al. Research on deep learning method for underwater weak target tracking[J]. Acta Armamentarii, 2024, 45(2):385-394. (in Chinese)
[10]	JIANG W, REN Y H, LIU Y, et al. Artificial neural networks and deep learning techniques applied to radar target detection:a review[J]. Electronics, 2022, 11(1):156.
[11]	段艳杰, 吕宜生, 张杰, 等. 深度学习在控制领域的研究现状与展望[J]. 自动化学报, 2016, 42(5):643-654.
	DUAN Y J, LÜ Y S, ZHANG J, et al. Deep learning for control:the state of the art and prospects[J]. Acta Automatica Sinica, 2016, 42(5):643-654. (in Chinese)
[12]	CUI Y Q, HE Y, TANG T T, et al. A new target tracking filter based on deep learning[J]. Chinese Journal of Aeronautics, 2022, 35(5):11-24.
[13]	熊伟, 朱洪峰, 崔亚奇. 在线学习的循环自适应机动目标跟踪算法[J]. 航空学报, 2022, 43(5):444-456.
	XIONG W, ZHU H F, CUI Y Q. Recurrent adaptive maneuvering target tracking algorithm based on online learning[J]. Acta Aeronautica et Astronautica Sinica, 2022, 43(5):444-456. (in Chinese)
[14]	GAO C, LIU H W, ZHOU S H, et al. Maneuvering target tracking with recurrent neural networks for radar application[C]// Proceedings of the 2018 International Conference on Radar.Brisbane,Australia:IEEE, 2018:1-5.
[15]	JOUABER S, BONNABEL S, VWLASCO-FORERO S, et al. NNAKF:a neural network adapted Kalman filter for target tracking[C]// Proceedings of the 2021 IEEE International Conference on Acoustics,Speech and Signal Processing.Toronto,Canada:IEEE, 2021:4075-4079.
[16]	FORTI N, MILLEFIORI L M, BRACA P, et al. Model-based deep learning for maneuvering target tracking[C]// Proceedings of the 2023 26th International Conference on Information Fusion. Charleston,SC,US:IEEE, 2023:1-6.
[17]	GAO C, YAN J K, ZHOU S H, et al. Long short-term memory-based deep recurrent neural networks for target tracking[J]. Information Sciences, 2019, 502:279-296.
[18]	LIU J X, WANG Z L, XU M. DeepMTT:a deep learning maneuvering target-tracking algorithm based on bidirectional LSTM network[J]. Information Fusion, 2020, 53:289-304.
[19]	LIU Z G, WANG Z K, YANG Y B, et al. A data-driven maneuvering target tracking method aided with partial models[J]. IEEE Transactions on Vehicular Technology, 2024, 73:414-425.
[20]	REVACH G, SHLEZINGER N, NI X Y, et al. KalmanNet:neural network aided Kalman filtering for partially known dynamics[J]. IEEE Transactions on Signal Processing, 2022, 70:1532-1547.
[21]	REVACH G, NI X Y, SHLEZINGER N, et al. RTSNet:learning to smooth in partially known state-space models[J]. IEEE Transactions on Signal Processing, 2023, 71:4441-4456.
[22]	BUCHNIK I, REVACH G, STEGER D, et al. Latent-KalmanNet:learned Kalman filtering for tracking from high-dimensional signals[J]. IEEE Transactions on Signal Processing, 2023, 72:352-367.
[23]	ZHANG Y S, LI G, ZHANG X P, et al. Transformer-based tracking network for maneuvering targets[C]// Proceedings of the 2023 IEEE International Conference on Acoustics,Speech and Signal Processing.Rhodes Island,Greece:IEEE, 2023:1-5.
[24]	ZHANG Y S, LI G, ZHANG X P, et al. A deep learning model based on transformer structure for radar tracking of maneuvering targets[J]. Information Fusion, 2024, 103:102120.
[25]	XU L F, LIANG Y, DUAN Z S, et al. Route-based dynamics modeling and tracking with application to air traffic surveillance[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 21(1):209-221.
[26]	XU L F, LI X R, LIANG Y, et al. Modeling and state estimation of linear destination-constrained dynamic systems[J]. IEEE Transactions on Signal Processing, 2022, 70:2374-2387.
[27]	HAO X H, LIANG Y, XU L F, et al. Mode separability-based state estimation for uncertain constrained dynamic systems[J]. Automatica, 2020, 115:108905.
[28]	XU L F, LI X R, DUAN Z S, et al. Modeling and state estimation for dynamic systems with linear equality constraints[J]. IEEE Transactions on Signal Processing, 2013, 61(11):2927-2939.
[29]	SIMON D. Kalman filtering with state constraints:a survey of linear and nonlinear algorithms[J]. IET Control Theory & Applications, 2010, 4(8):1303-1318.
[30]	SIMON D, CHIA T L. Kalman filtering with state equality constraints[J]. IEEE Transactions on Aerospace and Electronic Systems, 2002, 38(1):128-136.
[31]	ZHANG C, FU J. Multi-objective dynamic optimization of path-constrained switched systems[J]. Automatica, 2024, 159:111326.
[32]	FU J, LI H. Dynamic optimization of state-dependent switched systems with free switching sequences[J]. Automatica, 2023, 148:110747.
[33]	FU Q, LU K L, SUN C Y. Deep learning aided state estimation for guarded semi-Markov switching systems with soft constraints[J]. IEEE Transactions on Signal Processing, 2023, 71:3100-3116.
[34]	FU J, ZHANG C. Optimal control of path-constrained switched systems with guaranteed feasibility[J]. IEEE Transactions on Automatic Control, 2021, 67(3):1342-1355.