车辆目标实时跟踪是智慧交通的基础，毫米波雷达已广泛应用于交通监测。但城市交通场景复杂、车辆目标分布无序，传统关联算法进行目标关联时存在航迹匹配错误的问题，车辆非线性运动导致航迹不够准确。因此，开展多目标关联算法和滤波跟踪算法研究对提高车辆的跟踪性能具有十分重要的意义。

针对毫米波雷达交通监测中关联算法对多目标关联准确率低、计算量大的问题，提出一种次最优改进的联合概率关联算法。该算法将确认矩阵拆分成若干互联矩阵，根据航迹质量设置关联的优先级进行三次计算，减少互联事件的计算量，提高运算效率；质量差的点迹用于航迹补点计算，缓解量测数据缺失导致的航迹中断问题，提高关联准确性。相比传统联合概率数据关联算法，在两目标关联中，改进算法的运算耗时为联合概率算法的48.2%，同时改进算法的位置与速度的平均误差仅增加4%。该算法能较好地解决高杂波环境下，多目标关联波门重叠导致的量测点关联正确率低、计算量大等问题，提高目标关联的准确率和算法效率。

针对毫米波雷达交通监测中滤波算法对运动变化的目标跟踪性能差的问题，提出一种自适应强跟踪改进的扩展卡尔曼滤波算法。该算法对测量值进行估计，根据当前的测量新息强制校正一步预测的误差协方差，计算卡尔曼增益，用增益与量测值计算滤波估计值。改进算法基于强跟踪理论，引入次优渐消因子调节参数，调整后验协方差，从而满足算法的正交性，保持对非线性运动的跟踪精度；为减少参数扰动和状态突变对估计精度的影响，提高抗干扰能力，用多重次优渐消因子计算弱化因子，用弱化因子调整后验协方差，消除强跟踪滤波的发散问题，增加算法鲁棒性。相比传统强跟踪扩展卡尔曼滤波算法，改进算法的位置精度提升26.8%，速度精度提升38.4%。该算法能较好地保留传统算法运算收敛后丧失对目标运动状态突变的高跟踪能力，提高滤波跟踪的位置与速度精度。

使用CAL60S244射频芯片和Xilinx ZYNQ-7020信号处理芯片搭建毫米波交通雷达平台，在西安某路进行实际测试，对单向三车道与双向四车道两个场景的数百辆车进行跟踪，实现车道划分与车流量统计功能，流量统计正确率达到96%。结果表明，改进的多目标跟踪算法，在交通检测中可获取车辆的运动信息，为毫米波雷达在交通应用中提供一定的理论参考。

Real-time tracking of vehicle targets is the foundation of intelligent transportation, and millimeter-wave radar has been widely used in traffic monitoring. However, urban traffic scenes are complex and vehicle targets are distributed in disorder, which leads to errors in matching between points and tracks when using traditional correlation algorithms for vehicle target association. The nonlinear motion of vehicles also leads to inaccurate tracking of tracks. Therefore, conducting research on multi-target correlation algorithms and filtering tracking algorithms is of great significance for improving the tracking performance of vehicles.

To focus on the problem of low accuracy and large computational load of multi-target association algorithms in millimeter-wave radar traffic monitoring, a sub-optimal improved joint probability association algorithm is proposed. The algorithm divides the confirmation matrix into several interconnected matrices, and sets the priority of association based on the quality of the track for three calculations to reduce the computational load of interconnected events and improve the efficiency of operation; meanwhile the poor quality plots are used for track supplement calculations to alleviate the problem of track interruption caused by missing measurement data and improve the accuracy of association. In the association of two targets, to compare with traditional joint probability data association algorithms, the improved algorithm’s computation time is 48.2% of the joint probability algorithm, while the average error of the improved algorithm’s position and speed merely increases by 4%. This improved algorithm can effectively solve the problems of low accuracy and large computational load caused by overlapping measurement point association windows in high clutter environments, and improves the accuracy and efficiency of target association.

In view of the poor tracking performance of filtering algorithms for moving targets in millimeter-wave radar traffic monitoring, an adaptive strong tracking improved extended Kalman filter algorithm could be came up with. The algorithm estimates the measurement values, forces the correction of the error covariance of one-step prediction based on the current measurement innovation, calculates the Kalman gain that is used for calculation of the measurement values of the filtered estimation. The improved algorithm based on the strong tracking theory, introduces a suboptimal fading factor to adjust the parameters, adjusts the posterior covariance, and thereby satisfies the orthogonality of the algorithm and maintains the tracking accuracy for nonlinear motion. In order to reduce the impact of parameter disturbances and state mutations on estimation accuracy and improve anti-interference ability, a multiple-suboptimal fading factors are used to calculate the weakening factor, and the posterior covariance is adjusted using the weakening factor to eliminate the divergence problem of strong tracking filtering and increase the robustness of the algorithm. Compared with traditional strong tracking extended Kalman filter algorithms, the improved algorithm has a 26.8% increase in position accuracy and a 38.4% increase in speed accuracy. And it can better solve the problem of losing high tracking ability for target motion state mutations after convergence of traditional algorithms, and improve the position and speed accuracy of filtering tracking.

The millimeter-wave traffic radar platform based on the CAL60S244 RF chip and the Xilinx ZYNQ-7020 signal processing chip was tested on a road in Xi’an, tracking hundreds of vehicles in two scenarios: one-way three-lane and two-way four-lane. Statistical analysis of traffic flow was conducted on these two types of vehicle conditions, and the accuracy rate of the results reached 96%. The results showed that the improved multi-target tracking algorithm can obtain accurate vehicle information in millimeter-wave radar traffic detection, providing a theoretical reference for intelligent transportation informatization.

[1] 关积珍. 智能交通系统发展演进及其代际特征[J]. 人工智能, 2022, (04): 40-49.

[2] 郑健壮, 张嘉旎, 邵勇. 智能交通系统发展历程与我国数字化转型[J]. 公路交通技术, 2022, 38(04): 163-171.

[3] Jafari-Sadeghi V, Garcia-Perez A, Candelo E, et al. Exploring the impact of digital transformation on technology entrepreneurship and technological market expansion: The role of technology readiness, exploration and exploitation[J]. Journal of Business Research, 2021, 124: 100-111.

[4] 高畅, 金珊珊. 传感器在交通信息采集中的应用现状研究[J]. 传感器世界, 2020, 26(05): 18-23.

[5] Zhu B, Sun Y, Zhao J, et al. Millimeter-wave radar in-the-loop testing for intelligent vehicles[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23(8): 11126-11136.

[6] Zhao C, Ding D, Du Z, et al. Analysis of perception accuracy of roadside millimeter-wave radar for traffic risk assessment and early warning systems[J]. International journal of environmental research and public health, 2023, 20(1): 879.

[7] Zhang T, Jin P J. Roadside lidar vehicle detection and tracking using range and intensity background subtraction[J]. Journal of advanced transportation, 2022, 2022.

[8] Wang G, Wu J, Xu T, et al. 3D vehicle detection with RSU LiDAR for autonomous mine[J]. IEEE Transactions on Vehicular Technology, 2021, 70(1): 344-355.

[9] Deshmukh P, Satyanarayana G S R, Majhi S, et al. Swin transformer based vehicle detection in undisciplined traffic environment[J]. Expert Systems with Applications, 2023, 213: 118992.

[10] Othmani M. A vehicle detection and tracking method for traffic video based on faster R-CNN[J]. Multimedia Tools and Applications, 2022, 81(20): 28347-28365.

[11] Zhang L, Mao D, Niu J, et al. Continuous tracking of targets for stereoscopic HFSWR based on IMM filtering combined with ELM[J]. Remote Sensing, 2020, 12(2): 272.

[12] Zhao C, Ding D, Du Z, et al. Analysis of perception accuracy of roadside millimeter-wave radar for traffic risk assessment and early warning systems[J]. International journal of environmental research and public health, 2023, 20(1): 879.

[13] Wang K, Xiong H, Zhang J, et al. Sensemag: Enabling low-cost traffic monitoring using noninvasive magnetic sensing[J]. IEEE Internet of Things Journal, 2021, 8(22): 16666-16679.

[14] 赵毅寰, 天光. 雷神公司的相控阵技术——波束合成和信号/数据处理[J]. 航空兵器, 2016, (06): 39.

[15] 姜兴, 刘耀, 李晓峰等. 一种毫米波宽带宽波束双极化微带天线[J]. 电讯技术, 2023, 63(09): 1433-1438.

[16] 田丰, 刘万, 符渭波等. 重叠阵元时分复用MIMO雷达多目标参数估计方法研究[J]. 电子与信息学报, 2024, 46(01): 229-239.

[17] Anisia P, Diana H, Marianthi-Anna K, et al. Assessing neighborhood-scale traffic from crowd-sensed traffic data: Findings from an environmental justice community in New York City[J]. Environmental Science and Policy, 2022, 133155-163.

[18] Marco C, João A, C. P B, et al. Performance Assessment of Collective Perception Service Supported by the Roadside Infrastructure[J]. Electronics, 2022, 11(3): 347-347.

[19] 肖博, 霍凯, 刘永祥. 雷达通信一体化研究现状与发展趋势[J]. 电子与信息学报, 2019, 41(03): 739-750.

[20] 章永来, 周耀鉴. 聚类算法综述[J]. 计算机应用, 2019, 39(07): 1869-1882.

[21] Jothi R, Mohanty S K, Ojha A. DK-means: a deterministic k-means clustering algorithm for gene expression analysis[J]. Pattern Analysis and Applications, 2019, 22: 649-667.

[22] Campbell T, Kulis B, How J. Dynamic clustering algorithms via small-variance analysis of Markov chain mixture models[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 41(6): 1338-1352.

[23] Nguyen H H. Privacy-preserving mechanisms for k-modes clustering[J]. Computers & Security, 2018, 78: 60-75.

[24] Schubert E, Sander J, Ester M, et al. DBSCAN revisited, revisited: why and how you should (still) use DBSCAN[J]. ACM Transactions on Database Systems (TODS), 2017, 42(3): 1-21.

[25] Yuan G, Sun P, Zhao J, et al. A review of moving object trajectory clustering algorithms[J]. Artificial Intelligence Review, 2017, 47: 123-144.

[26] 林凤泰, 严蘋蘋, 张慧, 徐刚. 基于最近迭代点的毫米波雷达点云数据处理方法[J]. 信号处理, 2023, 39(02): 288-297.

[27] 刘鲁涛, 王璐璐, 李品, 陈涛. DSets-DBSCAN无参数聚类的雷达信号分选算法[J]. 国防科技大学学报, 2022, 44(04): 158-163.

[28] Liu Q, Zhang R, Hu R, et al. An improved path-based clustering algorithm[J]. Knowledge-Based Systems, 2019, 163: 69-81.

[29] Hahsler M, Piekenbrock M, Doran D. dbscan: Fast density-based clustering with R[J]. Journal of Statistical Software, 2019, 91: 1-30.

[30] Chen Y, Zhou L, Bouguila N, et al. BLOCK-DBSCAN: Fast clustering for large scale data[J]. Pattern Recognition, 2021, 109: 107624.

[31] Wang Z, Zhang Y. Estimation of ship berthing parameters based on Multi-LiDAR and MMW radar data fusion[J]. Ocean Engineering, 2022, 266: 113155.

[32] 戴永寿, 马鹏, 孙伟峰等. 基于JVC的紧凑型地波雷达海上目标点迹-航迹最优关联方法[J]. 电子与信息学报, 2021, 43(10): 2832-2839.

[33] Cheng L, Zhang H, Wei D, et al. An indoor tracking algorithm based on particle filter and nearest neighbor data fusion for wireless sensor networks[J]. Remote Sensing, 2022, 14(22): 5791.

[34] Abu Alfeilat H A, Hassanat A B A, Lasassmeh O, et al. Effects of distance measure choice on k-nearest neighbor classifier performance: a review[J]. Big data, 2019, 7(4): 221-248.

[35] Lerro D, Bar-Shalom Y. Interacting multiple model tracking with target amplitude feature[J]. IEEE Transactions on Aerospace and Electronic Systems, 1993, 29(2): 494-509.

[36] Sun S, Lyu H, Dong C. AIS aided marine radar target tracking in a detection occluded environment[J]. Ocean Engineering, 2023, 288: 116133.

[37] Reid D. An algorithm for tracking multiple targets[J]. IEEE transactions on Automatic Control, 1979, 24(6): 843-854.

[38] Lyu J, Cao Z, Song E. A data association algorithm for the robust confidence ellipsoid filter[J]. Signal Processing, 2023, 213: 109201.

[39] Fowdur J S, Baum M, Heymann F, et al. An Overview of the PAKF-JPDA Approach for Elliptical Multiple Extended Target Tracking Using High-Resolution Marine Radar Data[J]. Remote Sensing, 2023, 15(10): 2503.

[40] Rakai L, Song H, Sun S J, et al. Data association in multiple object tracking: A survey of recent techniques[J]. Expert Systems with Applications, 2022, 192: 116300.

[41] He S, Shin H S, Tsourdos A. Multi-sensor multi-target tracking using domain knowledge and clustering[J]. IEEE Sensors Journal, 2018, 18(19): 8074-8084.

[42] 韩蕾蕾. 一种基于D-CJPDA的多目标数据互联算法[J]. 电光与控制, 2020, 27(06): 6-10.

[43] Sun W, Li X, Ji Y, et al. Plot quality aided plot-to-track association in dense clutter for compact high-frequency surface wave radar[J]. Remote Sensing, 2022, 15(1): 138.

[44] 侯淋, 杨顺华. 光电雷达多目标跟踪技术发展分析[J]. 电光与控制, 2021, 28(08): 65-70.

[45] Kalman R E. A new approach to linear filtering and prediction problems[J]. 1960.

[46] Aidala V J. Kalman filter behavior in bearings-only tracking applications[J]. IEEE Transactions on Aerospace and Electronic Systems, 1979 (1): 29-39.

[47] Kalman R E, Bucy R S. New results in linear filtering and prediction theory[J]. 1961.

[48] Na W, Yoo C. Real-Time Parameter Estimation of a Dual-Pol Radar Rain Rate Estimator Using the Extended Kalman Filter[J]. Remote Sensing, 2021, 13(12): 2365.

[49] Tang M, Chen Z, Yin F. SLAM with improved schmidt orthogonal unscented Kalman filter[J]. International Journal of Control, Automation and Systems, 2022, 20(4): 1327-1335.

[50] Ge Q, Hu X, Li Y, et al. A novel adaptive Kalman filter based on credibility measure[J]. IEEE/CAA Journal of Automatica Sinica, 2023, 10(1): 103-120.

[51] Hongwei Y, Xinming S. Design of strong tracking Kalman filter under sudden change of vehicle[C]//2021 11th International Conference on Information Technology in Medicine and Education (ITME). IEEE, 2021: 27-31.

[52] Yin Z, Li G, Sun X, et al. A speed estimation method for induction motors based on strong tracking extended Kalman filter[C]//2016 IEEE 8th International Power Electronics and Motion Control Conference (IPEMC-ECCE Asia). IEEE, 2016: 798-802.

[53] 陈佳磊, 覃记荣, 徐小红等. 基于强跟踪的卡尔曼滤波算法道路坡度估计分析[J]. 专用汽车, 2023, (11): 19-22.

[54] Zerdali E, Yildiz R. A study on improving the state estimation of induction motor[J]. Electrical Engineering, 2023, 105(4): 2471-2483.

[55] Ge Q, Shao T, Chen S, et al. Carrier tracking estimation analysis by using the extended strong tracking filtering[J]. IEEE Transactions on Industrial Electronics, 2017, 64(2): 1415-1424.

[56] Zhang A, Shuida B A O, Fei G A O, et al. A novel strong tracking cubature Kalman filter and its application in maneuvering target tracking[J]. Chinese Journal of Aeronautics, 2019, 32(11): 2489-2502.

[57] Ge Z, Jia G, Zhi Y, et al. Strong tracking extended particle filter for manoeuvring target tracking[J]. IET Radar, Sonar & Navigation, 2020, 14(11): 1708-1716.

[58] Takyi-Aninakwa P, Wang S, Zhang H, et al. An ASTSEKF optimizer with nonlinear condition adaptability for accurate SOC estimation of lithium-ion batteries[J]. Journal of Energy Storage, 2023, 70: 108098.

[59] Su H, Zhao D, Heidari A A, et al. RIME: A physics-based optimization[J]. Neurocomputing, 2023, 532: 183-214.

[60] 何友, 修建娟, 关欣. 雷达数据处理及应用(第3版)[M]. 北京: 电子工业出版社,2013.

[61] 黄小平, 王岩. 卡尔曼滤波原理与应用—MATLAB仿真[M]. 北京: 电子工业出版社, 2015.

[62] 赵琳. 非线性系统滤波理论[M]. 北京: 国防工业出版社, 2012.