Cooperative localization of the connected vehicles is significant for many advanced intelligent transportation system (ITS) applications. Vehicle-to-vehicle communication using dedicated short-range communication (DSRC) has great potential to enhance global navigation satellite systems (GNSSs) for the capability of cooperative localization. In the integration of DSRC and GNSS, the tolerance against the unknown and time-varying observation conditions is a key factor to fulfill the requirements of several specific ITS applications. Under a GNSS/DSRC integrated architecture for cooperative localization, a novel robust cubature Kalman filter (CKF) is proposed in this paper to improve the performance of the data fusion under uncertain sensor observation environments. In the proposed solution, the structure of the standard CKF is enhanced using the Huber M-estimation technique, in which the original measurement update in the CKF is modified considering the probable anomalies in state estimation. Furthermore, based on the investigation of the adjustment effect from the constraint factor, an adaptive strategy for this parameter is introduced to optimize the performance comprehensively. The proposed method is validated using a specific simulation system. Results of experiment and simulations demonstrate the capability of improving the robustness and adaptive performance over the original filters under the unknown operation conditions.
7 September 2023 in IEEE Transactions on Intelligent Vehicles
Fault detection is crucial to isolate positioning risks for safety-critical applications using Global Navigation Satellite Systems. Conventional Kalman filter-based fault detection methods mainly focus on satellite measurement faults and presume that the test statistics follow the chi-square distribution. These methods ignore the adverse effect of undetected faults occurring previously, and the detection performance would be restrained under a mis-matched distribution assumption. To solve these issues, an enhanced fault detection method is proposed in this paper, which combines the Maximum Correntropy Criterion (MCC) and Local Outlier Factor (LOF). The MCC is introduced to derive a robust extended Kalman filter to deal with the undetected faults. Simultaneously, a specific Kernel Bandwidth (KB) for each measurement is calculated by the innovation and innovation covariance matrix to handle the inherent restriction of a fixed KB. Moreover, the LOF is used to reconstruct the test statistics, and the threshold is calculated by an offline model. Simulations are conducted to evaluate the proposed method under different fault scenarios. The results illustrate that the adaptive robust estimation reduces the negative influence of undetected faults, which makes the filter innovation follow the actual fault amplitudes. The proposed algorithm effectively improves the fault detection rate and positioning accuracy.
On-board train control equipment is an important component of the Train Control System (TCS) of railway trains. In order to guarantee the safe and efficient operation of the railway system, Predictive Maintenance (PdM) is significantly required. The operation data of the on-board equipment allow us to build fault prediction models using a data-driven approach. However, the problem of unbalanced fault samples makes it difficult to achieve the expected modeling performance. In this paper, a Conditional Generative Adversarial Network (CGAN) is adopted to solve the unbalancing problem by generating synthetic samples corresponding to specific fault labels that belong to the minority classes. With this basis, a CGAN-enhanced eXtreme Gradient Boosting (XGBoost) solution is presented for training the fault prediction models. From the pre-processing to the field data, artificial fault samples are generated and integrated into the training sample sets, and the XGBoost models can be derived with multiple decision trees. Both the feature importance sequence list and the knowledge graph are derived to describe the characteristics obtained by the models. Filed data sets from practical operation are utilized to validate the proposed solution. By comparison with conventional machine learning algorithms, it can be found that higher accuracy, precision, recall, and F1 scores, which are up to 99.76%, can be achieved by the proposed solution. By involving the CGAN strategy, the maximum enhancement to the F1 score with the XGBoost approach reaches 6.13%. The advantages of the proposed solution show great potential in implementing equipment health management and intelligent condition-based maintenance.
17 November 2021 in IEEE Transactions on Intelligent Transportation Systems
Advantages of Global Navigation Satellite System (GNSS) in precise and reliable localization can be exploited to enable low-cost and train-centric railway train control systems by reducing track-side facilities like Balises and track circuits. However, constrained observability of satellite signals in specific railway station areas leads to a great challenge to the continuity and availability of GNSS-based train positioning. The pseudolite (PL) technology has a great potential for seamless localization under GNSS-challenged or even signal-denied environments. In this paper, we consider the optimized solution of pseudolite constellation design for seamless train positioning in railway station environments. An integrated train positioning architecture is presented based on a combined GNSS/PL measurement model. Using the trackmap, the proposed solution firstly establishes a feature point set covering all tracks in the station by extracting key Points-of-interest (POIs) from track database. Based on that, a K-means-enhanced generalized center-guided firefly algorithm (KGFA) is proposed to improve the standard firefly algorithm (FA) for deriving an optimized pseudolite constellation solution. The performance indicator for each candidate pseudolite layout scheme is defined according to the scenario-based GNSS/PL constellation configuration. The capability of the KGFA-enabled solution is validated by comparisons with similar FA methods. Through a case study, performance of the optimized pseudolite constellation and its influence to GNSS/PL-based seamless train positioning have been demonstrated over the involved reference pseudolite layout strategies. It is noteworthy that the proposed solution enables the enhanced inherent capability of the local pseudolite network to achieve seamless train positioning over the conventional GNSS-alone train positioning mode.
Technological advances in the Global Navigation Satellite System (GNSS) industry have brought significant advantages in enhancing the cost-efficiency of railway applications. However, GNSS vulnerability to external jamming necessitates enhanced protection ability of the GNSS-based train localization system. This paper proposes a resilient train localization solution under the tightly-coupled integration scheme. This solution maximizes the utilization of multi-source information from train-borne sensors, including INS, odometer, and the trackmap database. Based on the existing localization scheme, it achieves a compatible way to address different jamming-intrusion situations without altering the GNSS receiver structure, addressing both the GNSS failure and degradation caused by jamming. Using an odometer/trackmap-enabled equivalent measurement logic, the continuity of localization can be guaranteed against GNSS failure under strong GNSS jamming. A robust filtering algorithm enabled by a jamming discrimination mechanism is proposed for GNSS/INS integration to mitigate the negative effect from degraded GNSS measurements, reducing the hazards by jamming with an intermediate power level. Based on the field data and a jamming test platform, results under two typical jamming scenarios are evaluated to demonstrate the necessity and superiority of the proposed solution. It also emphasizes the importance of the full-life-cycle resilience of GNSS-based train localization under the railway operation environment.