The rail transit equipment industry currently face numerous product models, largely differential configurations, dispersed R&D resources, and limited module reuse. To support platform-based product development and modular configuration, this study develops a modular product technology platform for rail transit equipment based on an Information Technology Application Innovation (ITAI) Product Lifecycle Management (PLM) system. The platform adopts a layered architecture, including the infrastructure, foundation, business and application layers; incorporates management objects, including the product platform, meta-modules, modules, module libraries, and simplified unified resource libraries; implements a process flow that covers platform definition, module partitioning, library entry of resources, single data source governance, configuration, and evaluation. As an engineering case study, the platform was used to support modular configuration, three-dimensional resource invocation and modularity statistics for typical locomotive products, achieving 68.9% modularity. The results indicate that the developed platform facilitates platform-based product design, module reuse and unified management of R&D resources.
To enhance global path planning efficiency for unmanned dump trucks in open-pit coal mines, this paper proposes a planning method that utilizes historical path data for numerical optimization to generate global paths. To reduce following difficulties caused by sharp turns over short distances, a curvature change rate constraint based on path distance is derived from a time-based vehicle kinematic model. To account for obstacle avoidance requirements, a path constraint is formulated using a multi-circle approximation model for both vehicles and obstacles. The start/end points of matched historical paths and the feasible region tunnel at collision points are adjusted to ensure the feasibility and safety of numerical optimization. These constraints are integrated to formulate an optimization model that minimizes control input energy, deviation from historical paths, and total travel time. Experimental results demonstrate that the proposed method effectively utilizes historical paths to generate global loading, queuing, and unloading paths across the designed eight typical scenarios, and reduces time consumption for path generation by more than 50% on average, while ensuring safety and followability.
Rail transit equipment typically involves complex product customization. The entire research and development (R&D) life cycle is characterized by extensive across-domain collaboration, strong system coupling, long data links, and frequent design changes. Traditional design change management exhibits four major pain points: non-standardized change sources, disconnected process links, uncontrollable implementation processes, and lack of cross-system information synchronization. These issues result in delayed change implementation, difficult traceability and high rework rates. To solve them, this paper proposes an end-to-end full-process monitoring method for design changes based on the 4A architecture. An integrated monitoring system is constructed, covering change requests, design implementation, technology implementation, production implementation, supply chain response, and verification for closure, through the integration and collaboration of four domains: business, application, data and technology. Engineering verification shows that the application of this method reduces the rework rate for design changes from 12.3% to 2.1%, shortens the full-process cycle of changes by approximately 9.9%, and achieves nearly 100% accuracy in change implementation, thereby effectively improving R&D quality, shortening delivery cycles and reducing manufacturing costs.
To address the strong coupling between operation sequencing and machine assignment in the flexible job shop scheduling problem (FJSP), as well as the difficulty of directly applying the traditional crayfish optimization algorithm to discrete encoding and its tendency toward premature convergence, this paper proposes a multi-strategy adaptive improved crayfish optimization algorithm, named ICOA_LQC, with the objective of minimizing the makespan. The proposed algorithm uses a two-layer encoding scheme consisting of operation sequencing and machine selection to represent scheduling solutions, and adopts Logistic-Tent chaotic mapping to improve the distribution of the initial population. A serial double-operator discrete updating mechanism is designed to transform the continuous position migration of the original COA into a structural search process suitable for FJSP. In addition, a heat-resistance threshold and Q-Learning are introduced to adjust the switching between exploration and exploitation online, while critical path search is incorporated in the exploitation stage to optimize bottleneck operations. Experimental results on Brandimarte benchmark instances show that ICOA_LQC obtains optimal or tie-optimal solutions on all 10 benchmark instances, with an average RPD of 0.92%, indicating that the proposed method has good solving capability and application potential for FJSP.
In August 2023, the European Union published the Technical Specification for Interoperability relating to the Control-Command and Signalling Subsystems(TSI CCS 2023/1695). This updated regulation introduces key modules including the future railway mobile communication system (FRMCS) and automatic train operation (ATO), to accommodate emerging technologies such as automated driving and next-generation communication systems, while addressing long-standing bottlenecks, particularly deficient interoperability for cross-border railways. By defining mandatory certification requirements and tightening market admittance restrictions, the regulation is intended to drive the green and digital transformation of the railway sector. Through comparison with the previous version of this TSI, this paper systematically analyzes regulatory changes related to compatibility testing for onboard and trackside subsystems, safety certification rules, and technical barriers, in the context of China's railway digitalization strategies. Notable changes include new requirements for FRMCS and ATO, transitional arrangements, and updated compatibility provisions. Based on this analysis, the paper proposes actionable recommendations for Chinese enterprises seeking EU railway certification and international market opportunities, in aspects including transition-period technologies and compatibility adaptation and with a particular focus on platform-based development, productization of interoperability constituents, and management process improvement.
Lithium-ion batteries are prone to thermal runaway under overcharge conditions, posing a serious safety risk. Although previous studies have conducted modeling and feature capture research on lithium battery overcharge thermal runaway, there remain two core issues due to the limitations of model applicability and experimental conditions: first, existing simplified models cannot accurately represent the subtle characteristic signals at the early stage of overcharge thermal runaway, making it difficult to meet the needs for early risk detection; second, there is a conflict between model reliability and the real-time response requirements of actual BMS, limiting the engineering application of early interventions for thermal runaway. Therefore, based on electrochemical principles and heat transfer theory, this study developed an electro-thermal coupled model for lithium-ion battery overcharge thermal runaway that considers sensitivity, accuracy, and real-time performance. This model simplifies the internal temperature distribution of the battery, uses a one-dimensional electro-thermal coupling approach, focuses on analyzing the roles of Joule heat and side reaction heat in the thermal runaway process, and incorporates convective heat transfer boundary conditions. By comparing experimental data, the model demonstrates high accuracy in predicting battery temperature and voltage changes: the deviation in predicting the onset of rapid temperature rise during thermal runaway is less than 10 seconds, meeting early warning requirements; the voltage prediction accurately captures the 'spike-cliff drop' feature, with a deviation of only 2 seconds at the rapid drop point, effectively simulating the key voltage and temperature characteristics during overcharge. This provides a theoretical basis for the development of overcharge thermal runaway warning and safety protection strategies.
In response to the challenge where train headway limits transport efficiency and safety under the high-density and mixed-mode development of urban rail transit, this paper investigates the calculation of train headway and the evaluation of line capacity under moving block systems. Traditional fixed-block capacity assessment methods are difficult to apply directly to moving block systems due to limitations such as insufficient model adaptability. To address these issues, a dynamic headway calculation method integrating characteristics of both CBTC and TACS systems is proposed. Three types of block section models are constructed for interval tracking, station operations, and turnaround movements, and an automatic block division mechanism is established. Combined with time-window compression technology, a closed-loop calculation process is formed. A simulation platform was developed, and a case study on a section of Beijing Metro Line 7 was conducted. The results show that the TACS system improves turnaround capacity by approximately 19% compared to CBTC, and the combined application of platform tracking and soft-wall collision technology can further reduce tracking headway. This study extends traditional fixed-block capacity assessment methods to moving block systems in urban rail transit, establishing an integrated framework for automatic block division and headway calculation. It provides theoretical support for system selection, signal configuration, and operational strategy formulation in high-density urban rail systems.
Retired battery echelon utilization offers significant economic and environmental benefits, and battery consistency is a key factor affecting performance in echelon applications. Current research methods suffer from the following deficiencies: feature selection fails to achieve a favorable balance between cost and effectiveness, restricting engineering popularization; clustering algorithms have drawbacks in robustness, adaptability and initialization sensitivity, limiting their application in multidimensional battery datasets; verification experiments often neglect comprehensive evaluation metrics and fail to consider long-term consistency in practical applications. To address these issues, this paper proposes an innovative method for improving the consistency of retired battery systems, which mainly includes four key improvements: firstly, coefficient of variation (CV) analysis is applied to differential voltage (DV) curves to identify critical differential voltage segments; a hybrid algorithm integrating nearest neighbor technology and fuzzy C-means clustering (improved fuzzy C-means, IFCM) is proposed for optimal battery grouping, which can increase long-term consistency by at least 12.5% compared with conventional methods; a distance-based outlier detection method is put forward, which improves the long-term system consistency by 13.8% compared with density-based methods; a weighted average evaluation system is established to systematically evaluate battery consistency. Experimental results demonstrate that the proposed method can effectively enhance the reliability and sustainability of retired batteries under echelon utilization scenarios.
The cold start-up process of once-through boilers used for thermal power units poses challenges to traditional strategies to achieve precise and efficient automatic control, due to large time delays, strong nonlinearity, and multivariable coupling. To address these challenges, an intelligent control strategy fusing a deep iterative decomposition Transformer (DID-Transformer) and a Kolmogorov-Arnold network (KAN) is proposed. First, the DID-Transformer is constructed to accurately predict the trends of key boiler state variables for the upcoming minute. Subsequently, the KAN is employed to fit the nonlinear mapping relationship between state variables and control inputs; based on the predicted values, an optimal coal feeding command is deduced to allow for anticipatory boiler regulation. Simulation verification was conducted based on real operational data from a 350 MW unit. Simulation results demonstrate that, compared to the baseline control strategy fusing Transformer and multi-layer perceptron (MLP), the proposed strategy reduces closed-loop tracking errors of the main steam temperature change rate by 17.0% and decreases the cumulative amplitude of coal-feeding command adjustments by 76.4%, effectively enhancing the control quality and automation level of the cold start-up process for once-through boilers.
Conventional kernel-based Fisher discriminant analysis (KFDA) is limited in adaptability of kernel functions for feature extraction from condition monitoring data. To address this, this paper proposes a feature extraction method based on multiple co-mapping kernel Fisher discriminant analysis (MCKFDA), which incorporates kernel function optimization into the conventional framework. Firstly, a data-dependent kernel function is constructed to adapt the kernel structure to various types of condition monitoring data collected; on this basis, a multiple co-mapping kernel structure is devised to accommodate diversity across data samples. Then, the maximun margin criterion (MMC) is employed to achieve the optimal combination weights of the multiple co-mapping kernel, while the Fisher criterion is used to find the optimal data-dependent kernel parameters. The proposed approach was assessed using a lithium-ion battery dataset. Experimental results show that, compared with traditional kernel discriminant analysis, the proposed MCKFDA method improves the mean square error by 15.4% for lithium-ion battery capacity estimation, representing a substantial enhancement in accuracy of performance evaluation.
This paper aims to enhance safety protection and management of locomotives when the train operation monitoring system (LKJ) is deactivated. It presents a systematic review of current operational practices for locomotives in LKJ-deactivated scenarios. Drawing on relevant LKJ regulations including maintenance rules, operation procedures, and technical specifications, and on established EMU requirements for safety protection following ATP/LKJ isolation, this paper proposes a technical solution of automatic protection for locomotives after LKJ deactivation and provides an in-depth feasibility analysis for retrofitting existing locomotives. Test results show that this solution can effectively identify LKJ deactivation and automatically activate speed-limit protection according to set speed limits, thereby ensuring train operation safety while maintaining railway transport efficiency.
In view of challenges faced when using remotely operated vehicles (ROVs) submarine cable laying, such as operational difficulty and low localization accuracy for cable detection, this paper introduces several common submarine cable detection systems carried by ROVs, surveys the development status at home and abroad in this field, as well as the composition and principles of different cable detection systems. Taking the Artemis detection system of SMD Company as a case study, this paper investigate its operating mechanisms for both passive and active modes and discusses in detail the system composition, detection principle, as well as technical features in data processing and real-time measurements. Further analysis of this cable detection system involves deviation, burial depth, position, and current through the upper computer display, along with adjustments for better detection performance in practical engineering applications. Finally, limitations of cable detection systems currently in service in China are analyzed in terms of accuracy, operating requirements, and anti-interference capability, and the future development directions of China's cable detection technology are provided.
Thermal runaway and overheating in lithium batteries are major obstacles to the rapid development of the energy storage industry. Existing battery thermal management systems are still significantly deficient in temperature monitoring and warning mechanisms, making it difficult to accurately perceive internal thermal states of batteries. Therefore, developing high-precision estimation methods is of great importance for ensuring safe and stable system operation. This study focuses on lithium iron phosphate (LiFePO4) energy storage batteries, investigating their electro-thermal behavior and internal temperature estimation. Hybrid pulse power characterization (HPPC) testing was conducted to obtain transient voltage-current responses of batteries, and an offline parameter identification strategy was employed to extract key electrical parameters. Based on these results, a resistor-capacitor (RC) equivalent circuit model was established. The accuracy of the identification model was verified by comparing simulation results with experimental temperature rise data. Heat generation during charging and discharging was then calculated based on the electrical parameters. Using the thermal parameters of lithium batteries, an electro-thermal coupling model was developed using a lumped parameter method. Comparison with 3D computational fluid dynamics (CFD) simulation models validated the accuracy of the proposed model, demonstrating high-precision estimation and relatively high computational efficiency under complex operating conditions.
To meet real-time and high-precision localization requirements in rail transit scenarios, a parallel computing localization method based on LiDAR and inertial measurement units (IMU) is proposed. Adopting a space-for-time design concept, this method utilizes the GPU characteristics: a large number of cores, strong parallel computing performance, and high video memory bandwidth. By constructing a voxel grid network, it enables efficient spatial indexing and nearest neighbor search of point clouds; voxel downsampling, LiDAR point cloud preprocessing, line and surface constraint construction, localization initialization, and relocation modules are transplanted to a graphics processing unit (GPU) for parallel operations. Experimental results show that the proposed method meets real-time localization requirements for automatic train operation from both localization accuracy and real-time performance, with an average decrease in CPU load rate by 25.66%, a 44.72% reduction in single-frame point cloud processing time from 85.87 ms to 47.47 ms, and an average absolute pose error of 0.10 m.
To address shortcomings of existing battery balancing techniques, including short balancing durations, low efficiency, and limited consistency optimization performance, this paper proposes an innovative consistency optimization strategy that integrates full-time balancing with intelligent top-up charging. By establishing a "monitoring-identification-compensation" tiered intervention framework, this strategy enables mild outlier balancing and severe outlier top-up charging for abnormal cells. Specifically, on one hand, a full-time balancing approach is designed to precisely identify target cells for balancing, quantify the duration of balancing, and determine the timing for starting and stopping the balancing process across the entire state-of-charge (SOC) range. On the other hand, a "rough compensation- precise compensation" dual-stage top-up charging solution is designed, where high current is supplied to correct cells with large deviations in the rough compensation stage, and the precise compensation stage involves supply of low current following capacity gap calculations based on the open circuit voltage (OCV)-SOC curve. In experiments over 30 charge-discharge cycles, the proposed strategy increased battery pack capacity by 5.94% and reduced voltage difference by 40%, demonstrating its effectiveness for consistency optimization in large-scale energy storage systems.
Deploying multiple train control systems on a single train typically requires separate dedicated driver-machine interface (DMI) devices, which increases both spatial occupation in the driver's cab and operational complexity, thereby compromising driving safety. To address this, this paper proposes a unified display solution for human-machine interaction that supports multi-mode train control systems using a single DMI. The introduction of a software switching function allows the single DMI to be compatible with different control modes (e.g., LKJ and ATP); separate communication interfaces provide data exchange between the DMI and various train control units. This solution simplifies hardware configuration, reduces system complexity, and enhances system flexibility without compromising system reliability. Laboratory verification and field trials under normalized operating conditions demonstrate the feasibility and stability of the proposed solution in practical applications and indicate its potential for widespread adoption in rail transit systems.
A semi-supervised anomaly detection and classification framework are proposed to address limited accuracy in anomaly detection of wind turbines caused by highly heterogeneous audio data, scarce anomalous samples, and the difficulty in identifying faults of unknown categories. The core innovation is a two-stage strategy: (i) an improved GANomaly model for efficient anomaly detection, and (ii) a dynamic feature memory bank that performs fine-grained anomaly classification via feature-similarity matching to facilitate identification of categories beyond the training scope. Experiments on a real-world wind-turbine audio dataset show that the proposed approach achieves an F1-score of 99.39% for anomaly detection and an overall accuracy of 96.75% for multi-class anomaly classification, offering an effective solution for class imbalance and challenges in identifying anomalies in unknown categories in industrial scenarios and providing a reliable tool for intelligent wind-turbine operation and maintenance.
For the purpose of carbon emission reduction as well as efficient and economic operation of park integrated energy systems (PIES), this paper proposes a low-carbon, robust, and economic dispatch strategy that accounts for flexible equipment operation characteristics and is based on a stepped carbon trading mechanism. First, a flexible equipment operation model that incorporates start and stop conditions was established within the PIES energy flow framework. Then, a scenario analysis approach was used to deal with uncertainty in source-side renewable energy output, and a multi-type electric-thermal flexible load model was established that reflects the characteristics of dispatchable resources on the load side, to improve operational robustness. Finally, based on the stepped carbon trading mechanism, an optimal economic dispatch model was constructed, which incorporates multiple and complex operational constraints across the source, load, and energy conversion/storage sides. Experimental results show that, when accounting for costs including energy purchase, equipment operation and maintenance, and carbon trading, the proposed strategy enables simultaneous optimizations of carbon emission reduction and economic performance improvement for PIESs.
To enhance collision prevention capability in locomotive operating environments, this paper presents a multi-source fusion-based obstacle detection system for locomotives. Using high-precision track maps as prior information, this system combines real-time kinematic (RTK) high-precision positioning data with track authorization data from the signaling system to calculate forward track-centerline coordinates in the vehicle coordinate system during operation. A kinematic clearance is defined based on these track-centerline coordinates that are also projected onto the vision image to obtain 2D pixel coordinates of the track centerline. This kinematic clearance is used to segment LiDAR point clouds for obstacle detection, generating LiDAR-detected obstacles within the clearance. The 2D pixel coordinates of the track centerline are also used to select the target track from visually detected paths, allowing for subsequent collision-risk assessment of visually detected objects. Finally, fused LiDAR-vision detection results are output. Field applications demonstrate the system's capability of providing long- to medium-range obstacle warnings using vision-only detection and medium- to short-range collision prevention through fusion-based detection.
Train traction motor systems exhibit significant process dynamics and nonlinear characteristics; accurately capturing their dynamic evolution patterns from system data helps improve the accuracy of fault detection. To this end, a fault detection method for train traction motors based on a state-space cascaded prediction network (SSCPN) is proposed in this paper, which enhances fault detection performance by capturing sufficient dynamic information. First, a cascaded structural mapping relationship is established among historical inputs-outputs, future inputs, and future outputs of dynamic data;learnable matrices and a neural network are then utilized to jointly fit this mapping relationship. Based on this, a matrix constraint for the state observer system is established to improve the prediction accuracy of normal-state outputs, and the constraint is incorporated into the network's loss function to achieve adaptive satisfaction. Furthermore, leveraging characteristics in abnormal prediction residuals under fault conditions, fault evaluation is performed by combining the Hotelling's statistic with a kernel density estimation threshold. Simulation experiments involving the injection of traction motor faults demonstrate that the proposed method effectively identifies faults by leveraging differences in model prediction accuracy between normal and faulty states. Compared to other methods, the proposed method reduces the false positive and false negative rates by approximately 14.90% and 68.46% on average, respectively, validating its effectiveness and superiority.
