International Journal of Vehicle Systems Modelling and Testing (18 papers in press)

Regular Issues

• Lightweight design of automotive parts based on the FPTO method  
  by Dengfeng Huang, Xuwei Hu, Xiaolei Yan 
  Abstract: Topology optimisation is a method to maximise or minimise an objective function by optimising material distribution under design constraints. In the theoretical research of topology optimisation, most algorithms are developed on regular geometric models in software such as MATLAB. However, applying these findings directly to complex structures with irregular geometries in engineering is challenging. Optimisation of such structures relies on commercial software using density-based methods, hindering open algorithm research. The floating projection topology optimisation (FPTO) is a stable, efficient method producing good results. This study introduces the FPTO principles and investigates its integration on the MATLAB-ABAQUS platform, including conducting analysis in ABAQUS, performing optimisation solution and result visualisation in MATLAB, and facilitating data exchange. The boundaries of the topology structure are smoothed to better meet actual engineering requirements. This research explores the application of FPTO to automotive components, achieving lightweight designs for wheel hubs and control arms, offering an effective engineering solution.
  Keywords: topology optimisation; floating projection; MATLAB-ABAQUS platform; complex structures.
  DOI: 10.1504/IJVSMT.2025.10073303
   
  
• Effects of endwall slot jet angles on the performance of high-load compressor cascade  
  by Junfu Yuan, Chen Li 
  Abstract: The three-dimensional corner separation represents a major obstacle to enhance performance of a high-load compressor cascade. To tackle this challenge, we conducted a computational fluid dynamics (CFD) numerical simulation to investigate impact of the endwall slit jet parameters on performance of the compressor cascade. Through numerical analysis, we found that the endwall slot jets substantially suppressed the corner separation and reduced flow loss. At a jet flow rate comprising a mere 0.435% of the mainstream flow rate, a 30 degree combined jet angle can cause a drop of 27.63% in losses and an increase of 9.2% in the static pressure coefficient. The primary mechanism behind the effectiveness of endwall slot jets lies in their ability to reduce flow losses and improve cascade performance. This is achieved by suppressing secondary flow in the endwall region, facilitating the mixing of low-energy fluid with the mainstream, proficiently regulating the development of passage vortices and inducing vortices.
  Keywords: high load diffuser cascades; end wall slit jet; jet flow rate; combined jet angle; flow control; angular separation; cascade performance.
  DOI: 10.1504/IJVSMT.2025.10075529
   
  
• Data-driven material constitutive modelling: a framework for method selection and performance evaluation  
  by Hengli Yu, Yantao Wang, Yingjing Wang, Tong Pang, Tangying Liu 
  Abstract: Constitutive models are crucial in engineering design and simulations. Traditional models require tedious parameter calibration and have limited generalization capabilities, while data-driven models offer adaptive learning advantages. However, selecting optimal models remains challenging. This study evaluates eight data-driven methods in material constitutive modeling across sparse and dense parameter spaces. By analyzing fitting accuracy, interpolation capability, and extrapolation performance, we found that deep neural networks provide the most stable generalization in sparse parameter spaces, while kriging achieves near-perfect performance in dense parameter spaces. Based on these findings, we propose a systematic model selection framework that considers data sampling density and prediction task types, providing a theoretical foundation for model selection across various material constitutive models.
  Keywords: constitutive modeling; data-driven methods; deep neural network; kriging.
  DOI: 10.1504/IJVSMT.2025.10074803
   
  
• The integration of neural networks in visual effects design for new energy vehicles  
  by Lijuan Sha, Jingyu Li, Chunxu Zhang, Xiangjuan Liu, Shiqing Lu, Bincheng Zuo, Shuxin Chen 
  Abstract: As the intelligent automotive industry thrives, artificial intelligence has deeply integrated into automotive design, driving the advancement of new energy vehicle (NEV) design. The focus lies in catering to users design preferences to provide utmost interior comfort. This study utilises representative automotive exterior samples to train a backpropagation neural network model and to establish correlations between design element encodings and sensory evaluation metrics. The effectiveness is tested via samples. Additionally, users sensory imagery words are collected for factor analysis. Subsequently, the exterior samples are refined; sensory word groups are identified; and the model is retrained through reassembly. This integration of AI and user sensory imagery analysis provides systematic methodologies for designers and aids in project design. Ultimately, based on factors such as colour, material and surface treatment, an ideal exterior sample for NEVs is predicted, offering a fresh perspective and approach to NEV design.
  Keywords: Kansei engineering; BP neural network; machine learning; style transfer; NEV; new energy vehicle.
  DOI: 10.1504/IJVSMT.2025.10074877
   
  
• TriXAI twin: a threefold explainability approach for battery health monitoring in electric vehicles  
  by P.G. Parvati, Mehbooba Shareef 
  Abstract: When electric vehicles are redefining modern transportation, the ability to monitor and predict battery health in real-time becomes critically important for ensuring safety, reliability and energy efficiency.This paper introduces deep learning framework based on a Long Short-Term Memory(LSTM) for accurate prediction of State of Charge(SoC) and State of Health(SoH) of lithium-ion batteries using NASA battery dataset.However the opacity of deep learning(DL) models often limits their adoption in safety critical application like battery management.To overcome the interpretability challenges of DL models, a hybrid explainability framework combining SHAP(SHapley Additive exPlanations), LIME(Local Interpretable Model-agnostic Explanations), and Layer-Wise Relevance Propagation(LRP) is introduced, which enables both global and local feature attribution.This unified framework not only delivers high predictive accuracy but also unveils temporal and contextual significance of features driving the model's decisions, making it suitable for reliable decision making in EV enabled battery systems.
  Keywords: BMS; battery management system; LSTM; long short-term memory; XAI; explainable artificial intelligence; LIME; local interpretable model-agnostic explanations; SHAP; Shapley additive explanations; LRP; layer-wise relevance propagation; DT; SoC; state of charge; SoH; state of health.
  DOI: 10.1504/IJVSMT.2025.10075067
   
  
• Adaptive fuzzy sliding mode controller to improve ride comfort based on a quarter car model  
  by Jiliang Jia, Qinghai Zhao, Kaiyu Ma, James Yang 
  Abstract: Based on the simulation of a four-degree-of-freedom human body model and a three-degree-of-freedom automobile model, this paper proposes an adaptive fuzzy sliding mode controller (AFSMC) for the seat suspension-body system with skyhook damping control as the reference. An adaptive algorithm is introduced to estimate system model errors and disturbances, improving control accuracy. The fuzzy algorithm optimises the convergence speed parameter of sliding mode control to achieve fast convergence and reduce jitter, while the hyperbolic tangent function replaces the sign function to ensure switching continuity. Simulation analysis on random and bumpy roads shows that AFSMC outperforms PID and passive control in vibration damping, significantly attenuating road vibrations.
  Keywords: quarter car model; sky-hook damping control; adaptive fuzzy sliding mode controller; hyperbolic tangent function; ride comfort.
  DOI: 10.1504/IJVSMT.2025.10075281
   
  
• Constructing question and answer system of urban rail transit emergencies  
  by Jiashuai Li, Bosong Fan, Xuejin Wan, Dong Liu, Dan Zhao 
  Abstract: Urban rail transit systems face challenges such as high data costs, complex knowledge construction, and significant differences across domain-specific datasets in constructing question and answer (Q&A) systems. This paper presents the urban rail transit operation emergency (URTOE)-large language model (LLM). The Q&A system construction method is based on LLM and domain-specific knowledge. URTOE-LLM introduces a domain knowledge enhancement mechanism and optimizes model parameters to meet the specific needs of urban rail transit. The model leverages an image-text multimodal expert and uses a phased pre-training approach to efficiently learn and retrieve of information of the operational emergency. The model's performance is evaluated using an urban rail transit operational emergency log dataset by comparing it with traditional models. The results demonstrate that URTOE-LLM outperforms other models, achieving precision of 0.87 in key phrase and relation recognition tasks.
  Keywords: urban rail transit; operational emergencies; large language modelling; domain question; answer system.
  DOI: 10.1504/IJVSMT.2025.10075692
   
  
• Stall prediction and identification for civil aircraft based on a long short-term memory model  
  by Long Xi 
  Abstract: Stall is a major contributor to aircraft safety incidents. Most civil aircraft lack stall prediction, and existing warnings afford little reaction time, limiting pilots ability to avert accidents. This paper proposes a deep learningbased stall identification approach. We first analyse flight-performance parameters for a representative civil aircraft and define identification methods of incipient stall. We then conduct cruise-phase stall simulations across 28 operating points for the same type and build a long short-term memory (LSTM) model to forecast the resulting time-series signals. Finally, we train and evaluate the model under varied train/test strategies and compare alternatives. Results show the proposed LSTM reliably anticipates stall and supports timely recovery guidance. The framework demonstrates that data-driven sequence modelling can extend crew reaction time and improve safety, offering a practical path toward onboard stall prediction systems for civil aviation. Performance remains robust under realistic noise and sensor latencies.
  Keywords: aircraft stall; LSTM; long short-term memory; state prediction; flight safety.
  DOI: 10.1504/IJVSMT.2025.10075693
   
  
• Data augmentation to enhance the accuracy of deep neural networks in subjective vehicle dynamics evaluation  
  by Cádmo Rodrigues, Jánes Landre Jr 
  Abstract: The accuracy of deep neural networks highly depends on the quality and representativeness of the data input. In the context of vehicle adaptation based on driver preferences, inconsistencies in data collection and noise in realworld driving conditions often hinder virtual model performance. In this context, this study proposes a novel data processing algorithm designed to enhance training data through augmentation and normalisation techniques. To do so, this algorithm relies on statistical transformations applied to the original telemetry signals. After perturbing the mean and variance of selected parameters within the observed experimental limits to generate synthetic samples, the new data retained the same statistical nature as the original dataset, and the plausibility of the augmented dataset was further validated using the KolmogorovSmirnov (KS) test. By refining the dataset and implementing an optimised training loop, a significant reduction in prediction errors and improved model generalisation is achieved. Since the developed system demonstrated a gain in accuracy from less than 25% to about 90% during vehicle evaluation in a driver-in-the-loop dynamic simulator, this work introduces two main innovations: The first involves attenuating the subjectivity inherent in individual driving style preferences, a critical challenge currently faced by the automotive industry; the second refers to the methodology for data collection and processing, validated through statistical tests, ensuring the reliability of the generated results. The experimental results demonstrated that the proposed approach enhances neural network accuracy, leading to more reliable adaptations of vehicle behaviour according to individual driver preferences. These findings highlight the importance of data preprocessing in machine learning applications and open new possibilities for personalised driving experiences.
  Keywords: driver-in-the-loop simulation; vehicle dynamics; subjectiveness in simulation; DiM 150; driver preferences; vehicle data augmentation; machine learning; deep learning.
  DOI: 10.1504/IJVSMT.2025.10076148
   
  
• Multi-closed-loop compound control strategy for braking gap adjustment in electromechanical brake systems  
  by Qiping Chen, Yuanhao Cai, Wuhao Xu, Qiang Shu, Wangliang Fu 
  Abstract: To address the issues of the increased volume and cost associated with traditional electromechanical brake systems (EMB) due to additional mechanical or sensing mechanisms for gap control, this paper proposes a braking gap adjustment strategy based on a multi-closed-loop compound control framework. First, a multi closed-loop EMB motor control system is firstly developed by employing a genetic algorithm-optimized proportional integral derivative (GA-PID) controller to regulate the clamping force. Subsequently, a braking gap adjustment method is then introduced to identify critical points for different braking stages and apply targeted control objectives accordingly. The EMB system and control strategy are modeled and validated through co-simulation using CarSim/Simulink. Simulation results demonstrate that the proposed GA-PID controller ensures rapid and precise clamping force regulation. Furthermore, the gap adjustment strategy effectively reduces response time, maintains optimal brake clearance, and mitigates adverse conditions caused by pad wear, thereby enhancing EMB safety and reliability.
  Keywords: electro-mechanical braking system; critical point; multi-closed loop; genetic PID; gap adjustment.
  DOI: 10.1504/IJVSMT.2025.10076149
   
  
• Dynamic modelling and state estimation of hub motor-driven amphibious vehicles for water-to-land transition  
  by Bin Huang, Wenbin Yu, Lianbing Suo, Zineng Yuan 
  Abstract: This study investigates the modelling and state estimation of a hub-motor-driven amphibious vehicle under water-to-land transition conditions. The focus is on modelling during land-to-water transitions and the design of an effective estimation framework. Accurate estimation must consider complex coupled effects such as buoyancy, fluid resistance, low-adhesion surfaces, and dynamic slopes. To address this, a Fading Memory Unscented Kalman Filter (FM-UKF) is proposed, integrating wheel speed and acceleration for vehicle velocity estimation. A longitudinal slope estimation method is also developed by fusing acceleration sensor data with a Forgetting Factor Recursive Least Squares (FF-RLS) algorithm. A multibody dynamics model is established, enabling real-time estimation of longitudinal speed, terrain slope, and adhesion coefficient through the proposed fusion framework. Simulation results demonstrate the method's effectiveness under transition conditions. Additionally, a real-vehicle experiment on a multi-gradient ramp confirms the accuracy and practicality of the proposed approach in real amphibious driving scenarios.
  Keywords: amphibious vehicle; state estimation; water-to-land composite conditions; multibody dynamics model.
  DOI: 10.1504/IJVSMT.2025.10076258
   
  
• Reinforcement learning-based dynamic pricing research on ride-hailing platforms  
  by Yue Bai, Wei Zhang 
  Abstract: Due to the significant spatiotemporal dynamics and the complex supply-demand relationship in ride-hailing operations, this study proposes a multi-module dynamic pricing model for ride-hailing (MDPM-RH) platforms, which is capable of real-time sensing of regional supply-demand fluctuations and constructing differentiated spatiotemporal price coefficients to enable dynamic pricing. The model considers the vacant vehicle dispatching and driver-passenger matching problems, and corresponding sub models were constructed separately. The dynamic pricing problem is formulated as a Markov decision process (MDP), and solved using the soft actor-critic (SAC) algorithm to maximise long-term profits. The experiments based on real ride-hailing order data in Haikou City show that the dynamic pricing method proposed in this study can effectively improve the platform profit. Compared with static pricing, the dynamic pricing strategy achieved a weekly profit improvement of
  Keywords: dynamic pricing; ride-hailing; reinforcement learning; soft actor-critic.
  DOI: 10.1504/IJVSMT.2025.10076340
   
  
• Selecting new energy vehicle product concepts: integrating best-worst method and quality function deployment  
  by Yan Gou, Jian-peng Chang, Jun Zhang, Qun Bai, Xing-yi Zou, Mei Cao 
  Abstract: Intense competition in the rapidly growing new energy vehicle market demands optimal product concept selection. Current methods often determine the importance of technical characteristics based primarily on customer requirements, neglecting critical factors like competitor performance and technological feasibility, and thus typically lack mechanisms to translate technical performance into subjective customer satisfaction for ranking. This paper presents a novel model integrating the Best-Worst Method and Quality Function Deployment to comprehensively evaluate technical characteristics by considering customer requirements, competitor performance, technological feasibility, and company strengths. Customer satisfaction is quantified using prospect theory's value function, enabling ranking based on comprehensive satisfaction scores. A case study of a traditional automaker entering the new energy vehicle market demonstrates the model's effectiveness. Findings emphasize aligning product development with customer preferences and market dynamics to improve competitiveness.
  Keywords: New energy vehicles; product concept selection; Best-Worst Method; Quality Function Deployment; prospect theory.
  DOI: 10.1504/IJVSMT.2026.10076384
   
  
• MATSim research on urban traffic simulation and optimisation strategy in collaboration with SUMO  
  by Jian Ma, Qianlong Fu, Liyan Zhang, Hairong Gu, Yuchen Zhang 
  Abstract: This paper presents mesoscopic and microscopic traffic simulations to address urban congestion challenges in a context of rapid urbanization. A co-simulation platform was developed by combining the multi-agent transport simulation (MATSim) agent-based modeling with the simulation of urban mobility (SUMO). The methodology incorporates deep Q-networks for adaptive traffic signal control and genetic algorithms for an optimal route planning, and the co-simulation approach can reduce bus waiting times by 29.10% and increased intersection throughput by 8.14% through a memory palace-enhanced DQN model for adaptive signal control., The genetic algorithms optimized route planning for 50,000 simulated residents. The utility-discrete choice framework improved travel demand realism by translating utility scores into probabilistic selections.Multi-stage optimization balanced system efficiency with individual preferences, proving 25% faster convergence than baseline DQN. This computationally efficient platform bridges macroscopic policy analysis with microscopic behavioral modeling, offering robust decision support for urban traffic management.
  Keywords: traffic simulation; MATSim; SUMO; joint simulation; congestion relief.
  DOI: 10.1504/IJVSMT.2025.10076966
   
  
• Task allocation for multi-unmanned vehicles under dynamic stochastic interference  
  by Xin Zhang, Hao Ge, Xuting Duan, Haiying Xia, Jianshan Zhou 
  Abstract: Dynamic environmental changes introduce random interference that can compromise performance of multi-unmanned vehicle systems. To mitigate these challenges, this paper proposes a novel dynamic task allocation approach based on a distributed market auction mechanism. Our method incorporates a priority-based conflict resolution strategy to effectively manage tie situations during auctions, ensuring fair and efficient task distribution. Additionally, an innovative task insertion scheme integrates new tasks into existing sequences without reordering all pending tasks, thereby reducing computational overhead while maintaining system stability. Simulation experiments in a Gazebo environment demonstrate that our approach significantly improves adaptability under conditions of vehicle failure and temporary task additions. The proposed framework is validated through comprehensive simulations, confirming its ability to handle both planned and unexpected scenarios with minimal performance degradation. This work sets a new benchmark for dynamic task management in unmanned systems.
  Keywords: dynamic task allocation; multi-unmanned vehicles; auction algorithm; vehicle priority; distributed market mechanism; conflict resolution; task insertion; stochastic interference; Gazebo simulation; travel cost optimisation.
  DOI: 10.1504/IJVSMT.2026.10077219
   
  
• Study on the effects of raster spacing and raster angle on the mechanical properties of FDM-printed parts  
  by Na Qiu, Jiaxuan Wu, Xiaomin Wu, Yifei Ren 
  Abstract: 3D printing is widely used for manufacturing various automotive components. However, structures fabricated using fused deposition modelling (FDM) technology exhibit anisotropic properties, making it crucial to study the impact of process parameters on their mechanical performance. This study investigates the influence of raster spacing and angle on the mechanical properties of FDM-printed structures. The results revealed that a 0
  Keywords: FDM; fused deposition modelling; automotive energy absorber; mechanical properties; fracture modes.
  DOI: 10.1504/IJVSMT.2025.10077270
   
  
• A lightweight methodology for trajectory prediction of powered two-wheeler in pre-crash scenarios  
  by Yong Han, Changyi Liu, Yanting LI, Di Pan, Hui LIu, Xiaojiang Lv, Koji Mizuno 
  Abstract: To tackle real-time trajectory prediction for powered two-wheelers (PTWs) in pre-crash scenarios, this study proposes TPtiny, a lightweight framework integrating an anchorfree detector (VPDet), a groupedconvolution tracker (MobileSort), and a VAEGAN hybrid predictor (VGAN). Evaluated on dense traffic and real accident data, TPtiny achieves comparable accuracy to stateoftheart models while being significantly smaller and faster, providing 218.8281.3 ms additional decision time. The framework demonstrates high efficiency, generalisability, and suitability for embedded vehicle safety platforms.
  Keywords: intelligent vehicle safety; trajectory prediction; deep learning; lightweight neural network; powered two-wheelers.
  DOI: 10.1504/IJVSMT.2025.10077538
   
  
• A web-based visualisation system for the impact of hazard on transportation using Cesium digital Earth  
  by Yang Feng, Shikun Xie, Zhen Yang 
  Abstract: As global climate change continues to worsen, natural hazards are occurring more frequently, affecting transportation operations. To effectively manage traffic control and hazard rescue operations during hazards, it is essential to have accurate geographic information on the location and impact of these hazards during such events, and to predict the future traffic affected by the hazards. To improve this process, a system by integrating advanced web geographic information system (GIS) technology with dynamic traffic network model was developed to visualise the impact of hazards on road traffic. A town named Hutiaoxia of Yunnan province was selected as a case study to perform 6 h simulation. The simulation results showed that the roads close to the landslide site experienced a reduction in average speed and congestion. This paper will aid in making accurate traffic control decisions and coordinating rescue efforts.
  Keywords: hazard; GIS; geographic information system; Cesium digital Earth; transportation structures; SUMO; simulation of urban mobility.
  DOI: 10.1504/IJVSMT.2025.10077539