International Journal of Vehicle Design (7 papers in press)

Regular Issues

• This is a test paper, pleaseignore it
  by ReviewerV ReviewerC 
  Abstract: This is a test submission. Please ignore it
  Keywords: test test test test test test test test test test test test test test test test.
  
  
• Transient dynamic behavior of vehicle cornering on ?-step road  
  by Shengyong Ye, Jianwei Lu, Lei Shi, Yingjie Zhu, Ruiming Xu 
  Abstract: In this paper, the transient dynamic behavior of vehicle cornering on ?-step road is studied. Particularly, the transient characteristic of tire lateral force caused by the variation of road friction coefficient is fully considered in the establishment of mathematical model. The transient response of the tire lateral force and vehicle motion under different conditions are obtained by solving the model. The tire lateral force saturation caused by sudden reduction of road friction coefficient is analyzed, and the critical initial steer angle is calculated. Then, the kinematic differential equation of vehicle cornering is used to analyze the vehicle trajectory. Moreover, an experiment of transient tire force testing is carried out to verify the effectiveness of the proposed method. The simulation and test results show that the change of road condition affects the vehicle motion state and tire lateral force, and causes the deviation of vehicle trajectory.
  Keywords: Vehicle; Cornering; Transient dynamics; ?-step road; Tire force saturation.
  
  
• Research on active suspension control based on safety constraint reinforcement learning  
  by Lin-feng Zhao, Xiao Feng, Wen-bin Shao, Zhen Mei, Jin-fang Hu 
  Abstract: To tackle the shortcomings of conventional active suspension control algorithms, which struggle to achieve a harmonious balance between ride comfort and handling stability, as well as obvious lack of practical safety considerations for suspension systems in ordinary reinforcement learning algorithms, a Deep Deterministic Policy Gradient(DDPG) Reinforcement Learning(RL) algorithm is proposed, which utilize suspension system safety boundary constraints as its foundation, and integrates the suspension dynamic deflection's limit stroke, as well as safety range of tire dynamic deformation into the reinforcement learning algorithm, and obtains the ideal active suspension control strategy through offline training. Based on MATLAB/Simulink, simulation comparison and HIL(Hardware In the Loop) verification were conducted with Linear Quadratic Regulator (LQR) and ordinary DDPG algorithm under convex road conditions. The results clearly indicates a distinction from the LQR algorithm, in contrast to which the sprung acceleration's rms(root mean square) value using algorithm proposed in this paper is reduced by 6.41%?
  Keywords: active suspension; reinforcement learning; safe boundaries; ride comfort; convex road surface.
  DOI: 10.1504/IJVD.2024.10066193
   
  
• Research on a collaborative optimisation strategy for the component parameters and control parameters of hybrid mining trucks  
  by Hongliang Li, Lihua Shang 
  Abstract: Considering the interactive effects of component parameters, energy management control parameters, and driving conditions on energy use optimisation in hydraulic hybrid mining trucks, synchronous-control and nested-control collaborative design strategy models were established to jointly optimise the component and control coupling parameters, and their performances were analysed. For fairness, the same combination optimisation algorithm and vehicle model were used for both optimisation strategies, and suitable energy management strategies were selected for each. Both strategies could quickly perform global searches, locate effective regions, and use combination algorithms that quickly converge. The nested-control strategy yielded better results than the synchronous-control strategy. The results of both strategies were better than the those of the optimal parameter combination, indicating that the combination optimisation algorithm was effective. However, using this algorithm and a high-configuration algorithm with the synchronous control strategy significantly improved the results despite increasing time costs. Such an improvement was not found for nested control.
  Keywords: hydraulic hybrid mining truck; integrated optimisation strategy; collaborative optimisation strategy; coupling parameters; joint optimisation.
  DOI: 10.1504/IJVD.2024.10067218
   
  
• Complete coverage path planning algorithm for multiple agricultural robots  
  by Changjie Liu, Haobo Zhang, Yangjie Ji 
  Abstract: Multi-robot complete coverage path planning (MCCPP) is an important direction in developing intelligent agricultural robots. Firstly, to address the problem that the existing region decomposition algorithm has too many subregions and contains concave subregions, this paper adopts the improved Maklink line to convexly decompose the workspace to obtain the minimum number of convex subregions. Secondly, the current MCCPP algorithm suffers from duplicate coverage of connection paths, uneven task allocation, and failure to consider the robot's extra energy consumption. This paper adopts the Dijkstra algorithm to plan the shortest non-duplicated connected paths between any subregions; improves the existing objective function by combining with the actual; and retains the high-quality gene fragments for chromosome crossover according to the breakpoints. Finally, the improved Non-dominated Sorting Genetic Algorithm (NSGA-II) is simulated in real planting areas, and the total connected paths and planting area balance were optimized compared to the traditional NSGA-II.
  Keywords: precision agriculture; multi-robots; complete coverage path planning; NSGA-II; non-dominated sorting genetic algorithm.
  DOI: 10.1504/IJVD.2024.10067304
   
  
• Research on gravity potential energy echelon recovery and speed control of industrial vehicle  
  by Guang Xia, Yongcheng Pan, Xiwen Tang, Yang Zhang, Shaojie Wang 
  Abstract: This paper proposes an innovative echelon recovery scheme to address inefficiencies in conventional forklift operations, particularly the dissipation of gravitational potential energy during frequent lifting and lowering cycles. The hydraulic oil circuit, with throttling control mechanisms impacting temperature and lubrication, contributes to reduced engine efficiency. The solution involves three load-specific accumulators with predetermined initial pre-inflation pressure and volume. Dynamic models are established for the lifting cylinder, accumulator, fork, and solenoid valve. Using a Model Predictive Control (MPC)-based hierarchical strategy, the upper decision-making layer selects the target accumulator for energy recovery, while the lower predictive control layer utilizes MPC to govern fork descent speed. Simulation and experimental results demonstrate a notable 40% enhancement in energy recovery efficiency compared to traditional hydraulic systems. The hierarchical control strategy effectively limits fork descent speed fluctuation to around 10%, significantly improving both gravitational potential energy recovery rates and operational smoothness in forklift fork functionality.
  Keywords: fork gravity potential energy echelon recovery; accumulator pre-charge pressure; model predictive control; fork falling speed control.
  DOI: 10.1504/IJVD.2024.10067922
   
  

Special Issue on: New Energy Vehicles' NVH and Lightweight and Control Technologies

• Comparison of deep learning methods for predicting charging energy of power batteries  
  by Xuefeng Zhu, Guoliang Xie 
  Abstract: Accurate prediction of the electric vehicle charging energy is essential for power grid companies to rationally allocate power resources, customise appropriate tariffs and select the location of charging piles. Currently, machine learning methods have been widely applied in this field. Aiming to predict Electric Vehicles (EV) charging energy more precisely, this paper compares several machine learning methods and concludes that Long Short-Term Memory (LSTM) neural network has better behaviour. As the initial training data was incomplete, we supplemented the training data with MissFrorest neural network. We compared Back Propagation (BP), Xtreme Gradient Boosting (XGBoost), and LSTM networks for the prediction of charging energy, and found that LSTM has the best prediction effect, XGBoost has the second best, and BP has the worst effect. LSTM addresses the issue of gradient dispersion due to introducing time series, and thus has a better prediction effect. The experimental results show that Mean Absolute Error (MAE) and Root-mean-square error (RMSE) indices for four of five experimental vehicles using the LSTM algorithm are smaller than those using BP and XGBoost methods. Compared with the BP, XGBoost algorithms, the average reduction of MAE is 42.79%, 23.48%, and RMSE is reduced by 43.42%, 19.65%.
  Keywords: predicting charging energy; deep learning; power batteries; electric vehicles.