Current Issue : October-December Volume : 2023 Issue Number : 4 Articles : 5 Articles
An online policy learning algorithm is used to solve the optimal control problem of the power battery state of charge (SOC) observer for the first time. The design of adaptive neural network (NN) optimal control is studied for the nonlinear power battery system based on a second-order (RC) equivalent circuit model. First, the unknown uncertainties of the system are approximated by NN, and a time-varying gain nonlinear state observer is designed to address the problem that the resistance capacitance voltage and SOC of the battery cannot be measured. Then, to realize the optimal control, a policy learning-based online algorithm is designed, where only the critic NN is required and the actor NN widely used in most design of the optimal control methods is removed. Finally, the effectiveness of the optimal control theory is verified by simulation....
Efficient navigation of off-road vehicles heavily relies on the ability to accurately model the interaction between the vehicle and the terrain. One of the most important parts of this interaction is the deformation of the terrain and the tire. Although high-precision methods like finite element method (FEM) simulation can be used for this purpose, they require significant computational power, which is impractical to install in a vehicle for real-time navigation purposes. Therefore, simplified and less-detailed models are essential for on-board installation in real-time applications. In this study, three two-dimensional static terrain–vehicle models are compared to a detailed FEM reference model, and the results are evaluated both from the perspective of accuracy and computational capacity requirements. The analysis sheds light on the effectiveness of each model in the real-time navigation of off-road vehicles....
As an advanced driver assistance system, automatic emergency braking (AEB) can effectively reduce accidents by using high-precision and high-coverage sensors. In particular, it has a significant advantage in reducing front-end collisions and rear-end accidents. Unfortunately, avoiding side collisions is a challenging problem for AEB. To tackle these challenges, we propose active seat belt pretensioning on driver injury in vehicles equipped with AEB in unavoidable side crashes. Firstly, records of impact cases from China’s National Automobile Accident In-Depth Investigation System were used to investigate a scenario in which a vehicle is impacted by an oncoming car after the vehicle’s AEB system is triggered. The scenario was created using PreScan software. Then, the simulated vehicles in the side impact were devised using a finite element model of the Toyota Yaris and a moving barrier. These were constructed in HyperMesh software along with models of the driver’s side seatbelt, side airbag, and side curtain airbag. Moreover, the models were verified, and driver out-of-position instances and injuries were evaluated in simulations with different AEB intensities up to 0.7 g for three typical side impact angles. Last but not least, the optimal combination of seatbelt pretensioning and the timing thereof for minimizing driver injury at each side impact angle was identified using orthogonal tests; immediate (at 0 ms) pretensioning at 80 N was applied. Our experiments show that our active seatbelt with the above parameters reduced the weighted injury criterion by 5.94%, 22.05%, and 20.37% at impact angles of 90◦, 105◦, and 120◦, respectively, compared to that of a conventional seatbelt. The results of the experiment can be used as a reference to appropriately set the collision parameters of active seat belts for vehicles with AEB....
Hydrogen and ammonia are primary carbon-free fuels that have massive production potential. In regard to their flame properties, these two fuels largely represent the two extremes among all fuels. The extremely fast flame speed of hydrogen can lead to an easy deflagration-todetonation transition and cause detonation-type engine knock that limits the global equivalence ratio, and consequently the engine power. The very low flame speed and reactivity of ammonia can lead to a low heat release rate and cause difficulty in ignition and ammonia slip. Adding ammonia into hydrogen can effectively modulate flame speed and hence the heat release rate, which in turn mitigates engine knock and retains the zero-carbon nature of the system. However, a key issue that remains unclear is the blending ratio of NH3 that provides the desired heat release rate, emission level, and engine power. In the present work, a 3D computational combustion study is conducted to search for the optimal hydrogen/ammonia mixture that is knock-free and meanwhile allows sufficient power in a typical spark-ignition engine configuration. Parametric studies with varying global equivalence ratios and hydrogen/ammonia blends are conducted. The results show that with added ammonia, engine knock can be avoided, even under stoichiometric operating conditions. Due to the increased global equivalence ratio and added ammonia, the energy content of trapped charge as well as work output per cycle is increased. About 90% of the work output of a pure gasoline engine under the same conditions can be reached by hydrogen/ammonia blends. The work shows great potential of blended fuel or hydrogen/ammonia dual fuel in high-speed SI engines....
In actual traffic scenarios, the environment is complex and constantly changing, with many vehicles that have substantial similarities, posing significant challenges to vehicle tracking research based on deep learning. To address these challenges, this article investigates the application of the DeepSORT (simple online and realtime tracking with a deep association metric) multitarget tracking algorithm in vehicle tracking. Due to the strong dependence of the DeepSORTalgorithm on target detection, a YOLOv5s_DSC vehicle detection algorithm based on the YOLOv5s algorithm is proposed, which provides accurate and fast vehicle detection data to the DeepSORTalgorithm. Compared to YOLOv5s, YOLOv5s_DSC has no more than a 1% difference in optimal mAP0.5 (mean average precision), precision rate, and recall rate, while reducing the number of parameters by 23.5%, the amount of computation by 32.3%, the size of the weight file by 20%, and increasing the average processing speed of each image by 18.8%. After integrating the DeepSORTalgorithm, the processing speed of YOLOv5s_DSC + DeepSORTreaches up to 25 FPS, and the system exhibits better robustness to occlusion....
Loading....