The traditional development of Child Restraint System (CRS) safety relies on simulation and physical test results, which suffers from high costs, low efficiency and obvious limitations. The paper constructes a simplified high-precision coupled model under standard working conditions to efficiently generate training datasets. By fusing the time-domain, frequency-domain and second-order waveform fitting features of crash pulses, an intelligent prediction model based on deep multilayer perceptron is adopted to rapidly predict 4 core safety indicators, namely the Head Injury Criterion (HIC), head acceleration, axial neck force and chest acceleration. Validation results on new vehicle models show that the prediction model achieves an average accuracy of 93.3%, which can support the rapid development of CRS safety and demonstrates excellent engineering applicability.

To address the high-voltage electrical safety risk of a hybrid SUV during 80 km/h rear impacts as specified by FMVSS 305, component-level investigations are conducted, and vehicle collision structure optimization is performed. At the component level, through designing two types of component tests, quasi-static extrusion and dynamic impact, the ultimate bearing capacity of Intelligent Power System (IPS) is obtained. At the level of vehicle collision structure optimization, by optimizing the rear impact force transmission path and improving the IPS collision mode, designing the rear longitudinal beam structure to bend upward makes IPS be subjected to planar extrusion. The simulated IPS collision force decreases by 64% from 137 kN to 49 kN, which is less than its ultimate bearing capacity.

The paper focuses on the sideslip problem in the small overlap crash test in the vehicle structural safety field. A two-degree-of-freedom first-order dynamic model of the vehicle is established, based on which a segmented sideslip motion planning method is proposed, and a vehicle sideslip motion planning software for small-overlap crash conditions is developed. The experimental results show that the proposed planning method enhances the vehicle's lateral displacement by 30%, thereby facilitating successful sideslip. This elevates the vehicle structural rating from Acceptable (A) to Good (G) and significantly improves the safety performance in the small-overlap ceashes.

To improve the body crashworthiness of a battery electric SUV under extreme conditions, the research employs a collision energy management method to conduct simulation under two side impact conditions of the China Insurance Automotive Safety Index (C-IASI) and truck. The body transmission path, load-bearing target definitions for critical structures, structural deformation and failure modes are compared and analyzed. Furthermore, the body structure is optimized targeting the extreme truck side impact condition. Simulation results and physical vehicle test results demonstrate that the body crashworthiness under extreme condition is significantly improved after optimization.

In the analysis of large-scale traffic accident data, traditional statistical methods suffer from low computational efficiency and limited ability to identify causes, making it difficult to effectively uncover deep-seated associations behind accidents. To address this issue, this paper proposes a traffic accident cause analysis method based on the Human Factors Analysis and Classification System (HFACS) framework combined with parallel Apriori algorithms. The method is developed from five aspects, including feature engineering, HFACS framework mapping and transactional structure transformation. Experimental results show that this approach significantly enhances the efficiency and accuracy of cause analysis in large-scale traffic accident data compared to traditional statistical analysis methods, providing more scientific data support for traffic accident prevention and safety management.

To address the issue that static path planning methods struggle to respond to dynamic traffic conditiond and easily cause local congestion in underground parking lot guifance, this paper proposes a parking space guidance method tailored for Intelligent Connected Vehicles (ICV). The proposed method leverages Vehicle-to-Everything (V2X) communication technology to acquire real-time information on parking space availability and traffic flow. It employs a time window model to predict potential spatiotemporal conflicts during vehicle movement and coordinates vehicle passage according to predefined priority rules. To mitigate congestion arising from these conflicts, the estimated waiting time due to conflicts is quantified as a dynamic increment to the edge weights of the affected road sections. On this basis, the D*Lite algorithm is integrated to perform path re-planning, achieving spatiotemporal cooperative optimization. Based on a model replicating a real parking structure, Simulation tests demonstrate that compared to traditional methods, the proposed approach significantly reduces the average parking search time across various scenarios with different occupancy levels within the facility, thereby effectively enhancing the overall parking efficiency.

To address the issue of high fuel consumption in heavy-duty truck druing hill-climing conditions, a fuel comsumption predicton model balancing accuracy and real-time performance is developed by inregrating parameters such as engine speed and torque. This paper proposes a fuel-saving strategy for heavy trucks using Predictive Adaptive Cruise Control (PACC) on sloped roads. A hierarchical architecture is developed: the upper layer employs Dynamic Programming (DP) for global economic speed planning, while the lower layer uses Model Predictive Control (MPC) for secondary planning to ensure safe following and energy efficiency. A fuel consumption model considering engine torque and speed is integrated. Vehicle test demonstrates an average fuel reduction of 1.5% per kilometer.