No abstract

No abstract

"The automotive OEMs undertake a large amount of research and engineering work for the development of passive safety systems of their new vehicle models. The development of automotive passive safety system is susceptible to errors due to even minor variations in structural properties, material properties, and/or manufacturing process parameters. Assuming the sensor mounting positions and other external dependencies are fixed, there are data analysis models that can cause variations in the triggering of restraint systems. Every new development requires a fine calibration of the algorithm to trigger respective restraint systems. The algorithm is sensitive to the crash behavior of vehicle components and assemblies. Every iteration in the product development cycle must be subjected to crash tests and validation process. As the VUTs (vehicles under test) are pre-series production models, the costs of these tests are not negligible. Hence, a model that can virtually utilize the crash mechanics and deformation behavior and provide a fast baseline prediction of the crash pulse can be extremely helpful in the virtual calibration of the crash algorithm, thus reducing the effort, time, and cost for the development. The virtual methodologies for simulating crash behavior are capable of providing its variation trend due to product evolution in the development process. However, there are considerable variations between simulated crash pulses and real-time crash pulses. The respective crash test data of the completed product development cycles can be analyzed to model the delta between the virtual and real crash behavior. The analytical and predictive capabilities of such a mathematical model can be further expanded by adding new node points using respective sets of virtual and real crash tests. The key challenge in quantifying and modeling the differences between crash behaviors and establishing patterns is to develop a consistent methodology that comprehensively considers all the physical dependencies of this complex problem. In the scope of this research, a methodology is proposed, which builds on the conservation of total crash energy. The total input energy is traced to total energy spent during the crash to analyze the dissipation behavior. The input energy available at T0 (time instance of the first contact) is primarily a function of the differential velocity of the vehicle(s). The energy dissipated from T0 to Tend (end of evaluation time) can be categorized into – energy spent on deformation, kinematic energy, and energy losses due to e.g. friction, etc. The quantification of the first component is achieved using acceleration sensors at strategic points on the vehicle body. The quantification of kinematics is done using the photogrammetry approach. The energy losses are ignored. The individual components of energy provide a factor to attribute the difference in crash behavior to component level causes. Hence, the quantification model can place the nodes causing the delta, their respective magnitudes, and the effects in the overall variation of crash behavior. The variations could be further analyzed to identify patterns and regularities using machine-learning algorithms. Therefore, the study explores a novel approach for crash comparison and discusses the application using crash test results."

The development of finite element vehicle models for crash simulations is a highly complex task. The main aim of these models is to simulate a variety of crash scenarios and assess all the safety systems for their respective performances. These vehicle models possess a substantial amount of data pertaining to the vehicle's geometry, structure, materials, etc., and are used to estimate a large set of system and component level characteristics using crash simulations. It is understood that even the most well-developed simulation models are prone to deviations in estimation when compared to real-world physical test results. This is generally due to our inability to model the chaos and uncertainties introduced in the real world. Such unavoidable deviations render the use of virtual simulations ineffective for the calibration process of the algorithms that activate the restraint systems in the event of a crash (crash-detection algorithm). In the scope of this research, authors hypothesize the possibility of accounting for such variations introduced in the real world by creating a feedback loop between real-world crash tests and crash simulations. To accomplish this, a Reinforcement Learning (RL) compatible virtual surrogate model is used, which is adapted from crash simulation models. Hence, a conceptual methodology is illustrated in this paper for developing an RL-compatible model that can be trained using the results of crash simulations and crash tests. As the calibration of the crash-detection algorithm is fundamentally dependent upon the crash pulses, the scope of the expected output is limited to advancing the ability to estimate crash pulses. Furthermore, the real-time implementation of the methodology is illustrated using an actual vehicle model.

No abstract