Search Results

To adapt to Battery Electric Vehicle (BEV) integration, the significance of protective designs for battery packs against ground impact caused by road debris is very high, and there is also a keen interest in the feasibility assessment technique using Computer-Aided Engineering (CAE) tools for prototype-free evaluations. However, the challenge lies in obtaining real-world empirical data to verify the accuracy of the predictive CAE model. Collecting real-world data using actual battery pack can be time-consuming, costly, and accurately ascertaining the precise direction, magnitude, and location of the force applied from the road to the battery pack poses a challenging task. Therefore, in this study, we developed a methodology using machine learning, specifically Gaussian process regression (GPR), to perform inverse analysis of the direction, magnitude, and location of vehicle-road contact forces during rough road conditions.

Rubber isolators are widely used under random vibrations. In order to predict their fatigue life, a study on the fatigue analysis methodology for rubber isolators is carried out in this paper. Firstly, taking a mount used for isolating air conditioning compressor vibrations as studying example, accelerations versus time of rubber isolator at both sides are acquired for a car under different running conditions. The acceleration in time domain is transformed to frequency domain using the Fourier transform, and the acceleration power spectral density (PSD) is the obtained. Using the PSD as input, fatigue test is carried for the rubber isolator in different temperature and constant humidity conditions. A finite element model of the rubber isolator using ABAQUS is established for estimating fatigue life, and model validity is verified through static characteristic testing. Dynamic responses of the rubber isolator at frequency domain are calculated if a unit load is applied.

Vibration from a mechanical system not only produces unwanted noises annoying to people around, but also runs a risk of fatigue failure that would actually hinder its functionality. There are several forms of vibration depending on the sources of excitation forms. Mechanical systems with rotating components can be subjected to sinusoidal excitation due to the fact the center of mass is not perfectly aligned with the rotating axis. If the rotating speed is strictly ramping up or ramping down, this can create an excitation whose frequency is changing with time in a frequency range corresponding to the speeds swept. Compared with a single sinusoidal excitation, the issue with fatigue at swept sinusoidal excitation, is that as it sweeps through a wide frequency range, some swept frequencies will definitely coincide with the natural frequencies of the system. Certainly, the stress response exactly at the resonant frequency becomes the highest and could account for a lot of fatigue damage.

Recently, topology optimization (TO) has seen increased usage in the automotive industry as a numerical tool, greatly enhancing the accessibility and production-readiness of optimal, lightweight solutions. By natural extension of classic single material TO (SMTO), a wealth of research has been completed in multi-material TO (MMTO), enabling simultaneous determination of material selection and existence. MMTO is effective for linear static analyses, making use of structural responses that are continuously differentiable, giving itself to efficient gradient-based optimization engines. A structural response that is inherently nonlinear and transient, thus providing difficulty to the mainstay MMTO process, is that of crashworthiness. This paper presents a multi-objective MMTO framework considering crashworthiness using the equivalent static load (ESL) method. The ESL method uses a series of linear static sub-models to approximate the transient crashworthiness model.

In recent years, an increase in vehicle weight due to the electrification of automobiles, specifically EVs, has increased the input loads on anti-vibration rubber parts. Moreover, the characteristics of these loads have also changed due to the rotational drive of electric motors, regenerative braking, and other factors. When designing a vehicle, in advance it is necessary to set specifications that take into account the spring characteristics and durability of the anti-vibration rubber parts in order to meet functional requirements. In this study, the hyperelastic and fatigue characteristics (S-N diagram and Haigh diagram) of Rubbers which is widely used for anti-vibration rubber parts, were experimentally obtained, and structural and fatigue analyses using FEM (Finite Element Method) were conducted in conjunction with spring and fatigue tests of anti-vibration rubber parts to determine the correlation between their spring and fatigue characteristics.

The fatigue prediction model of an air spring based on the crack initiation method is established in this study. Taking a rolling lobe air spring with an aluminum casing as the studying example, a finite element model for analyzing force versus displacement is developed. The static stiffness and dimensional parameters of limit positions are calculated and analyzed. The influence of different modeling methods of air springs bellow are compared and analyzed. Static stiffness measurement of an air spring is conducted, and the calculation results and the measured results of the static stiffness are compared. It is shown that the relative error of the measured stiffness and calculated stiffness is within 1%. The Abaqus post-processing stage is redeveloped in Python language.

This study deals with the fatigue life prediction methodology of welding simulation components involving arc welding. First, a method for deriving the cyclic deformation and fatigue properties of the weld metal (that is also called ER70S-3 in AWS, American Welding Standard) is explained using solid bar specimens. Then, welded tube specimens were used with two symmetric welds and subjected to axial, torsion, and combined in-phase and out-of-phase axial-torsion loads. In most previous studies the weld bead’s start/stop were arbitrarily removed by overlapping the starting and stop point. Because it can reduce fatigue data scatter. However, in this study make the two symmetric weld’s start/stops exposed to applying load. Because the shape of the weld bead generated after the welding process can act as a notch (Ex. root notch at weld start / Crater at weld stop) to an applied stress. Accordingly, they were intentionally designed to cause stress concentrations on start/stops.

A special spot weld element (SWE) is presented for simplified representation of spot joints in complex structures for structural durability evaluation using the mesh-insensitive structural stress method. The SWE is formulated using rigorous linear four-node Mindlin shell elements with consideration of weld region kinematic constraints and force/moments equilibrium conditions. The SWEs are capable of capturing all major deformation modes around weld region such that rather coarse finite element mesh can be used in durability modeling of complex vehicle structures without losing any accuracy. With the SWEs, all relevant traction structural stress components around a spot weld nugget can be fully captured in a mesh-insensitive manner for evaluation of multiaxial fatigue failure.

This paper presents deep learning-based prognostics and health management (PHM) for predicting fractures of an electric propulsion (eP) drivetrain system using real-time CAN signals. The deep learning algorithm, based on autoencoders, resamples time-series signals and converts them into 2D images using recurrence plots (RP). Subsequently, through unsupervised learning of DeepSVDD, it detects anomalies in the converted 2D images and predicts the failure of the system in real-time. Also, reliability analysis based on fracture mechanics was performed using the detected signals and big data. In particular, the severity of the eP drivetrain system is proportional to the maximum shear stress (τmax) in terms of linear elastic fracture mechanics (LEFM) and can be calculated by summarizing the relationship between cracks (a) and the stress intensity factor (KIII).

Fretting is a phenomenon in which a fatigue crack is initiated by a small relative slip between two objects, resulting in crack propagation and fracture at stresses far below the fatigue limit [1, 2]. Since the mechanism behind fretting is complex and covers multiple disciplines, it is not easy to develop a consistent evaluation method. In the field of engine development, fretting events can also pose an issue due to the complexity of the mechanism [3]. In particular, it has been a challenge to help predict changes in the presence and severity of fretting events, as the engine temperature fluctuates with operating conditions. As one method for evaluating fretting, Sato, et al. have made predictions using analytical models based on the finite element method (FEM) [4, 5]. However, their predictions did not take into account temperature fluctuations in the system, and they were unable to predict events in which the occurrence of fretting fatigue changed with temperature fluctuations.

An advanced multi-layer material model has been developed to simulate the complex behavior in case-carburized gears where hardness dependent strength and elastic-plastic behavior is characterized. Also, an advanced fatigue model has been calibrated to material fatigue tests over a wide range of conditions and implemented in FEMFAT software for root bending fatigue life prediction in differential gears. An FEA model of a differential is setup to simulate the rolling contact and transient stresses occurring within the differential gears. Gear root bending fatigue life is predicted using the calculated stresses and the FEMFAT fatigue model. A specialized rig test is set up and used to measure the fatigue life of the differential over a range of load conditions. Root bending fatigue life predictions are shown to correlate very well with the measured fatigue life in the rig test.

Optimizing the specifications of the parts that make up the vehicle is essential to develop a high performance and quality vehicle with price competitiveness. Optimizing parts specifications for quality and affordability means optimizing various factors such as engineering design specifications and manufacturing processes of parts. This optimization process must be carried out in the early stages of development to maximize its effectiveness. Therefore, in this paper, we studied the methodology of building a database for parts of already developed vehicles and optimizing them on a data basis. A methodology for collecting, standardizing, and analyzing data was studied to define information necessary for specification optimization. In addition, AI technology was used to derive optimization specifications based on the 3D shape of the parts. Through this study, body parts specification optimization system using AI technology was developed.

As a key component of in-vehicle intelligent voice technology, speech enhancement can extract clean speech signals contaminated by environmental noise to improve the perceptual quality and intelligibility of speech. It has extensive applications in the field of intelligent car cabins. Although some end-to-end speech enhancement methods based on time domain have been proposed, there is often limited consideration given to designing model architectures based on the characteristics of the speech signal. In this paper, we propose a new U-Net based speech enhancement framework that utilizes the temporal correlation of speech signals to reconstruct higher-quality and more intelligible clean speech.

The accurate prediction of engine performance maps can guide data-driven optimization of engine technologies to control fuel use and associated emissions. However, engine operational maps are scarcely reported in literature and often have missing data. Assessment of missing-data resilient algorithms in the context of engine data prediction could enable better processing of real-world driving cycles, where missing data is a more pervasive phenomenon. The goal of this study is, therefore, to determine the most effective technique to deal with missing data and employ it in prediction of engine performance characteristics. We assess the performance of two machine learning approaches, namely Artificial Neural Networks (ANNs) and the extreme tree boosting algorithm (XGBoost), in handling missing data.

Modeling uncertainties pose a significant challenge in the development and deployment of model-based vehicle control systems. Most model- based automotive control systems require the use of a well estimated vehicle dynamics prediction model. The ability of first principles-based models to represent vehicle behavior becomes limited under complex scenarios due to underlying rigid physical assumptions. Additionally, the increasing complexity of these models to meet ever-increasing fidelity requirements presents challenges for obtaining analytical solutions as well as control design. Alternatively, deterministic data driven techniques including but not limited to deep neural networks, polynomial regression, Sparse Identification of Nonlinear Dynamics (SINDy) have been deployed for vehicle dynamics system identification and prediction.

In an era characterized by the rapid proliferation and advancement of AI-based technologies across various domains, the spotlight is placed on the integration of these technologies into trustworthy autonomous systems. The integration into embedded systems necessitates a heightened focus on dependability. This paper combines the findings from the TEACHING project, which delves into the foundations of humanistic AI concepts, with insights derived from an expert workshop in the field of dependability engineering. We establish the body of knowledge and key findings deliberated upon during an expert workshop held at an international conference focused on computer safety, reliability and security. The dialogue makes it evident that despite advancements, the assurance of dependability in AI-driven systems remains an unresolved challenge, lacking a one-size-fits-all solution.

Speech enhancement can extract clean speech from noise interference, enhancing its perceptual quality and intelligibility. This technology has significant applications in in-car intelligent voice interaction. However, the complex noise environment inside the vehicle, especially the human voice interference is very prominent, which brings great challenges to the vehicle speech interaction system. In this paper, we propose a speech enhancement method based on target speech features, which can better extract clean speech and improve the perceptual quality and intelligibility of enhanced speech in the environment of human noise interference. To this end, we propose a design method for the middle layer of the U-Net architecture based on Long Short-Term Memory (LSTM), which can automatically extract the target speech features that are highly distinguishable from the noise signal and human voice interference features in noisy speech, and realize the targeted extraction of clean speech.

Data-driven driving safety assessment is crucial in understanding the insights of traffic accidents caused by dangerous driving behaviors. Meanwhile, quantifying driving safety through well-defined metrics in real-world naturalistic driving data is also an important step for the operational safety assessment of automated vehicles (AV). However, the lack of flexible data acquisition methods and fine-grained datasets has hindered progress in this critical area. In response to this challenge, we propose a novel dataset for driving safety metrics analysis specifically tailored to car-following situations. Leveraging state-of-the-art Artificial Intelligence (AI) technology, we employ drones to capture high-resolution video data at 12 traffic scenes in the Phoenix metropolitan area. After that, we developed advanced computer vision algorithms and semantically annotated maps to extract precise vehicle trajectories and leader-follower relations among vehicles.

During the development of an Internal Combustion Engine-based powertrain, traditional procedures for control strategies calibration and validation produce huge amount of data, that can be used to develop innovative data-driven applications, such as emission virtual sensing. One of the main criticalities is related to the data quality, that cannot be easily assessed for such a big amount of data. This work focuses on an emission modeling activity, using an enhanced Light Gradient Boosting Regressor and a dedicated data pre-processing pipeline to improve data quality. First thing, a software tool is developed to access a database containing data coming from emissions tests. The tool performs a data cleaning procedure to exclude corrupted data or invalid parts of the test. Moreover, it automatically tunes model hyperparameters, it chooses the best set of features, and it validates the procedure by comparing the estimation and the experimental measurement.

Accurate prediction temperature variation of electric drive transmission (EDT) can effectively monitor its abnormal temperature rise that may occur under high speed and heavy load working conditions, so as to ensure the vehicles’ safe operation. In this paper, combined with real temperature and input/output characteristic data collected from EDT test platform under different working conditions, a spatio-temporal relationship dynamic graph convolution neural network based on least square method (OLS-DRGCN) for temperature prediction is proposed. Firstly, OLS is used to estimate the EDT’s internal temperature based on partial sensor information as the input of OLS-DRGCN. Secondly, the spatial dependence relationship of each temperature node is dynamically learned through node embedding and the dynamic thermal network topology of EDT is constructed. Meanwhile, the timing rule of each temperature node is obtained through the gated recurrent unit.