Search Results

High-fidelity finite element (FE) tire-snow interaction models have the advantage of better understanding the physics of the tire-snow system. They can be used to develop semi-analytical models for vehicle design as well as to design and interpret field test results. For off-terrain conditions, there is a high level of uncertainties inherent in the system. The FE models are computationally intensive even when uncertainties of the system are not taken into account. On the other hand, field tests of tire-snow interaction are very costly. In this paper, dynamic metamodels are established to interpret interaction forces from FE simulation and to predict those forces by using part of the FE data as training data and part as validation data. Two metamodels are built based upon the Krieging principle: one has principal component analysis (PCA) taken into account and the other does not.

An application of variation simulation for predicting vehicle interior driveline vibration is presented. The model, based on a “Monte Carlo”-style approach, predicts the noise, vibration and harshness (NVH) response of the vehicle driveline based on distributions of imbalance and runout derived from manufacturing production variation (the forcing function) and the vehicle's sensitivity to the forcing function. The model is used to illustrate the change in vehicle interior vibration that results when changes are made to production variation for runout and imbalance of driveline components, and how those same changes result in different responses based on vehicle sensitivity.

Selecting the right transportation platform is challenging, whether it is at a personal level or at an organizational level. In settings where predominantly the functional aspects rule the decision making process, defining the mobility of a vehicle is critical for comparing different offerings and making acquisition decisions. With the advent of intelligent vehicles, exhibiting partial to full autonomy, this challenge is exacerbated. The same vehicle may traverse independently and with greater tolerance for acceleration than human occupied vehicles, while, at the same time struggle with obstacle avoidance. The problem presents itself at the individual vehicle sensing level and also at the vehicle/fleet level. At the sensing and information level, one can be looking at issues of latency, bandwidth and optimal information fusion from multiple sources including privileged sensing. At the overall vehicle level, one focuses more on the ability to complete missions.

Durability of automotive structures is a primary engineering consideration that is evaluated during a vehicle's design and development. In addition, it is a basic expectation of consumers, who demand ever-increasing levels of quality and dependability. Automakers have developed corporate requirements for vehicle system durability which must be met before a products is delivered to the customer. To provide early predictions of vehicle durability, prior to the construction and testing of prototypes, it is necessary to predict the forces generated in the vehicle structure due to road inputs. This paper describes an application of the “virtual proving ground” approach for vehicle durability load prediction for a vehicle on proving ground road surfaces. Correlation of the results of such a series of simulations will be described, and the modeling and simulation requirements to provide accurate simulations will be presented.

Countless articles and publications3,4,5 have documented and proven the efficacy, benefits and value of operating within a lean system. Furthermore, there exists common agreement amongst leading organizations successfully implementing a lean system that in order to do so it must take into consideration the entire enterprise, that is, from supplier to customer and everything in between6. One of the core issues this paper addresses is when the optimal time is to train and educate the people who currently have, or will have, influence over the ‘enterprise’.

Motivated by experiences in the Formula SAE® competition, an engine control unit (ECU) was designed, developed and tested at Oakland University. A systems approach was taken in which the designs of the electronic architecture and software were driven by the mechanical requirements and operational needs of the engine, and by the need for dynamometer testing and tuning functions. An ECU, powered by a 68HC12 microcontroller was developed, including a four-layer circuit board designed for EMC. A GUI was written with Visual C++® for communication with a personal computer (PC). The ECU was systematically tested with an engine simulator, a 2L Ford engine and a 600cc Honda engine, and finally in Oakland's 2004 FSAE vehicle.

Controller Area Network (CAN), the de facto standard for in-vehicle networks, has insufficient security features and thus is inherently vulnerable to various attacks. To protect CAN bus from attacks, intrusion detection systems (IDSs) based on advanced deep learning methods, such as Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN), have been proposed to detect intrusions. However, those models generally introduce high latency, require considerable memory space, and often result in high energy consumption. To accelerate intrusion detection and also reduce memory requests, we exploit the use of Binarized Neural Network (BNN) and hardware-based acceleration for intrusion detection in in-vehicle networks. As BNN uses binary values for activations and weights rather than full precision values, it usually results in faster computation, smaller memory cost, and lower energy consumption than full precision models.

Machine learning methods, such as decision trees and deep neural networks, are becoming increasingly important and useful for data analysis in various scientific fields including dynamics and control, signal processing, pattern recognition, fluid mechanics, and chemical synthesis, etc. For future engine design and performance optimization, there is an urgent need for a robust predictive model which could capture the major combustion properties such as autoignition and flame propagation of multicomponent fuels under a wide range of engine operating conditions, without massive experimental measurement or computational efforts. It will be shown that these long-held limitations and challenges related to complex fuel combustion and engine research could be readily solved by implementing machine learning methods.

The coefficient of friction between surfaces is an important criterion for predicting metal behavior during sheet metal stamping processes. This research introduces an innovative technique to find the coefficient of friction on a lubricated aluminum sheet metal surface by simulating the industrial manufacturing stamping process while using 3D digital image correlation (3D-DIC) to track the deformation. During testing, a 5000 series aluminum specimen is placed inside a Stretch-Bend-Draw Simulator (SBDS), which operates with a tensile machine to create a stretch and bend effect. The friction coefficient at the contact point between an alloy sheet metal and a punch tool is calculated using an empirical equation previously developed. In order to solve for the unknown friction coefficient, the load force and the drawback force are both required. The tensile machine software only provides the load force applied on the specimen by the load cell.

The need for transient thermal simulations in vehicle packaging studies has grown rapidly in recent years. To date, the computational costs associated with the transient simulation of 3D conjugate heat transfer phenomena has prohibited the widespread use of full vehicle transient simulations. This paper presents results from a recent study that explored a method to circumvent the computational costs associated with long transient conjugate heat transfer simulations. The proposed method first segregates the thermal structural and fluid physics domains to take advantage of time scale differences. The two domains are then re-coupled to calculate a series of steady state conjugate heat transfer simulations at various vehicle speeds. The local convection terms are then used to construct a set of surrogate models dependent on vehicle speed, that predict the local heat transfer coefficients and the local near wall fluid temperatures.

Springback is a common phenomenon in automotive manufacturing processes, caused by the elastic recovery of the internal stresses during unloading. A thorough understanding of springback is essential for the design of tools used in sheet metal forming operations. A DBS (Draw-bead Simulator) has been used to simulate the forming process for two different sheet metals: aluminum and steel. Two levels of pulling force and two die radii have been enforced to the experimental process to get different springback. Also, the Digital Image Correlation (DIC) system has been adopted to capture the sheet contour and measure the amount of side-wall-curl (sheet springback) after deformation. This paper presents the influence of the material properties, force, and die radius on the deformation and springback after forming. A thorough understanding of this phenomenon is essential, seeing that any curvature in the part wall can affect quality and sustainability.

The development of efficient algorithms has been the focus of automobile engineers since self-driving cars become popular. This is due to the potential benefits we can get from self-driving cars and how they can improve safety on our roads. Despite the good promises that come with self-driving cars development, it is way behind being a perfect system because of the complexity of our environment. A self-driving car must understand its environment before it makes decisions on how to navigate, and this might be difficult because the changes in our environment is non-deterministic. With the development of computer vision, some key problems in intelligent driving have been active research areas. The advances made in the field of artificial intelligence made it possible for researchers to try solving these problems with artificial intelligence. Lane detection and tracking is one of the critical problems that need to be effectively implemented.

In the automotive industry, a Malfunction Indicator Light (MIL) is commonly employed to signify a failure or error in a vehicle system. To identify the root cause that has triggered a particular fault, a technician or engineer will typically run diagnostic tests and analyses. This type of analysis can take a significant amount of time and resources at the cost of customer satisfaction and perceived quality. Predicting an impending error allows for preventative measures or actions which might mitigate the effects of the error. Modern vehicles generate data in the form of sensor readings accessible through the vehicle’s Controller Area Network (CAN). Such data is generally too extensive to aid in analysis and decision making unless machine learning-based methods are used. This paper proposes a method utilizing a recurrent neural network (RNN) to predict an impending fault before it occurs through the use of CAN data.

Engineering design-decisions often involve many attributes which can differ in the levels of their importance to the decision maker (DM), while also exhibiting complex statistical relationships. Learning a decision-making policy which accurately represents the DM’s actions has long been the goal of decision analysts. To circumvent elicitation and modeling issues, this process is often oversimplified in how many factors are considered and how complicated the relationships considered between them are. Without these simplifications, the classical lottery-based preference elicitation is overly expensive, and the responses degrade rapidly in quality as the number of attributes increase. In this paper, we investigate the ability of deep preference machine learning to model high-dimensional decision-making policies utilizing rankings elicited from decision makers.

This paper presents a nonlinear model predictive controller (NMPC) coupled with a pre-trained reinforcement learning (RL) model that can be applied to lateral control tasks for autonomous vehicles. The past few years have seen opulent breakthroughs in applying reinforcement learning to quadruped, biped, and robot arm motion control; while these research extend the frontiers of artificial intelligence and robotics, control policy governed by reinforcement learning along can hardly guarantee the safety and robustness imperative to the technologies in our daily life because the amount of experience needed to train a RL model oftentimes makes training in simulation the only candidate, which leads to the long-standing sim-to-real gap problem–This forbids the autonomous vehicles to harness RL’s ability to optimize a driving policy by searching in a high-dimensional state space.