Search Results

A Machine Learning-Genetic Algorithm (ML-GA) approach was developed to virtually discover optimum designs using training data generated from multi-dimensional simulations. Machine learning (ML) presents a pathway to transform complex physical processes that occur in a combustion engine into compact informational processes. In the present work, a total of over 2000 sector-mesh computational fluid dynamics (CFD) simulations of a heavy-duty engine were performed. These were run concurrently on a supercomputer to reduce overall turnaround time. The engine being optimized was run on a low-octane (RON70) gasoline fuel under partially premixed compression ignition (PPCI) mode. A total of nine input parameters were varied, and the CFD simulation cases were generated by randomly sampling points from this nine-dimensional input space. These input parameters included fuel injection strategy, injector design, and various in-cylinder flow and thermodynamic conditions at intake valve closure (IVC).

To improve the braking energy recovery rate of pure electric garbage removal vehicles and ensure the braking effect of garbage removal vehicles, a strategy using particle swarm algorithm to optimize the regenerative braking fuzzy control of garbage removal vehicles is proposed. A multi-section front and rear wheel braking force distribution curve is designed considering the braking effect and braking energy recovery. A hierarchical regenerative braking fuzzy control strategy is established based on the braking force and braking intensity required by the vehicle. The first layer is based on the braking force required by the vehicle, based on the front and rear axle braking force distribution plan, and uses fuzzy controllers.

In the foreseeable future, the growing energy crisis and environmental pollution problem pose severe challenges to the automobile powertrains and exhaust systems. However, conventional optimization methods, including multi-dimensional computational fluid dynamics model and bench experiments, are very time-consuming or expensive. Adding the application of data-driven models to engine research and development has the potential to reduce computational costs or the number of in-depth experiments. This purpose of this study was to compare the performance of widely used artificial neural network (ANN) and random forest (RF) model for predicting the fuel consumption and engine-out emissions of a calibrated spark ignition (SI) engine for any given condition.

Gasoline compression ignition (GCI) has been shown to offer benefits in the NOx-soot tradeoff over conventional diesel combustion while still achieving high fuel efficiency. However, due to gasoline’s low reactivity, it is challenging for GCI to attain robust ignition and stable combustion under cold operating conditions. Building on previous work to evaluate glow plug-assisted GCI combustion at cold idle, this work evaluates the use of a spark plug to assist combustion. The closed-cycle 3-D CFD model was validated against GCI test results at a compression ratio of 17.3 during extended cold idle operation under laboratory-controlled conditions. A market representative, ethanol-free, gasoline (RON92, E0) was used in both the experiment and the numerical analysis. Spark-assisted simulations were performed by incorporating an ignition model with the spark energy required for stable combustion at cold start.

Due to the poor ability of millimeter wave radar in recognizing distant static objects, target loss and incomplete information will occur when it recognizes the static target in front, thus increasing the false alarm rate and missing alarm rate of the radar-dependent driving assistant system, which will reduce the driving safety and the acceptability of the assistant system. Aiming at the radar's poor ability to recognize static targets, this paper uses a model based on machine learning algorithm to recognize and track targets. The radar signals are collected and processed in different conditions, and the results show that the radar has a poor recognition effect when the distance is more than 100 meters and the speed is more than 19m/s.

Mcity at the University of Michigan in Ann Arbor provides a realistic off-roadway environment in which to test vehicles and drivers in complex traffic situations. It is intended for testing of various levels of vehicle automation, from advanced driver assistance systems (ADAS) to fully self-driving vehicles. In a recent human factors study of interfaces for teen drivers, we performed parallel experiments in a driving simulator and Mcity. We implemented driving scenarios of moderate complexity (e.g., passing a vehicle parked on the right side of the road just before a pedestrian crosswalk, with the parked vehicle partially blocking the view of the crosswalk) in both the simulator and at Mcity.

The goal of this project was to investigate how to make driver distraction state classification more efficient by applying selected machine learning techniques to existing datasets. The data set used in this project included both overt driver behavior measures (e.g., lane keeping and headway measures) and indices of internal cognitive processes (e.g., driver situation awareness responses) collected under four distraction conditions, including no-distraction, visual-manual distraction only, cognitive distraction only, and dual distraction conditions. The baseline classification method that we employed was a support vector machine (SVM) to first identify driver states of visual-manual distraction and then to identify any cognitive-related distraction among the visual-manual distraction cases and other non-visual manual distraction cases.