Projektarbeten
Länka till denna samling:
Browse
Senast publicerade
- PostMachine Learning-Based Optimization for Battery Pack Cooling(2024) Abukar, Abubakar; John, Pedro; Boudagh, Francisco; Verde, Salvatore; Wendel, Gabriel; Chalmers tekniska högskola // Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime Sciences; Vdovin, Alexey; Vdovin, Alexey; Vivek, Anthony; Koutsimanis, Dimitrios; Alatalo, ViktorThe thermal management of electric vehicle (EV) batteries is a critical factor in safety, performance and durability. Traditional design methods for cooling channels rely on manual adjustments, trial and error processes and intensive topology optimizations leading to time con suming approaches and significant limitations in efficiency. This study introduces an innovative framework that combines machine learning (ML) and computational fluid dynamics (CFD) to optimize cooling channel designs that circumvent the challenges faced by traditional methods. These improvements are achieved by using two different components: The first component is a surrogate model. It is a machine learning model trained on a large dataset produced through CFD simulations using Star-CCM+. This model significantly reduces computational costs and time by predicting pressure drops and temperature distributions based on input geometries and system parameters. The second component is a genetic algorithm for geometry optimization. This component generates an optimal geometry by balancing pressure drop and an effective thermal regulation by using a Non-Dominated Sorting Genetic Algorithm (NSGA-II). This algorithm creates a number of random solutions and iteratively improves on them to find a Pareto front, which represents the optimal trade-offs between competing objectives: minimizing pressure drop and maximizing thermal regulation. The role of the surrogate model is to provide instant feedback on potential solutions throughout the algorithm, which is vital as the algorithm itself is inherently time consuming. By combining these two components, the framework accelerates the design process ensuring the creation of advanced cooling solutions. The results demonstrate the potential to reduce computational time while achieving superior performance. This work provides a framework for future advancements in the thermal management of EV batteries, highlighting the importance of combining CFD with ML in modern engineering solutions.
- PostDrawbar Eye Identification and Guiding(2024) Garg, Mahin; Winbo, Andreas; Liu, Biying; Qui, Shiyi; Renberg, Felix; Chalmers tekniska högskola // Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime Sciences; Vdovin, Alexey; Von Corswant, Fredrik; Johansson, TobiasThe purpose of this research project is to study if a machine learning algorithm can utilize a video feed from a camera mounted on the back of a truck to detect a drawbar eye under varying environmental, instrumental and lightning conditions. In order to achieve this, lab based and real world testing data was collected from a setup which simulated connecting a truck and a trailer. With these setup cases, different video feeds were collected and later, images were extracted from this feed, which were labeled for the machine learning algorithm. In order to train the algorithm, YOLOv8n (You Only Look Once Version 8) was used, which is a real time object detection algorithm to identify objects in an image. This was used to classify the drawbar eye and its position in real time. In order to better guide the driver, distance between camera and drawbar eye was calculated using the pinhole camera principle, and here the angle of the camera was introduced. From the calculation, we had the horizontal distance (X), vertical distance (Y) and distance in depth (Z). Two additional tests were done in order to verify the accuracy of distance and height calculation, and take the distortions of a fisheye camera into account while calculating the said distances. The final detection accuracy (ability to detect position of drawbar eye with 50-95 percent accuracy of the labeled position) of the model came at around 75.4 percent.
- PostWheelset maintenance and monitoring: Current practices and strategies(2024) Cervenka, Jakub; Bergkvist, Herman; Wirdheim, Eskil; Chalmers tekniska högskola // Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime Sciences; Ekberg, Anders; Ekberg, AndersRailway wheelsets are a central component of railway vehicles. Maximizing the lifespan of wheelsets requires proper health monitoring and maintenance strategies. From a background in deterioration phenomena and the operational conditions of railway vehicles, this project gives an overview of maintenance and monitoring within the railway industry. This has been done through interviews with and visits to multiple industry actors. Practices within monitoring, maintenance, and possible business plans are highly variable, depending on operating conditions, involved actors, and the size and uniformity of the fleet. Potential benefits could be achieved through more cooperation and data sharing between actors, however this brings new challenges.
- PostBrake load characterization of heavy-duty vehicles(2024) Achmad Munthahar, Sayid; Dhamane, Bhumika; Randby, David; Suresh, Nandu; Wurzinger, Jakob; Chalmers tekniska högskola // Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime Sciences; Sjöblom, Jonas; Srivastava, Suraj; Petersson, MartinIn the fast-changing automotive world, more vehicles have zero tail-pipe emissions due to electrification. Global and local air quality has increased in recent times due to the implementation of legislation regarding tail-pipe emissions. Regulators are now looking at other sources of emissions, like particulate matter from brakes and tires, which have severe health effects. The particulate matter can enter the airways and if it is small enough, it can travel down to the lungs and even further into the bloodstream. The particle sizes can be divided into PM10, PM2.5, and Ultra-fine particles. Legislation for PM10 will be implemented regarding brake wear in the new Euro 7 legislation for light-duty vehicles in 2025 and something similar to this is expected for heavy-duty vehicles. The stakeholder for this project is Volvo Group, which has been involved during the whole project. This project mainly focuses on the emissions caused by disc brakes on Volvo’s Internal Combustion Engine (ICE) and Battery Electric Vehicle (BEV) heavy-duty vehicles. Field-test datasets from several ICE and BEV trucks, along with Volvo’s brake test rig data are used to estimate brake wear based on brake temperature and braking energy. Data analysis was performed on the dataset and brake wear estimation was done to characterize the braking behaviors of ICE and BEV trucks and determine which affects the brake wear the most. An in-depth analysis was performed by doing a multivariate analysis. The in-depth analysis resulted in the identification of four different clusters that were characterized by their unique driving behavior. From this exploration, it becomes evident that several factors influence brake wear, with temperature emerging as a prominent indicator linked to driving behavior.
- PostUsing machine learning to estimate road-user kinematics from video data(2024) Fang, Luhan; Malm, Oskar; Wu, Yahui; Xiao, Tianshuo; Zhao, Minxiang; Chalmers tekniska högskola // Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime Sciences; Sedarsky, David; Dozza, Marco; Rajendra Pai, Rahul; Rasch, AlexanderEach year, there are over one million people who sustain fatal injuries in traffic-related crashes, with vulnerable road users, often abbreviated as VRUs, being involved in more than half of the crashes. In the context of road safety, VRUs are mainly pedestrians, cyclists, motorcyclists, and e-scooterists. To mitigate the crashes, it is essential to understand the causation mechanisms. Naturalistic data has been recognised as a good tool to understand the trafficant’s behaviour and address safety concerns within the field of traffic safety research. Traditionally, critical events are identified from naturalistic data using the kinematic information from the sensors onboard the vehicles. Video footage for the trip corresponding to the critical events is then used to validate and annotate the events. While this is a reliable method when it comes to identification of crashes, near-crashes may not display any anomalies in the sensor readings thereby going unidentified. These near-crash events would be visible in the video footage, but the manual identification of these by watching videos is not feasible because the amount of videos is too large for the human eyes. Therefore, developing tools that can identify and estimate the position of different road users using the video footage is essential and will enable automation of process of identifying critical events. This report describes such models and also delves into the application of machine learning to allow identify the severity of imminent critical interactions among road users in the future. This project investigates how models can be developed to estimate and predict the position and kinematics of various road users from video data from a camera mounted on an e-scooter. The initial generation of bounding boxes and categories for road users utilized You Only Look Once (YOLOv7) algorithms. The detection for cyclists was achieved by a simple rule-based model calculating the overlap area between the pedestrian and bicycle detected by YOLOv7. The e-scooterists detection model was implemented by combining YOLOv7 and MobileNetV2 models. Different machine learning models were trained to estimate distance for the four different road users: pedestrians, cyclists, e-scooterists, and cars separately using LiDAR and GPS data as the position ground truth. The input for these models was derived from bounding box data extracted from videos. Furthermore, a DBSCAN-based noise remover was used to remove the outlier point of the distance estimation model to filter out points with excessive errors. Finally, a Rauch-Tung-Striebel smoother was applied to the output of the noise remover to improve the distance estimation accuracy and generate both the relative position and velocity of the target road user. It was concluded that the object detection model could achieve an accuracy over 90%, and the distance estimation achieved the highest accuracy when using polar coordinates for all road users, compared with using the cartesian system. The highest R2 score for distance estimation was obtained with a k-nearest neighbors regression model (with n = 2) using the pixel position in x and y direction of center point of the bottom of bounding box, height, and width of the bounding box as input. As a consequence, the e-scooterist model achieved a R2 score of 0.978, while the cyclist and car models attained commendable scores of 0.92 and 0.96 respectively. This means that the distances predicted from the model are highly accurate. These models can now be used to detect critical interactions among road users using the naturalistic data collected using e-scooters.