Skip to main content
Fachgebiete
Automobil + Motoren Bauwesen + Immobilien Business IT + Informatik Elektrotechnik + Elektronik Energie + Nachhaltigkeit Finance + Banking Management + Führung Marketing + Vertrieb Maschinenbau + Werkstoffe Versicherung + Risiko
DE
EN
Bücher
Zeitschriften
Themenseiten
Organisationspsychologie Projektmanagement Marketing Smart Manufacturing
Weitere Formate
Podcasts Webinare Technik Webinare Wirtschaft Kongresse Veranstaltungskalender Awards
Jetzt Einzelzugang starten
Zugang für Unternehmen
Referenzkunden
SLX-Digitalkonferenz: Zukunftswerkstatt 2024

Mehr Umsatz mit Top-Kontakten

Am 7. Mai erfahren Sie, wie Vertriebsteams Leadgenerierung, Leadnurturing und Leadstrategien effizient steuern können.
Erweiterte Suche
Anmelden
nach oben

2018 | Buch

Introduction Kapitel lesen Erstes Kapitel lesen

Enhanced Machine Learning and Data Mining Methods for Analysing Large Hybrid Electric Vehicle Fleets based on Load Spectrum Data

verfasst von: Philipp Bergmeir

Verlag: Springer Fachmedien Wiesbaden

Buchreihe : Wissenschaftliche Reihe Fahrzeugtechnik Universität Stuttgart

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten
insite
SUCHEN

Über dieses Buch

Philipp Bergmeir works on the development and enhancement of data mining and machine learning methods with the aim of analysing automatically huge amounts of load spectrum data that are recorded for large hybrid electric vehicle fleets. In particular, he presents new approaches for uncovering and describing stress and usage patterns that are related to failures of selected components of the hybrid power-train.

Inhaltsverzeichnis

Frontmatter
Chapter 1. Introduction
Abstract
Nowadays, modern vehicles, like hybrid electric vehicles (HEV), are equipped with many electronic control units (ECUs) such as an engine control unit or a battery management system (BMS). In general, among the main tasks of such a unit are monitoring the state of a component, controlling various of its functions, and protecting it from abnormal use. For this purpose, an ECU reads data that are measured by sensors and calculates control values and statistics that describe the component’s load, e.g., load spectrum data [61], based on the sensor input.
Philipp Bergmeir
Chapter 2. Data foundation
Abstract
The success of each Data Mining task strongly depends on the quality of the data that are used. Thus, assuring a high data quality before performing any data analysis is a crucial, maybe the most important step of any data analysis project. This is also pointed out by the popular saying “garbage in, garbage out” [10].
Philipp Bergmeir
Chapter 3. Classifying component failures of a vehicle fleet
Abstract
In this chapter, the applicability of several state-of-the-art classification algorithms such as random forests and support vector machines are studied for the purpose of distinguishing non-faulty HEV from those suffering from a failure of a particular component of the hybrid power-train, when these algorithms are fed with load spectrum data. Furthermore, it is analysed whether these classifiers can be combined with feature selection approaches to not only improve the classification performance of the models, but also to select a small set of failure related features.
Philipp Bergmeir
Chapter 4. Visualizing different kinds of vehicle stress and usage
Abstract
At the end of the previous chapter a motivation is given, why an algorithm is strongly needed that is able to identify whether two faulty vehicles are likely to suffer from the same type of failure or not. Thereby, it is only known that the same component of these two cars has failed. Moreover, the method should draw its conclusion exclusively on the basis of the load spectrum data of these vehicles.
Philipp Bergmeir
Chapter 5. Identifying usage and stress patterns in a vehicle fleet
Abstract
In Chapter 3, methods have been proposed that facilitate a load spectrum based distinction between vehicles with and those without a failure of a hybrid component, whereas algorithms that allow a visual detection of structure such as clusters in the data have been discussed in Chapter 4. However, a common disadvantage of the approaches, which performed best on the studied datasets, is that they are all “black box” models, i.e., they do not allow to gain interpretable insights into the analytical relationship between the data and the obtained results. More precisely, it remains unknown which patterns in the data provoke the classifier rf Gini to assign a certain label to an instance on the one hand, and which ones induce particular objects to form a cluster in the low-dimensional maps, produced by method RF-t-SNE, while others do not, on the other hand.
Philipp Bergmeir
Chapter 6. Conclusion
Abstract
This thesis addressed the problem of analysing a huge amount of a load spectrum data, i.e., a special kind of automotive data that are recorded and computed on-board in modern vehicles such as HEVs. The aim has been manifold, where the main goal has been to determine usage and stress patterns that are related to failures of selected components of the hybrid power-train, like the hybrid car battery. The identified patterns can help the engineers to find out the reasons for component failures and, thus, to improve the dimensioning as well as the reliability of future versions of these vehicle parts.
Philipp Bergmeir
Chapter 7. Outlook
Abstract
At the end of this thesis, possible enhancements of the proposed approaches as well as future research directions are discussed briefly.
Philipp Bergmeir
Backmatter
Metadaten
Titel
Enhanced Machine Learning and Data Mining Methods for Analysing Large Hybrid Electric Vehicle Fleets based on Load Spectrum Data
verfasst von
Philipp Bergmeir
Copyright-Jahr
2018
Electronic ISBN
978-3-658-20367-2
Print ISBN
978-3-658-20366-5
DOI
https://doi.org/10.1007/978-3-658-20367-2

    Premium Partner

    Bildnachweise