Evaluation of Head Toss Based on Sensor Data Collected from a Car on a Four-Poster Rig: A study in how to quantify, measure and replicate a subjective feeling in car ride in an objective manner using a four-poster rig

Typ
Examensarbete för masterexamen
Program
Publicerad
2021
Författare
Brokelind, Hugo
Thörneby, Kasper
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
Comfort development in vehicles has been a factor for a long time. Recent research examines urban driving at low speed which causes jerking motion of the car that can be unpleasant for the passengers’ necks and heads. These undesirable comfort disruptions are known as head toss. The report aims to find an objective way to measure and compare the occurrence and harshness of the subjective feeling of head toss between different cars and chassis setups. Furthermore, the aim is to measure the head toss on a 4-poster rig, commonly known as a shake rig. This was done through collection of chassis and head acceleration data and the rating of each head toss on a scale of 1-9. The relationship between head toss occurrence and chassis accelerations would show to be complex and non-linear. A design of models and tools that were able to accurately predict the occurrences and rating of head toss was therefore executed. The models that performed best was neural networks and ensemble learning models. Using these models, a program was developed which could predict the occurrence of head toss and its harshness accurately. The final result of this project thus became an experiment set up to execute a rig program in order to collect data, and a program that evaluated the collected data to display an objective evaluation of head toss.
Beskrivning
Ämne/nyckelord
Head toss , subjective measurement , car comfort , objective testing , four-poster rig , non-linear model , ensemble learning , neural network , system identification , sensor data , time series
Citation
Arkitekt (konstruktör)
Geografisk plats
Byggnad (typ)
Byggår
Modelltyp
Skala
Teknik / material
Index