University of Twente Student Theses

Login

Improving and expanding the algorithms of Gouden Regelen

Hazenberg, L.W. (2022) Improving and expanding the algorithms of Gouden Regelen.

[img] PDF
2MB
Abstract:In this study, algorithms are designed and evaluated to improve Gouden Regelen. Gouden Regelen is an application for traffic control systems. It uses real life data from detection loops to optimize the maximum green times and the order in which signal groups get green. In this study, this application is optimized by adding algorithms that are made to better incorporate cyclists and long vehicles (or trucks) in the application. For the algorithm that incorporates cyclists into Gouden Regelen was found that already research for this was conducted. Therefore, the algorithm was based on this research. The algorithm designed exists of two parts, namely the part that determines the number of cyclists that pass the distant loop detector per occupation and the part that adds cyclists to the queue. The number of cyclists that pass the detector loop is determined using the follow-up time and the occupation time. The follow-up time is mainly used to determine the direction the cyclist(s) is going and the occupation time is used to determine the number of cyclists passing the detector. This number is then added to the number of cyclists in the queue. The algorithm that is made to incorporate long vehicles, classifies whether a vehicle passing the distant loop detector is a passenger car or freight traffic. Depending on the situation, the algorithm uses different methods for this. When the queue fits within the detection field, the algorithm uses the method that bases the expected occupation time on the expected braking distance. By comparing the expected occupation time with the actual occupation time, the vehicle classification can be determined. When a vehicle is standing still on the distant loop detector, the vehicle classification is determined by comparing the distance between the end of the queue and the distant loop detector with the (expected) length of a passenger car. If the distance is much bigger, the vehicle is classified as freight traffic. Otherwise, it is classified as a passenger car. The last situation, is the situation where the queue grew outside the detection field. In that case, the classification is determined by comparing the occupation time of the current vehicle with the occupation time of the previous vehicle. The accuracy of the algorithms is tested in order to evaluate their performance. This is done using multiple tests, such as the base scenario, a scenario where the traffic volume is increased, etc. The results from these tests are that the accuracy is high for almost all cases. However, some things stand out. Namely that the algorithm is not that good at recognizing when multiple cyclists pass the detection loop at once. This is mainly due to the fact that the occupation time is not different when only one cyclist passes the detection loop. The algorithm that classifies motor vehicles has its deficiencies as well. Firstly, from the results could be seen that especially the method which uses the braking distance to classify vehicles performs well. However, the other methods could be improved to increase the accuracy. Next to that, the algorithm is designed to only make a distinction between a truck and a passenger car. While it could be desired to make more distinctions in order to optimise Gouden Regelen even more. Besides the accuracy, also the effects on the traffic flow, traffic safety and credibility of traffic control systems equipped with Gouden Regelen are determined. This is done by comparing the results of four different scenarios (no algorithms, both algorithms, only the cyclist algorithm and only the freight algorithm) with each other. As indicators of the effects, the waiting time and the chance that vehicles have to stop multiple times during a cycle are used. It was found that there were no significant differences between the different scenarios. Meaning that the algorithms do neither improve nor deteriorate the traffic flow, traffic safety and credibility. From the results can be concluded that the algorithms perform in most cases like it is supposed to do. However, due to some of the deficiencies of the algorithms and the results which indicate the effects of implementing the algorithms, it is advised to conduct some more research before the algorithms are implemented. Besides that, the algorithms are only tested using simulations, in order to find all results it is desired to also test them in real life.
Item Type:Essay (Bachelor)
Faculty:ET: Engineering Technology
Programme:Civil Engineering BSc (56952)
Link to this item:https://purl.utwente.nl/essays/93594
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page