Developing Passenger Car Equivalent by Modeling Average Travel Speed Using Artificial Neural Network
| dc.contributor.advisor | Bikila, Teklu (PhD) | |
| dc.contributor.author | Tadiyos, Marie | |
| dc.date.accessioned | 2021-10-19T08:41:13Z | |
| dc.date.accessioned | 2023-11-11T12:53:43Z | |
| dc.date.available | 2021-10-19T08:41:13Z | |
| dc.date.available | 2023-11-11T12:53:43Z | |
| dc.date.issued | 2021-07 | |
| dc.description.abstract | In order to develop the model for studying the effect of traffic volume and composition variations observed in Addis Ababa on PCE, mainline road midblock sections in Addis Ababa ring road with uninterrupted flow in one direction was taken as a case study. A multistage sampling technique was adapted to collect the data from five sections. The data was taken using two video cameras for recording traffic flow; the cameras were placed in entry and exit location of the section. The data used are the traffic flow and average traffic speed of every vehicle type for a 5-minute time interval that meets minimum number of vehicles to be observed as a result a total of 675 datasets are extracted and calculated from the five sections. In order to achieve the desired objectives, average speed was modeled using artificial neural network first and PCE is estimated using Equation 2-1. For measuring the accuracy of the model result the study uses coefficient of correlation (R2 ). The model is developed using MATLAP, the model divides datasets in to three groups which are training used 70% of the datasets, tasting and validation each uses 15% of the datasets. The developed model use Levenberg– Marquardt as training and tan-sigmoid as activation function because they provide the best generalization from other training and activation functions tried. The model provides 94% and above R2 value for training, tasting, validation, and all datasets. The analysis of results from the speed model indicates; increasing the volume of the traffic stream from 300 to 2400veh/hr. decrease speed from 76 to 26.8km/hr for PC, 75.6 to 28.6km/hr for pickup and LC, 74.2 to 28km/hr. for minibus, 58.9 to 27.3km/hr for bu s, and 48.7 to 24.1km/hr for truck and PCE decreases from 1.54 to 1.42 for pickup and LC, 1.65 to 1.53 for minibus, 4.02 to 3.05 for bus, 4.16 to 2.96 for truck. Changing the proportion of vehicle types by keeping the volume constant shows a decrease in traffic stream speed and increase in PCE for all vehicle types except minibus but the effect is pronounced on bus and truck. For example, increasing the percentage of bus from 0 to 130veh/hr. decreases the speed from 57.8 to 48km/hr. for PC, 59 to 47.78km/hr. for pickup and LC, 58.5 to 45.7km/hr. for minibus, 60 to 40.6km/hr. for bus, 59.6 to 37km/hr. for truck and PCE increases from 1.52 to 1.55 for pickup and LC, 1.64 to1.69 for minibus, 3.4 to 3.69 for bus, 2.94 to 3.42 for truck. Finally, it’s concluded that PCE is different for different traffic and volume scenarios observed in Addis Ababa. | en_US |
| dc.identifier.uri | http://etd.aau.edu.et/handle/12345678/28247 | |
| dc.language.iso | en_US | en_US |
| dc.publisher | Addis Ababa University | en_US |
| dc.subject | passenger car equivalence | en_US |
| dc.subject | average travel speed | en_US |
| dc.subject | artificial neural network | en_US |
| dc.subject | mixed traffic | en_US |
| dc.title | Developing Passenger Car Equivalent by Modeling Average Travel Speed Using Artificial Neural Network | en_US |
| dc.type | Thesis | en_US |