Monitoring and comparing various approaches for short-term forecasting of urban traffic parameters and simulation using GIS: (Case study of the city of London)
The main objective of this research is to compare different methods for short-term forecasting of urban traffic parameters, as well as simulation of traffic parameters in the MATLAB environment and optimal selection of their effective parameters with a Geographic Information System (GIS) as a supplement to a transportation information system. To that end, three distinct short-term traffic parameter forecasting algorithms, traditional polynomials (TP), genetic basis traditional polynomials (GBT), and neural networks (NN), were used to predict traffic parameters using two error reduction strategies. In addition, to control future traffic, urban traffic flow and velocity parameters were simulated. Due to a lack of regular traffic data in Iran, the research data for this study was drawn from data from 2012 to 2014 in London, with similar traffic patterns during the week. The routes investigated total 15.84 km and are known as LM561-LM563-LM557-LM555. Training, validation, and reference data were obtained in 2012, 2013, and 2014, respectively. Overall, the findings revealed that the TP approach failed to forecast traffic flow and speed characteristics, but the GBT and NN methods were effective. Furthermore, the quantitative findings of the study routes in terms of root mean square error revealed that the three techniques of TP, GBT, and NN for traffic flow parameters were 13.91, 0.78, and 0.22, respectively, and for the speed parameter, 5.20, 0.78, and 0.19. In other words, the accuracy of the traffic flow parameter in GBT and NN is about 18 and 63 times better than the TP technique, while the accuracy of the speed parameter is approximately 7 and 27 times better than the TP method, respectively.
- حق عضویت دریافتی صرف حمایت از نشریات عضو و نگهداری، تکمیل و توسعه مگیران میشود.
- پرداخت حق اشتراک و دانلود مقالات اجازه بازنشر آن در سایر رسانههای چاپی و دیجیتال را به کاربر نمیدهد.