Research on intervals in traffic flow on multilane roadways
Students Name: Matsko Vladyslav Yuriiovych
Qualification Level: magister
Speciality: Traffic Control and Organization
Institute: Institute of Mechanical Engineering and Transport
Mode of Study: full
Academic Year: 2022-2023 н.р.
Language of Defence: ukrainian
Abstract: The already existing results of researches of certain indicators of a transport stream on VDM of the city are analyzed. Thus, the actual distribution of the values of time intervals at different traffic intensities shows that with decreasing intensity the time intervals increase. Scientists also analyzed the distribution of time intervals between vehicles with different composition of the traffic flow under the influence of traffic intensity. Silyanov derived the dependence of time intervals between cars at different speeds. It is seen that with increasing intervals between cars, the speed increases. However, additional research in this area will provide a more detailed description of certain dependencies of traffic flow indicators. Thus, at the first stage the intervals of traffic between vehicles on the multi-lane carriageway were investigated. Studies of traffic intervals between vehicles were conducted on the main street of regulated traffic with a dividing lane (Naukova Street, Lviv). On this section of the street there are three lanes in each direction, the pavement is asphalt concrete, in good condition. The width of the lanes is 3.75 m. The research was conducted at a distance of 75 m at the intersection with Trolleybusna Street. The research was conducted using a video camera that covered both directions. The corresponding values of traffic flow indicators were obtained with the help of video. Traffic flow surveys were conducted at five weeks with an interval of 15 minutes. The distribution of traffic flow by type and lane is given in the table. Taking into account the obtained results of the intensity and composition of the traffic flow, we can summarize the following. All studies of traffic time intervals on the main street will be conducted at a traffic intensity in the range of 949-1193 units / h and an average share of cars of more than 65%. It was decided to determine how many drivers adhere to safe traffic intervals between cars on the multi-lane street of Lviv (Naukova Street). At the first stage the actual intervals of movement between vehicles on days of the week in the software product "R-project" are investigated. The procedure for installing the R-project program is given, as well as the main functions (copy from the file distance_time.csv, where TV is the time series (vector) of movement intervals with 335 values for the research period). The time series of movement intervals consists of 335 values for each week. Thus, in the program "R-project" the dependence "movement intervals - time" is graphically deduced. The next step is to create time series objects in the R-project program (the program is assigned a number of values each week). There is no function for this. Decomposition in the program "R-project". The general model of the series can be represented as x = m + s + z, where m is the trend, s is the seasonality, and z is the error. "R-project" allows you to easily decompose the series into these components. You can use the decompose and stl functions to do this. decompose uses moving averages to decompose, and stl smoothes loess. The function x = m + s is used as a prediction of the following values. Seasonality in our case - weeks. Time series decomposition can be performed on the following components: trend-cyclic, seasonal and random, using a multiplicative model of the time series. STL decomposition for the time series of traffic intervals between cars - the division of the time series into seasonal, trend and irregular components, using loess smoothing (STL). Based on the obtained values, it is possible to predict the time series of traffic intervals for the future period. To check the forecasting results, the values obtained during the decomposition of time series from the program "R-project" will be transferred to Microsoft Excel. Using a trend and a seasonal variable, you can get the predicted result of a time series of vehicle intervals in a given period. For example, in Microsoft Excel, the next working day is taken - Monday. When comparing the obtained values with the existing ones, the error in milliseconds and percentages was determined. Based on these values, such indicators as MAE (Mean Absolute Error - ms), RMSE (Root Mean Square Error - root of the root mean square error,%) and MAPE (Mean Absolute Percentage Error - the average component of the relative error, ms) . Therefore, it was determined that MAE is 665ms, RMSE - 21.8%, MAPE - 688ms. Based on the obtained data, we can draw conclusions: the data has a pronounced dependence on the days of the week; when decomposed by both functions for the forecast model x = m + s received the same data; the chosen model for forecasting our series is quite accurate, as the error is 21%. The proposed model is universal and self-regulating. To improve the forecast, it is necessary to continue data collection and improve the model. The more data, the better the forecast. Knowing the predicted intervals of vehicles, you can determine the intensity of traffic, density, dynamic size and speed. Because these metrics are related. Object of research: a section of Naukova Street in Lviv. Subject of research: intensity, composition, speed, traffic intervals between vehicles. The purpose of the study: to investigate the time series of traffic intervals between vehicles on the multi-lane carriageway for five weeks and to try to predict the traffic intervals between cars for the future. To achieve this goal, it was decided to use a modern computer program "R-project". As a result of the conducted researches: the analysis of methods and algorithms of traffic management in cities is carried out; the intervals of traffic flows on the main streets are determined. Key words - main street, traffic intervals, forecasting of traffic flow behavior. References. 1. Webster F. Traffic signals / F. Webster, B. Cobbe. – Road Research Technical Paper. – № 56, HMSQ. – London. – 112 p. 2. Teply S. Canadian Capacity Guidefor Signalized Intersections. Second Edition / S. Teply, D. I. Allingham, D. B. Richardson, B. W. Stephenson. – Toronto: Institute of Transportation Engineers, District 7, 1995. – 116 p. 3. Highway Capacity Manual / Washington: TRB, 2000. – 1134 p. 4. Queue Discharge Flow and Speed Models for Signalised Intersections 5. Halasa P.V. Ekspertnyi analiz dorozhno-transportnykh pryhod / P.V. Halasa, V.B. Kyselov, A.S. Kuibida, Yu.O. Lakhno, H.M. Nadhornyi. – K.: «Ekspert-servis», 1995, 193s. 6. Puhachёv Y. N. Orhanyzatsyia y bezopasnost dorozhnoho dvyzhenyia : ucheb. posobye dlia stud. vыssh. ucheb. zavedenyi / Y. N. Puhachёv, A. Э. Horev, E. M. Oleshchenko. – M. : Yzd-yi tsentr «Akademyia», 2009. – 272 s.