Automated traffic control systems (ATCS) are an interconnected set of technical, software and organizational measures that collect and process information about traffic flow data and, on the basis of this, optimize traffic control. The task of automated traffic control systems (ATCS) is to provide traffic safety organizations on the roads.
ASUDD are divided into several types:
Main automated traffic control systems (ASUDD) of coordinated control - centerless, centralized and centralized intelligent.
- · centerless ASUDD - there is no need to create a control center. There are 2 modifications of centerless ASUDD. In one of them, the work is synchronized by the main controller, to which there is communication from the other controllers (one line for all). In the next modification of the centerless ASUDD, all controllers have their own communication line.
- · centralized ASUDD - have a control center, with controllers associated with it, their own communication lines. Often, ASUDD can carry out multi-program KU with a change of programs during the day.
- · centralized intelligent ASUDD - they are equipped with transport determinants, and depending on traffic congestion, they can change traffic coordination plans.
City-wide automated traffic control systems (ATCS) - simplified, intelligent, with traffic control on city roads of continuous traffic and with reverse traffic.
· intelligent ASUDD - contain powerful control computer systems (UVK), and a network of changing information displays. These ATCS can carry out continuous traffic flow control and can manage automatic adaptive traffic control and allow redistribution of traffic flows over the network.
ACS DD, as part of the ITS, performs control and information functions, the main of which are:
- management of traffic flows;
- providing transport information;
- organization of electronic payments;
- Security and emergency management.
In general, the ACS DD subsystems can be represented as a set of road telematics devices, controllers and automated workstations (AWS) included in the data exchange network, with the organization of central and local control centers - depending on the density and intensity of traffic.
As road telematics devices, variable information signs (VPI), multi-position road signs, variable information display (TPI), vehicle detectors, automatic road weather stations (ADMS), video cameras, etc. are used.
The telecommunications part of the ACS DD is a road integrated communication system. The stable functioning of communication systems on roads makes it possible to increase the level of road safety and ensure the efficient operation of road maintenance services, as well as operational and rescue services in case of emergency.
The following functional subsystems can be organized as part of DISS:
- information exchange of ACS DD;
- communication with mobile objects (includes subsystems of operational-technological radio communication and radio access);
- management and technical operation;
- ensuring information security of DISS;
- provision of information and communication services on a reimbursable basis.
Improving the efficiency of traffic management is associated with the creation of automated traffic control systems (ACS DD), which are integral components of intelligent transport systems (ITS). ITS is an integrated information support and management system for land road transport, based on the use of modern information and telecommunication technologies and management methods.
To ensure the functioning of the ACS DD and the provision of infocommunication services to road users, DISS are being created, which are currently subject to the following generalized requirements:
- multifunctionality;
- stability;
- economy.
ACS "GOROD-DD" - is designed to ensure effective control of the movement of traffic and pedestrian flows in cities using means, traffic signaling, video monitoring and registration of violations on the roads, operational analysis of the environmental situation in the city, control of the movement of route transport, etc.
The main advantages and benefits of the automated control system "GOROD-DD"
- - a significant increase in the efficiency of traffic management and control of the state of affairs on the roads, which allows you to save about 5-8 million dollars a year on the scale of the regional center (the savings consist of reducing fuel consumption, reducing the travel time of vehicles, the time spent by passengers on the road, etc. .d.);
- - more efficient use of organizational and preventive measures to normalize traffic on the roads;
- - an integrated approach to the organization of traffic;
- - use of domestic hardware and software, focused on modern technologies and modern methods of traffic management in accordance with the requirements of ISO 9001;
- - new opportunities for monitoring the state of affairs on the roads: visual control of city intersections, video recording of traffic accidents, video recording of violations of the speed limit and rules for crossing intersections, operational analysis of the environmental situation, etc.;
- - the possibility of phased commissioning, by gradually replacing existing traffic control systems with an expired service life and full compatibility of any of the parts of the proposed system (controllers, MCC, MZTS) with all types of existing equipment.
Automated system "Gorod-DD":
- · Central control point;
- · Modules of zonal centers (if necessary);
- · Controllers (in three versions - S, SM, SL);
- · Additional equipment;
- · Software package.
Classification and purpose
Traffic management in conditions of extreme saturation of roads with traffic and pedestrian flows requires more and more advanced methods of traffic control. Recently, the use of automated traffic control systems(ASUDD), which is a complex of technical means that implements certain technological algorithms for managing traffic flows.
The main purpose of the introduction of ATCS is to reduce the total delays of vehicles at intersections in the coverage area of this system - at an intersection, in a district or city. General requirements for ASUDD are defined by GOST 24.501 - 82 “Automated traffic control systems. General requirements".
ATCS classification with separation by management methods shown in fig. 5.3.
Rice. 5.3. ASUD classification
(automated traffic control system)
Local is an ATMS if only information on traffic flows on the approaches to this intersection and in the area of the intersection is used to determine the parameters of regulation at an intersection. With the help of local algorithms, the control cycle, the sequence of control phases, their durations or phase switching moments, and the parameters of intermediate cycles are determined.
feature network ATCS is their use to determine the parameters of regulation of information about the traffic situation at several intersections, usually connected into a single network, characterized by a significant amount of traffic between neighboring intersections and small (up to 600...700 m) distances between them.
As a rule, control cycles for a group of intersections and time shifts for individual traffic light objects are determined at the network level. To determine these parameters, in addition to the data necessary for local control, information is used on the network topology, the relationship of traffic flows on neighboring stop lines and (or) the geometric directions of travel through intersections, the travel times between adjacent stop lines.
By time criterion all traffic light control algorithms are subdivided into algorithms that implement traffic control according to the forecast ( software, hard), and real-time algorithms ( adaptive).
Forecast control does not rule out fairly frequent (up to 3-5 times in a daily cycle) changes in control parameters, however, these parameters are determined not based on the current traffic situation, but by its forecasting based on previous observations.
An intermediate position between adaptive and non-adaptive algorithms is occupied by algorithms based on situational management. The algorithms of this group involve the preliminary calculation of control parameters for various classes of transport situations and the creation of a library of typical control modes. The choice of a specific mode from the library is made in real time based on the current information about the traffic situation and its assignment to one of the classes of traffic situations.
Thus, the methods of automated control of traffic flows in ASUDD can be attributed to one of four classes, as shown in Fig. 5.4 (for each class, the most common control algorithms are indicated).
At present, the most common method in Russia is local hard single-program traffic signal control.
This method is based on a preliminary calculation of the duration of the control cycle and the control phases.
Rice. 5.4. Automated Control Methods
Introduction
The concept of adaptive traffic control in a transport network node
Comparison of Time-Dependent and Vehicle-Dependent Traffic Management Strategy
Statement and analysis of simulation
Development of a base of fuzzy rules, determination of parameters for controlling the movement of traffic flows in a node of the transport network
1 Building a membership function
2 Building rules for matching a specific class of control parameter
3 Fuzzy rule base
Conclusion
Bibliography
Introduction
The changed conditions of mobility, characterized by an increase in the number of cars in recent years, have led to an increase in the burden on transport infrastructure and the environment. The growing need for improved travel conditions cannot be fully satisfied (neither within the settlements nor outside them) only by the creation of new transport links or other construction measures. To get out of this situation, it is necessary to introduce a whole range of measures to organize and manage traffic. Adaptive traffic control systems (ATCS) represent a new approach to organizing traffic control and, together with high-performance transport computers controlled by them, implement appropriate control technologies.
The constant increase in the number of vehicles in conditions of insufficient road capacity leads to difficulties in the movement of traffic flows. Intelligent transport systems (ITS) make it possible to minimize the formation of congestion situations and increase the throughput of the transport network. Developments in the field of ITS are tried on for the organization of traffic in settlements and highways. Optimization of traffic control is achieved through the interaction of managing, classifying, predicting, expert, decision-making or supporting these processes of ITS subsystems. In this regard, the task is to find methods for processing information about emergency situations on the road network (SDN).
In this paper, the following issues will be considered: the concept of adaptive traffic control in a transport network node, on a network, as well as a comparison of time-dependent and transport-dependent traffic control strategies.
1. The concept of adaptive traffic control in a transport network node
Opportunities to improve traffic conditions through the optimal organization of traffic are largely underestimated, and the development of transport infrastructure is understood mainly as activities related to the construction of new roads and highways, the reconstruction of existing overpasses and interchanges. At the same time, the introduction of modern innovative technologies, called "Intelligent Transport Systems" (ITS), makes it possible to significantly improve the transport situation. The introduction of ITS technologies in Russia makes it possible to better manage traffic flows, increase the level of throughput of the road network and reduce the load on its individual elements.
The growth of the car park and the volume of traffic leads to an increase in traffic intensity, which in the conditions of cities with historical development leads to a transport problem. It is especially acute in the nodal points of the road network. Here, transport delays increase, queues and congestion are formed, which causes a decrease in the speed of communication, unjustified excessive fuel consumption and increased wear of vehicle components and assemblies. The changed conditions of mobility, characterized by an increase in the number of cars in recent years, have led to an increase in the burden on transport infrastructure and the environment. The growing need for improved travel conditions cannot be fully satisfied (neither within the settlements nor outside them) only by the creation of new transport links or other construction measures. To get out of this situation, it is necessary to introduce a whole range of measures to organize and manage traffic.
Adaptive traffic control systems (ATCS) represent a new approach to organizing traffic control and, together with high-performance transport computers controlled by them, implement appropriate control technologies. At present, the following traffic control technologies are most common in the world practice as part of the automated control system:
Control technology according to fixed plans (coordinated control);
Network adaptive control technology;
Situational management technology.
HAUDD is a traffic management system with a centrally distributed intelligence, consisting of:
central control point (CPU);
points of adaptive traffic control, equipped with intelligent controllers and traffic detectors, providing:
local adaptive control of the most complex and important intersections and sections of the road network;
information interaction with the CPU;
system detectors reporting information about traffic flows to the CPU;
system controllers controlled from the CPU permanently or periodically.
The organization of traffic at the level of traffic services is a set of engineering and organizational measures on the existing road network that ensure the safety and sufficient speed of traffic and pedestrian flows. These activities include traffic management, which, being an integral part of the organization of traffic, as a rule, solves narrower tasks. In the general case, management is understood as the impact on a particular object in order to improve its functioning. With regard to road traffic, the object of control is traffic and pedestrian flows.
The essence of traffic control is to oblige drivers and pedestrians, to prohibit or recommend certain actions to them in the interests of ensuring speed and safety. It is carried out by including the relevant requirements in the Rules of the Road, as well as using a set of technical means and administrative actions of inspectors of the road patrol service and other persons with appropriate powers.
2. Comparison of time-dependent and transport-dependent traffic control strategies
The current state of traffic management in most cities can be generally characterized in such a way that control devices (nodes) are controlled according to a fixed schedule or according to the state of the traffic flow. The essential difference is that no detectors are needed for time schedule control and the system is unable to respond to any traffic flow changes. In the case of traffic-dependent control of the stop lines, there are detectors that detect the instantaneous presence of vehicles, and the control device thus responds to instantaneous conditions at the node by increasing the duration of the green signal. Therefore, we are talking about control in the second grid of time.
Time-dependent (autonomous) control - transport states are determined on the basis of a statistical analysis of the historical values of the traffic flow characteristics (traffic intensity) and on their basis the output values of the control process are determined.
Transport-dependent (current time mode - online) control, in the Anglo-Saxon literature, also called Traffic Responsive, is that the intervention of the control system is calculated from the instantaneous traffic situation. Online methods provide real-time operation and, based on variable input traffic data, change and optimize control parameters every second, i.e. the duration of the green signal in the corresponding direction. Control devices in this mode operate independently or, in extreme cases, are located in a line and linearly coordinated.
Management is carried out at the local level. If a control center is used, then the status of the control devices or the status of traffic flows is often monitored afterwards. Real-time control of traffic lights is well known and is commonly referred to as vehicle dependent control or dynamic control. Its principle is that the transport hub is usually equipped with two types of sensors: interval and call sensors, which are in most cases inductive loops. The transport control device controls a program that continuously tests the state of the traffic stream over individual sensors and, based on predetermined algorithms, increases the duration of the signals, modifies the phase sequence or inserts a phase on call. These changes are typically made within a predetermined cycle time and predetermined maximum green signal durations. The interval sensor, located approximately 30-50 m before the stop line, got its name from the fact that it continuously measures the time intervals between vehicles and if they are less than this value (usually 3-5 seconds), then it increases the duration of the green signals up to a predetermined maximum. This measurement method is called "Time Interval Measurement Control". The second possibility is that the individual nodes are connected to a traffic control centre, which coordinates and manages the operation of the nodes at the district level. The following modes are used to control the area:
Time-dependent (autonomous) control - information on the characteristics of the state of traffic flows in the area is obtained by statistical analysis, data on the characteristics of the movement of traffic flows (intensity and composition of traffic) for the past years, measured at the main points of the transport network, and on their basis the mode is determined operation of transport control devices. Then they are entered into control devices depending on the time of day or day of the year. The calculation optimizes the duration of the green signals, the duration of the cycle and the time shift. An example of an offline-based method is the TRANSYT method, where dummy vehicles are “released” according to predetermined rules into an area and pass through the area based on and according to a traffic pattern. Their movement is affected by a change in the controlled parameters of the node. With the help of numerical mathematical methods for various parameters, such as cycle time, green signal duration and time shift, the minimum of a certain objective function is found (parameter optimization).
Transport-dependent (online mode) control is characterized by the fact that for various conditions of traffic flows on the network, systems of signal plans are calculated in advance, which are stored in control devices or in the traffic control center. The TRANSYT method is typically used to calculate the maximum values for green duration, cycle time, and time offset. At the same time, strategic sensors are selected in the area and logical equations are compiled that describe different combinations of the states of all or selected sensors. Depending on the instantaneous traffic situation, the program that best suits the given situation is selected by means of an appropriate equation. An example is the description of the state of the traffic flow according to the strategic sensors SDV1 and SDV5, which means: if at the point SDV1 there is degree 2 and at the same time at the point SDV5 - degree 4, then you should select the signal program number 6.=2 &SDV5=4 THENSP6
If the network does not classify the state of the traffic flow, then only one parameter is used for description, which is the traffic intensity. Vehicle dependent control is used in real time and receives signals from selected sensors every second. However, the switching of signal programs is carried out with a certain hysteresis to ensure stability in the transport network. In practice, this means changing the program of the control device in a grid of several tens of minutes.
Offline optimization makes it possible to calculate the main controlled variables: cycle time, phase sequence, time shift and green signal duration for the historical database (historical data). This data is obtained by long-term measurement with transport detectors. On the basis of long-term recorded data, a statistical model is usually developed, which, for the traffic volume, usually makes it possible to determine typical working days and especially Saturday and Sunday, as a result of which the changes in variables are highly limited. The essential feature is that we are talking about off-line macroscopic control based on deterministic flow modeling and optimization algorithms, when the systems of signal plans are calculated from the space-time vector of intensity data for previous years. Optimization models are used for offline calculations of signal time plans of transport control devices in a transport network or line.
In this case, the control process chooses, depending on time, the most advantageous of the set of pre-prepared signal plans. This method is called time-dependent control.
Advantages of Time-Dependent Control:
the possibility of simple control;
ease of modification of signal programs;
relatively low equipment and installation costs.
Disadvantages of time dependent control:
it is not possible to improve the efficiency of using the time of signals (permission of movement for individual directions);
intensity peaks cannot be covered (a certain intensity reserve is required);
it is impossible to enter into the process of control by individual vehicles or pedestrians;
traffic congestion cannot be eliminated.
3. Statement and analysis of simulation
The task of modeling traffic control strategies in a transport network node, as well as on a network, is to develop a classical fuzzy control module. Its components:
Fuzzification Block: The fuzzy logic control system operates with fuzzy sets, so a particular value of the input signal of the fuzzy control module is subject to a fuzzification operation, as a result of which a fuzzy set will be associated with it.
The rule base is a set of fuzzy rules for determining the fuzzy set to which the system output signal belongs.
Decision-making block: direct determination of the membership set of the output signal for specific sets of input signals.
The defuzzification block represents the procedure for mapping the fuzzy set obtained at the output of the decision block into a specific value, which represents the impact control.
To build control strategies, it is proposed to use the TRANSYT software package, based on assessing the behavior of the traffic flow using traffic simulation and allowing you to choose the optimal parameters for the traffic signal operation mode. According to the results of traffic simulation in the program for various combinations of traffic intensity, the optimal time for the green traffic light to burn is determined.
4. Development of a base of fuzzy rules for determining the parameters of traffic flow control in a transport network node
Building a base of fuzzy rules for determining the optimal time for the green signal of a traffic light to burn at an intersection characterized by maximum traffic on intersecting roads. The necessary data were obtained using a transport detector.
We create a rule base for the classification of control strategies for a system with two inputs and one output:
1. Data is required in the form of a set. Next, we find the domains of definition of the elements of the set , which we divide into domains (segments), and the value of N is selected individually, and the segments can have the same or different lengths. Separate areas can be designated as follows: …, S,,…,.
We construct membership functions for a certain class of elements of a given set of training data. We propose to use triangular-shaped functions according to the principle: the top of the graph is located in the center of the splitting area, the branches of the graph lie in the centers of neighboring areas. The degree of data belonging to a certain class will be expressed by the value of the membership functions.
Then, for each pair, we determine the rule of correspondence to the class of the control strategy. The final rule for each pair of training data can be written 1 rule, that is
Since there are a large number of pairs available, there is a high chance that some of the rules will be inconsistent. This refers to rules with the same premise (condition) but different means (conclusions).
One way to solve this problem is to assign a so-called degree of truth to each rule, and then choose the contradictory rules of the one with the highest degree of truth. After that, the rule base is filled with qualitative information.
For example, according to the rules described above, the degrees of truth have the form
4. To determine the quantitative values of the control strategy optimization parameter, it is necessary to perform the defuzzification operation. To calculate the output value of the impact control, it is possible and recommended to use the defuzzification method according to the center of gravity method.
1 Building a membership function
For the elements of the set of training data, we denote the following domain of definition
Having divided X 1 X 2 and G into 2n+1 segments, we construct membership functions of the form
Figure 4.1 General view of the graph of membership functions
We end up with:
Figure 4.2 Graphs of the membership functions of the intensity x 1 to the classes of the partition of the set X 1.
We determine the membership functions µ(x 1) on segments of the division of the region X 1 by assigning µ(x 1) to a certain class.
Table 4.1. Membership functions µ(x 1) on the segments of the division of the region X 1 (n=4)
Split segment Designation Membership function µ(x 1) , , , , , Figure 4.3 Graphs of the membership functions of the intensity x 2 to the classes of the partition of the set X 2 . We determine the membership functions µ (x 2) on segments of the division of the region X 2 by assigning µ (x 2) to a certain class according to Figure 4.3. Table 4.2 Membership functions µ(x 2) on the segments of the division of the area X 2 (n=5)
Split segment Designation , , , , , , , Figure 4.4 Graphs of the membership functions of the intensity g to the classes of the partition of the set Q. We determine the membership functions µ(g) on segments of the partition of the domain G by assigning µ(g) to a certain class Table 4.3 Membership functions µ(g) on segments of the partition of the domain G(n=6)
Split segment Designation Membership function µ(x 2) , ; ,; ,; ,; 2 Building rules for matching a specific class of control parameter We define the rule of correspondence to the class of control strategies and assign a degree of truth to each rule. Table 4.4 Values of data membership functions for certain classes
(i)µ((i))(i)µ((i))g(i)µ(g (i)) We get a table with the assigned degrees of truth and the degree of truth for each of the pairs x 1 , x 2 . transport management road passenger Table 4.5 Fuzzy rules generated from training data and the degree of truth of these rules
3 Fuzzy rule base According to the rules defined in Table 4.7, we compose a base of fuzzy rules that determines the optimal value of the green traffic light. Table 4.6 Fuzzy rule base
Conclusion In this paper, the following issues were considered: the concept of adaptive traffic control in a transport network node, on a network, as well as a comparison of time-dependent and transport-dependent traffic control strategies. The main concepts of adaptive control implemented in different countries and the advantages such as: ensuring high performance in the face of changing properties of the controlled object, environment and goals, through the development of new functioning algorithms. The organization of the movement of urban passenger public transport during the operation of an adaptive traffic control system, the implementation of this condition occurs due to the installation of radio tags on vehicles and readers on traffic light objects. Recognition of the vehicle will allow "stretching" the time of burning the green signal and ensure unhindered passage of public transport. It is also possible to use the principle of data exchange directly between the controllers of neighboring intersections. The data of the detectors connected to the road controller are complemented by the data of those detectors that are installed at neighboring intersections. This allows you to directively set the state of signaling groups, and also provides priority for public transport Since adaptive control is very expensive, an alternative method was proposed to determine the optimal time for the green traffic light to be on at the intersection. Namely, the method of developing a classical fuzzy control module, the initial data for which were sets of data on the intensity of two intersecting roads. In this paper, the first three blocks of this method were considered and calculations were carried out. Bibliography 1. P. Przhibyl, M. Switek "Telematics in transport", 2004; Konoplyanko, V.I., Gudzhoyan O.P., Zyryanov V.V., Berezin A.S. Traffic safety. Kuzin M.V. Simulation modeling of traffic flows in a coordinated control mode Omsk - 2011; V.G. Kocherga, E.E. Shatalova Technical means of modern automated traffic control systems. Rostov-on-Don 2011; E.A. Petrov article "Adaptive traffic control system as part of urban ITS"; Abramova L.S. Journal Bulletin of Kharkov National Automobile and Highway University.
;
;
;
;
;
;
;
;
;
;
;
; 
; 
;
;
;
;
; 
;
;
;
;
;
;
;
;
;
UDK 517.977.56, 519.876.5
adaptive traffic control based on a system of microscopic simulation of traffic flows
A. S. Golubkov,
engineer, junior researcher
B. A. Tsarev,
cand. tech. Sci., Associate Professor Institute of Management and Information Technology Cherepovets Branch of St. Petersburg State Polytechnic University
The composition and features of the functioning of modern automated traffic control systems are described. A method for adaptive traffic control based on traffic flow prediction and fast intersection optimization models is proposed. The characteristics of the system of microscopic simulation of traffic flows used in the system of adaptive traffic control are presented.
Keywords - adaptive traffic control, traffic control optimization, traffic flow simulation, microscopic simulation.
Introduction
Currently, in many large cities the problem of traffic congestion is very acute. At the same time, studies show that the potential of existing road networks (SRNs) is far from being fully used. Increasing the traffic capacity of the road network can be achieved through the introduction of automated traffic control systems (ATMS). With the introduction of ASUDD, the following indicators are improved: the travel time of vehicles (TC) is reduced by 10-15%; the number of general transport stops is reduced by 20-40%; fuel consumption is reduced by 5-15%, the amount of harmful emissions into the atmosphere is reduced by 5-15%; improves road safety.
Modern ASUDD
The main components of modern automated control systems, in addition to traffic lights and traffic light controllers, are:
1) transport detectors (DT), which provide detection of vehicles and counting their number when driving along the lanes;
2) one or more computers for data processing with DT and calculation of optimal control signals;
3) a set of software tools that implement algorithms for detecting transport and optimizing traffic control;
4) means of informing the drivers of the vehicle (various information boards);
5) means of communication and telecommunications used to combine the ASUDD software and hardware into a single system.
Various types of transport detectors are used in modern automated control systems: loop (induction); infrared active and passive; magnetic; acoustic; radar; video detectors; combined (in various combinations of ultrasonic, radar, infrared and video detectors). All diesel engines have different efficiency in different operating conditions. However, due to the high level of development of computer and television technology, in many cases, video detectors based on image processing and analysis technologies, as well as combinations of video detectors with detectors of other types, are most preferable.
In the existing ASUDD of various manufacturers, three main methods of adaptive traffic flow control are used in various combinations.
1. A control method using libraries, characterized by pre-calculation of a plurality of coordination plans and switching them based on the current average readings of strategic DTs by selecting the appropriate appropriate plan from the library.
2. The actual control method, characterized by the preliminary calculation of traffic light coordination plans, their switching according to the calendar schedule and the implementation of changes in these plans in accordance with traffic requests recorded by local detectors in certain directions.
3. An adaptive control method characterized by constant recalculation of coordination plans and calendar modes based on information received from local and strategic (track) detectors in real time.
Optimization of traffic flow management in modern ATCS is carried out by various methods. The Balance system (Germany) uses genetic optimization algorithms. In the Utopia system (Netherlands), the calculation is based on a price function that takes into account the delay time, the number of stops, specific priority requirements, and the relative position of intersections. In the Spektr system (St. Petersburg, Russia)
the following algorithms are used: search for traffic flow breaks; calculation according to the Webster formula; switching programs by intensity. The ASUDD manufactured by OAO Elektromekhanika (Penza, Russia) uses the following algorithmic support: an algorithm for searching for a break in traffic flows; search for a gap while maintaining the total duration of the coordination cycle; algorithm for switching pre-calculated modes by control points of traffic intensity; algorithm for dynamic recalculation of cycle parameters based on the Webster formula. In ASUDD "Agat" (Minsk, Belarus) the following heuristic control algorithms are used: selection of the coordination plan according to the time map; phase, mode selection according to the coordination plan; selection of the coordination plan according to the motion parameters at characteristic points, etc.
Adaptive Traffic Flow Management Based on Intersection Optimization Models
The developed traffic control system (figure) consists of one central point and many local points.
■ Diagram of adaptive traffic control system
control kts, the number of which corresponds to the number of controlled intersections in the system. All local points have a connection via communication channels with the central control point.
The central control point performs the functions of collecting and processing information about the traffic intensity of vehicles in the road network. Information processing is the prediction of traffic flow values based on the following data:
Current intensities of traffic flows;
Vehicle speeds;
Distances between adjacent controlled intersections in the system;
Prediction of vehicle routes based on statistics for the current day of the week and time of day;
The current lengths of the phases of traffic light objects at UDS intersections.
The local points in the system perform direct traffic management optimization at the respective intersections. Each local control center includes:
Transport detectors;
A computer that performs data preprocessing with DT, if necessary, and optimization of traffic control;
Traffic light controller that allows external setting of the phase lengths of a traffic light object;
Traffic lights.
It is proposed to use video detectors as DT. In this case, the signal from the cameras enters the computer of the local control center, where the pre-processing software module performs video image analysis and estimates of traffic flow intensities in all controlled lanes. Further, the intensity of traffic flows are transmitted to the central control point.
Optimization of traffic control is performed as follows. The computer has an accurate software microscopic model of the intersection. When calculating the optimal phase lengths for the next phase cycle of controlling a traffic light object (the duration of the phase cycle is usually 2-5 minutes), the following actions are performed.
The model specifies the input intensity of traffic flows for the next 5 minutes (intensity forecast from the central control point) with an accuracy of an individual vehicle.
The optimization module launches runs of the intersection model with a duration of 5 minutes of model time, for each run it sets new phase lengths of the model traffic light object
and calculates the value of the objective function based on the results of each run.
As a result of an optimization cycle consisting of several runs of the model, the optimization module finds the optimal phase lengths of the model traffic light object corresponding to the extremum of the objective search function.
The phase lengths of the traffic light object are a vector of optimization parameters j = (fr f2, f3, f4) (no more than four phases are usually set at a crossroads). As the objective function F(j) can serve as the average waiting time for the passage of the intersection of the vehicle. In this case, the optimization criterion will be the minimum average waiting time for a passage
min .P(f) = F(^*),
where Ф is an admissible set of values of the coordinates of the vector of phase lengths; j* - vector of optimal values of phase lengths. The admissible set of coordinate values of the phase length vector has the following form:
Ф = (Ф|Tmin< Фi < Tmax.i = 1.-. 4} С r4.
where T. and - respectively, the minimum
and the maximum value of the phase length.
The calculation of the derivatives of the objective function on the model is impossible, therefore, only direct methods can be used as optimization methods. The use of alternating cyclic variation of the phase lengths of a traffic light object from run to run with a constant step along the phase length is proposed. The length of the phase length variation step can be set to 2-3 s.
A necessary condition for the possibility of implementing the described system of adaptive traffic control is the presence of a system of microscopic simulation of traffic flows, the speed of which would be sufficient to optimize the lengths of the phases of a traffic light object during one phase cycle.
System for microscopic simulation of traffic flows
The authors of the article have developed a system for microscopic simulation of traffic flows in the street network, which can be used to optimize traffic management as part of an adaptive traffic control system. The main feature of the simulation system is the use of a discrete-event approach in modeling
due to which the system has a high speed.
The performance of the system was evaluated in a series of experiments with models of individual typical intersections. The experiments were performed on a computer with an Intel Core 2 Quad Q6600 processor with a frequency of each core of 2.4 GHz (actually, only one core was used in the experiments, since the simulation is performed in one program thread). As a result, simulation of traffic flows through a single intersection for 45 days (3,888,000 s) took 2864 s of CPU time. Thus, the excess of the simulation speed over the real time flow rate was 3 888 000/2864 « » 1358 times, i.e., during the real phase cycle at the intersection, the optimization module is able to perform more than 1300 runs of the optimization experiment.
A feature of the discrete-event approach in modeling is the independence of the simulation results from the model execution speed, i.e., even in the full processor load mode, the simulation will show completely identical results to the execution results, for example, in real time.
On the contrary, in the system-dynamic approach, when the simulation is accelerated by increasing the sampling time step, the accuracy of the simulation decreases. The system-dynamic approach implements the vast majority of modern systems for microscopic modeling of traffic flows: Aimsun (Spain), Paramics Modeler (Scotland), DRACULA (Great Britain), TransModeler (USA), VISSIM (Germany) . In all the above simulation systems, a time sampling step of 0.1-1.0 s is used.
In a system-dynamic road and transport model, a simulation step in time equal to 1 s is quite capable of depriving the model of adequacy. Thus, a vehicle at a speed of 60 km/h travels more than 16 m in 1 s, i.e., at typical speeds, a model vehicle is positioned only with an accuracy of about 10 m.
In the proposed discrete-event model, the positioning accuracy of model objects remains constant at almost any speed and depends on the bit depth used.
1. Brodsky G. S., Aivazov A. R. Automated traffic control in the urban environment. 2007. No. 26. S. 2-3.
variables and the type of arithmetic operations performed on them. When using double-precision floating-point numbers (64 bits, 15 significant decimal digits of the mantissa), the positioning accuracy of model vehicles in the discrete-event model at any time will be no more than 1 cm.
Conclusion
The proposed adaptive traffic control system is able to demonstrate high efficiency due to the exhaustive optimization of each individual intersection and taking into account traffic flows between neighboring intersections with an accuracy of individual vehicles. If there is a high density traffic flow in the road network in any direction, the control is automatically adjusted at all adjacent intersections with the organization of a green wave in this direction. At the same time, all other directions with traffic flows of lower density are also subject to optimization.
Optimization of the control of each individual intersection in real time is possible due to the use of a system of microscopic discrete-event modeling of traffic flows in the road network developed by the authors of the article. This modeling system, due to the use of a discrete-event approach, has high performance and accuracy. In the near future, a trial version of the modeling system will be available on the developers' website.
The quality of traffic management optimization is highly dependent on the accuracy of traffic density prediction. In this case, the prediction accuracy is the higher, the smaller the prediction time interval. When using sufficient performance hardware at local intersections, recalculation of the optimal lengths of the phases of the traffic light object regulation cycle can be performed with the beginning of each next phase. In this case, the actually used prediction time interval will be reduced to the duration of one phase, i.e., to 15–100 s, which will increase the optimization efficiency.
2. Brodsky G. S., Rykunov V. V. Let's go! ASUDD - world experience and economic sense // Mir roads. 2008. No. 32. P.36-39.
3. GNPO AGAT. http://www.agat.by (date of access:
4. Crowdhury M. A., Sadek A. Fundamentals of Intelligent Transportation System planning. - Boston - London: Artech House, 2005. - 190 p.
5. Kremenets Yu. A., Pechersky M. P., Afanasiev M. B. Technical means of organizing traffic. - M.: Akademkniga, 2005. - 279 p.
6. GEVAS software: Traffic Control. http://www.gevas.eu/index.php?id=149&L=1 (accessed 16.06.2010).
7. UTOPIA - Peek Traffic. http://www.peektraffic.nl/ page/484 (date of access: 16.06.2010).
8. CJSC "RIPAS": Development and production of automated systems. http://www.ripas.ru (date of access: 06/16/2010).
9. ASUDD - OJSC Electromechanics. http://www. elmeh.ru/catalog/3/asud (date of access:
10. Karpov Yu. G. Imitation modeling of systems. Introduction to modeling with AnyLogic 5. - St. Petersburg: BHV-Petersburg, 2006. - 400 p.
11. Sovetov B. Ya., Yakovlev S. A. Modeling systems. - M.: Higher. school, 2001. - 343 p.
12. Nagel K. High-speed microsimulations of traffic flow. Thesis: University Cologne, 1995. - 202 p.
13. Aimsun. The integrated transport modeling software. http://www.aimsun.com (accessed:
14. Quadstone Paramics. Traffic Simulation Solutions. http://www.paramics-online.com (accessed:
15. SATURN Software Web Site. https://saturnsoftware.com co.uk (Accessed 20 May 2010).
16. TransModeler Traffic Simulation Software. http://www.caliper.com/transmodeler/ (accessed:
17. PTV Vision - transport planning. http://www.ptv-vision.ru (date of access: 05/20/2010).
18. Company "Mullen". http://www.mallenom.ru (date of access: 20.05.2010).
Each of you needs to register on the RUNEB website (http://www.elibrary.ru) in order for you to be assigned an individual digital code (during registration, the code is assigned automatically), which is necessary to create a correct RUNEB database that objectively reflects information about your scientific activity, as well as to calculate your citation index (RSCI).
