EP2280383B1 - Method for determining traffic information for a section of a road network and traffic calculator to implement the method - Google Patents

Description

The invention relates to a method for determining traffic information for a road section of a road network and to a traffic computer for carrying out the method, to a machine-readable program code for the traffic computer and to a storage medium with program code stored thereon.

The determination of traffic information for traffic signal controlled roads, especially the accurate detection of the traffic situation on a traffic signal system, is possible with high manual effort by observation. The measurement data of the commonly used vehicle detectors for traffic control, usually 10 to 40 m in front of a traffic signal arranged induction loops, can not or only partially be used to accurate traffic information on the number of stopping vehicles, waiting times or in the manual for the design of Road traffic facilities, known by the short name HBS, defined levels of traffic quality to win. However, it is precisely the recording of the coordination quality of successive traffic signal systems over the proportion of stopping vehicles, ideally even knowledge of the arrival distributions of the vehicles, that is an essential parameter for assessing the traffic flow quality at traffic light installations.

In the case of the offline determination of the traffic situation, for example for expert reports, it is known for precise determination of the traffic quality to carry out fixed observations. This is done, for example, by manually counting the proportion of vehicles coming to a stop at a driveway to a traffic signal for a certain period or by the subsequent evaluation of video recordings of the driveways. Waiting times can only be determined qualitatively in this way. A distinction as to whether the vehicles traveling on the road route arrive at the upstream node point via the main direction or are inflected from one of the secondary directions is generally not possible.

Very expensive, however, is the driving of the road in question by vehicles and the measurement of the course of the journey, for example, to lead a proof of a network control measure. This is done either by recording the arrival times at the end of a queue or at a stop line or by logging using GPS receivers (GPS, short for Global Positioning System, to German Global Positioning System). The evaluation of the recorded measurement data then allows statements about the travel times or waiting times, about the proportion of stops for the routes traveled by the measuring vehicles, and thus also about the quality according to HBS. Statements about the traffic quality for non-drivers whose routes were not used are not possible.

From the publication DE 103 59 977 A1 For example, a method for determining a travel time for a vehicle in a route network that can be subdivided into sections is known. In this case, the route network is subdivided into route sections such that a route section ends downstream on a traffic light system controlling the traffic flow on this route section. At a measuring point along the route section, the traffic flow characterizing traffic data is detected by means of a road-side measuring device. From the temporal behavior of the continuously recorded traffic data and the derived current traffic situation on this section, a travel time for this section is calculated. Although the data collected for jam lengths and travel times are accurate enough to be used for control measures in a traffic management system, they are not used for expert opinions. This known Furthermore, the method does not provide any information about the coordination quality of successive traffic signal systems.

From the European patent application EP 0 501 193 A1 is a method for automatic traffic coordination of an independent node control unit of a road traffic signaling system with adjacent nodes known. The traffic flowing in from there is detected and analyzed, wherein the number of vehicles per time interval is determined and stored for a given measurement period with a plurality of time intervals. The measuring period is divided into test cycles with the same cycle time. Within the test cycles, the number of vehicles per respective time interval is added up. Such a mapping of the entire measurement period to a test cycle allows averaging and variance formation. For further test cycles, each with different cycle times, such mappings and variance calculations are made. Since the magnitude of the variance is a measure of the cyclic property of the incoming traffic, an existing vehicle cycle and its cycle time for the coordinated control of the signaling system is determined from the calculated variances.

The European patent application EP 1 276 085 A1 shows a method for determining a congestion indicator and for determining tailback lengths. A method for determining a congestion number at operator stations for handling individually moved units is disclosed. With the aid of the traffic jam number, two methods for estimating the back-up length at the operating station are obtained. The first method exploits a linear relationship between backpressure length and smoothed congestion figure. In each dispatch phase, the slope of the queue length function is calibrated by comparing the current traffic jam number with a lower limit for the traffic jam length. In the second method, the backlog length is calculated from the traffic jam count and the saturation time requirement using a macroscopic queuing model. Although this known method is suitable for online use, However, it does not provide any information about the coordination quality of successive traffic signals.

The publication DE 101 08 611 A1 shows a method for simulating and predicting the movement of individual vehicles on a network of traffic routes with network nodes and these connecting sections by microscopic quantities using currently measured and historical traffic data. In a first step, macroscopic traffic variables are determined and in a further step, the microscopic individual vehicle sizes are generated separately for each vehicle.

From the US publication " Configuring Micro-Genetic Algorithms for Solving Traffic Control Problems: The Case of Numbers of Generations, published by Abu-Lebdeh and Al-Omari in IEEE Proceedings of the Fourth International Symposium on Uncertainty Modeling and Analysis, 2003, pages 1-8 , the use of genetic algorithms in optimization tasks in transport networks is known.

From the DE-publication "Online evaluation of green waves: A fuzzy expert system for estimating the loss time before light signals by means of near-line detectors", published in 2004 in traffic engineering, author: Braun, Mück, discloses a method for online evaluation of the coordination quality between two traffic lights , It needs a stop near detector in front of the downstream traffic signal system, which measures the occupancy rate every second and counts the number of vehicles. In addition, the current signal position of this traffic light system must be known. From the detector data and the signal position six characteristic quantities are determined for each circulation. In order to evaluate the coordination quality between two traffic light systems, a fuzzy expert system was developed, which estimates the average loss time from these quantities per revolution. This can be used for unconstrained undisturbed traffic conditions on the coordination quality be closed between this and the upstream traffic signal system. A fuzzy expert system makes it possible to linguistically formulate complex relationships. For the present problem, this approach is very well suited, since the traffic-technical relationships between the characteristic quantities and the loss time can be linguistically formulated, and processed as expert knowledge. The online evaluation method of coordination quality can be used to verify the quality of existing green waves. A use in the context of traffic-dependent controls of traffic lights is basically conceivable, since the method works under real-time conditions. However, the method does not provide any information about the compositions of the vehicle spool in the inflow. In addition, the method is complex and contains a lot of heuristics. It does not include a historical history of the data and thus omits information. Finally, the accuracy of the procedure is not suitable for expert opinion and quality assurance.

The EP 1 480 183 A1 1 shows, according to the preamble of claim 1, a method for determining traffic parameters at operator stations for handling individually moving units with alternating blocking and transmission phases and with a detector located in front of the operating station, with the steps of providing the points of a basic diagram for the operating station using the Detector data and determining at least a subset of points of the fundamental diagram corresponding to a traffic condition.

From the European patent application EP 1 480 184 A2 A method is known for determining traffic parameters at operating stations for handling individually moving units. The operating stations have alternating blocking and passage phases and a detector arranged in front of the operating station. First, points of a plurality of comparison basis aliagrams and points of a characteristic diagram are added to each point of a comparison foundation alias a traffic characteristic value. Further, using detector data, points of a fundamental diagram are provided for the operator station. Finally, points of the fundamental diagram are automatically compared with points of each one of the comparison foundation balance diagrams according to a predetermined criterion until a predetermined similarity condition is satisfied.

All known methods have in common that they allow either no or only a rough statement about the distribution of vehicle arrivals at a light signal approach. In any case, the achieved accuracies are not satisfactory. It is also common to the method that, because of the high calibration effort or the missing accuracy or characteristic data, they are not used as methods in the context of an automated quality assurance of the coordination of successive traffic signal installations, e.g. by a daily retrospective evaluation, are suitable.

The invention is therefore based on the object to provide a method for determining traffic information and a traffic computer for performing the method, which in an automated manner an accurate determination of traffic information, such as waiting times and stops of individual vehicles, quality characteristics, and the like, is possible.

The object is achieved by a method for the determination of traffic information for a road section of a road network, the road section an entry cross section, through which a road route inquiring traffic flows, an exit cross section through which flows a controlled by a traffic signal traffic flow, and at least one between Entry and exit cross-section arranged measuring cross section at which a vehicle detector detected by passing vehicles generated measurement data, wherein the traffic flow along the road route simulated by means of a traffic model and, depending on an incoming model traffic flow, model measurement data associated with the measurement data is generated, wherein the incoming model traffic flow is varied with respect to the time distribution of model vehicles entering the model road route and with respect to a match of the respectively generated model measurement data is optimized with the corresponding measurement data acquired by the vehicle detector, and wherein the traffic information is determined from the simulated model traffic flow resulting from the optimized model traffic stream. The simulation of the traffic flow takes place, for example, on the basis of a macroscopic traffic model which is known per se, wherein the modeling of the traffic flow can take place, for example, over surfaces of constant traffic density moving along the modeled road route. The method according to the invention is therefore based on a simulative simulation of the real traffic flow and the resulting measurement data by means of a suitable traffic model. The simulated model traffic flow is modified in its generation characteristic at the entrance cross-section - ie the time distribution of vehicles entering through the entry cross section - until the modeled measurement data generated by it are as similar as possible to the measured data actually measured. The optimized incoming model traffic flow, or the resulting simulated model traffic flow then serves to derive the sought-after parameters, which are customary in practice, as traffic information. It is essential for the method according to the invention not to model the inflowing traffic flow via an upstream traffic signal system, but rather via a targeted modeling of the inflowing traffic flow. With the obtained traffic information, it is possible to automatically determine or evaluate traffic parameters as well as the quality of traffic signal controls and green waves with little effort and without carrying out test drives.

In an advantageous embodiment of the method according to the invention are subdivided over a time intervals Examination period as measured data a count of detected vehicles per time interval and an occupancy value of the vehicle detector per time interval are detected, with count and occupancy values are determined from raw data of the vehicle detector. An essential point here is the initial systematic evaluation of the raw data of the vehicle detector in the form of time data of its rising and falling edges or in the form of finely resolved count and occupancy values every second for the determination of the coordination quality. In the case of the detector edge data, these are evaluated in a suitable, derived from the physics of traffic flow manner by the distance behavior and the transit time via the vehicle detector for each vehicle in macroscopic characteristics that may be finer in time than the time intervals of the investigation period, converted and possibly smoothed. The use of the raw data enables, in comparison to the previously known methods, the exact determination of results in examination periods which are considerably shorter than hitherto customary, for example 10 to 30 minutes. By using this method, in contrast to comparing direct edge data from the model and the measurement, no further smoothing is needed to calculate the goodness of the match. By avoiding smoothing, the process becomes considerably more accurate. The measurement data generated in this way, in accordance with the count and occupancy values determined by macroscopic models, are significantly better than directly smoothed raw data.

In a preferred embodiment of the method according to the invention, the inflowing model traffic flow is respectively related to a circulation time of a signal time schedule running in a traffic light system controlling the incoming traffic flow. In this case, the fact is taken into account that a light signal-controlled incoming traffic flow, the vehicles enter in orbital periodic pulses through the entrance cross section in the road section. In the traffic model, therefore, the inflowing model traffic flow based on the real orbital period of the inflow controlling traffic signal system.

In an advantageous embodiment of the method according to the invention, the measurement data acquired during the time intervals of the examination period are allocated correspondingly to the individual signal cycles of the signal time schedule. If the time intervals of the examination period are already adapted to the signal cycles of the signal time schedule, the measurement data can be used directly. If the signal circulations overlap the time intervals with different time durations and / or starting times, then the measured data of a time interval must be divided up in relation to its overlap with the signal circulations. Thus, the measurement data are tuned to the signal circulations of the incoming traffic flow stamping traffic signal.

In an advantageous refinement of the method according to the invention, the model traffic flow flowing in during a circulation time is formed by multiplying the sum of the counted values (z _i ) associated with circulation with detected vehicles (F) having a normalized pulse profile (p ') which has a time distribution of Vehicle proportions of an inflowing vehicle pulse within a cycle time indicates, wherein the inflowing model traffic flow is varied by varying the underlying pulp profile. Here, the pulk profile as a central, to be estimated characteristic of the erfindungsgemäβen method is based. By means of this pulp profile, the per se known, actually present pulp profile should be approximated. The pulse profile includes the duration of a round trip time of the signal time schedule of the inflow controlling traffic signal system. The pulse profile is normalized, for example, to a unit area and indicates in which time periods of a circulation time of the incoming traffic flow controlling traffic signal, which proportion of the total number of vehicles retracted during the orbital period through the entrance cross section into the stretch of vehicles.

Advantageously, the formation of the model traffic flow flowing in during the investigation period is based on the same pulse profile for each revolution time. This assumption is justified for sufficiently short examination periods, for example up to one hour, and simplifies the method by using a constant pulse profile for all signal circulations.

In an advantageous embodiment of the method according to the invention, the traffic flow is simulated by simulating movements of model vehicles of the incoming model traffic flow along a model roadway, which generate model measurement data when passing a model measurement cross section and which by a light signal controlled model exit cross section flow away. Compared to a macroscopic traffic model, the use of a microscopic traffic model based on a targeted insertion of individual model vehicles at the model entrance cross section provides detailed traffic information.

In an advantageous embodiment of the method according to the invention, the inflowing model traffic flow is formed from model vehicles of different vehicle classes, each with a mean vehicle length, wherein a composition of the model traffic flow from vehicle classes and their average vehicle lengths are predetermined or varied. Thus, the incoming model traffic flow can be examined not only in terms of its time distribution during a round trip time, but also in its composition with respect to different vehicle classes, such as passenger cars, trucks or buses, which have different acceleration and cruising speed values. After optimizing the incoming model traffic flow, characteristic values for different vehicle classes are also available. By including characteristics of a traffic model, such as the class distribution or the class properties - such as the average vehicle length - in the optimization process, The traffic model used for the simulation can be calibrated.

In an additional preferred embodiment of the method according to the invention, the repetitive cycle time or a sequence of changing cycle times of the signal time schedule, which takes place in the light traffic system controlling the incoming traffic flow, is varied, the acquired measurement data being correspondingly associated with the repetitive cycle time or the changing cycle times become. This method can be used with advantage in uncertainty with respect to the coincidence of the transit times of the transmitting and considered traffic signal system. Optionally, a simple pulp profile composed of, for example, a main direction pulp and a pitching pulse may be presumed so that the optimization is limited to determining the revolutions of the transmitting traffic signal.

In a further preferred embodiment of the method according to the invention, a distance between the respectively generated model measurement data and the corresponding measurement data acquired by the vehicle detector is calculated for at least a portion of the time intervals and an average value of the distances for the portion of the time intervals of Examination period minimized. If the traffic volume in the units of vehicles per hour and the occupancy rate in percent are present as measurement data per time interval, the square root could be formed as the distance measure for a specific time interval from the sum of the squares of the differences between the real measurement data and the model measurement data. From all distance measures of the time intervals of an examination period, an arithmetic mean value is now formed which is minimized iteratively by varying the pulse profile generating the model measurement data.

In a further preferred embodiment of the method according to the invention, the pulse profile on which the incoming model traffic stream is based is varied by using genetic algorithms. This method, which is known per se, is particularly suitable for the present application, iteratively adapting a normalized burst profile.

In another preferred embodiment of the method according to the invention, for each variant of the incoming model traffic flow, a degree of correspondence of the model measured data generated by the latter with the corresponding measured data recorded by the vehicle detector is determined, wherein the variants exceeding a predetermined threshold for the quality measure Bandwidth of possible variation parameters is determined. As a result, in addition to the determination of the characteristic quantities, a measure of the accuracy or reliability of the solution can also be determined and displayed.

In a further advantageous embodiment of the method according to the invention, movement lines and / or holding and / or waiting times of the model vehicles are determined from the optimized incoming model traffic flow by statistical evaluation of the model vehicle movements as traffic information. Thus, if an optimized estimated pulp profile is available, then the resulting simulative simulation of the traffic flow on the road route is also available. The statistical evaluation of the simulated traffic flow makes it possible to determine waiting times and stops of all vehicles, possibly differentiated according to vehicles which have flowed in from the main direction or from a secondary direction. From the shape of the found pulp profile, the Pulehenile the main direction and the bending secondary directions can be assigned. If it is noted in the simulation which vehicles were generated from the main directional part of the pulp profile, their travel profiles can be evaluated separately after the simulation. It is even a distinction of vehicles of the main direction in those who are in the road section possible at green start or during a later time of the green time.

With particular advantage, a quality value for the road route is calculated as traffic information. This may be the so-called "level-of-service" quality score set out in the Road Safety Assessment Manual (HBS). Thus, according to the invention, an automated quality determination for a road route and thus also the proof of the traffic quality of planning measures in the context of a project acceptance is possible.

In another advantageous embodiment of the method according to the invention, an optimal coordination of the traffic light system at the exit cross section to the traffic light system at the entrance cross section is determined. Essential here is the offset of the two signal time schedules to achieve a green wave. If waiting times and stops of the main and secondary direction vehicles are weighted, a recommendation for an optimal shaft position or coordination on this road section can be given via a downstream optimization algorithm; it can also be determined by what percentage the current situation is away from the optimum.

In a further preferred refinement of the method according to the invention, threshold values are specified for holding and / or waiting times and / or coordination deviations, whose overshoots or undershoots are determined during an analysis period and reported after the analysis period as quality analysis. In the evaluation, it is possible to identify from several access roads those in which the controls of the traffic signals should be checked for their quality. By setting thresholds, automated mechanisms can be used to perform a quality check in the background and, for example, generate a quality and abnormality analysis with a report on traffic quality and abnormalities after one day.

In another preferred embodiment of the method according to the invention, the road section has a plurality of lanes, wherein at least one lane at least one lane-related vehicle detector, wherein the inflowing model traffic flow with respect to the temporal distribution of entering into the respective model lanes of the model road route model vehicles varies and is optimized with respect to a match of the respectively generated model measurement data with the corresponding measured by the at least one lane-related vehicle detector measurement data. According to the invention, the method can also be used for more complex node topologies, wherein the estimate may include several lanes per road section. The individual lanes may have none, one or more consecutive vehicle detectors.

The object is further achieved by a traffic computer for determining traffic information for a road section of a road network, which is provided with a program code containing control commands that cause the traffic computer to carry out a method according to one of claims 1 to 16. For this purpose, the traffic computer has correspondingly designed data processing means, interfaces for data input and output as well as a visualization unit for displaying the traffic information.

Furthermore, the invention relates to a machine-readable program code for a traffic computer, which contains control commands that cause the traffic computer to carry out a method according to one of claims 1 to 16.

Finally, the invention also relates to a storage medium having a machine-readable program code stored thereon according to claim 18.

FIG 1

ein Ablaufplan des erfindungsgemäßen Verfahrens in einem Verkehrsrechner für eine Straßenstrecke,

FIG 2

Rohdaten eines Detektors,

FIG 3

Zählwert-Zustand variabler Zeitdauer,

FIG 4

Belegungswert-Zustand variabler Zeitdauer,

FIG 5

Zähl- und Belegungswert-Zustand konstanter Zeitdau- er,

FIG 6

ein normiertes Pulkprofil,

FIG 7

ein Diagramm für den Mittelwert der Abstandsmaße in Abhängigkeit der Versatzzeit

schematisch veranschaulicht sind.Further features and advantages of the invention will become apparent from an embodiment illustrated in the drawings, in which

FIG. 1

a flowchart of the method according to the invention in a traffic computer for a road,

FIG. 2

Raw data of a detector,

FIG. 3

Count state of variable duration,

FIG. 4

Occupancy value state variable duration,

FIG. 5

Count and occupancy value state of constant time duration,

FIG. 6

a normalized bulk profile,

FIG. 7

a diagram for the mean of the distance measures as a function of the offset time

are illustrated schematically.

FIG. 1 shows a road S ₁₂ a road network that connects, for example, two nodes not shown. The road S ₁₂ has an entrance cross-section 1 at the Vorknoten, an exit cross-section 2 at the main node and an intermediate measuring cross-section 3. Through the entrance cross-section 1 flows a the route S ₁₂ querying traffic flow, formed by vehicles F to. The incoming traffic flow is controlled by a traffic signal 10 at the Vorknoten. The traffic signal system 10 has signal transmitters 11 for the main traffic flow and the secondary traffic flows whose signal times are switched according to a signal schedule SP ₁ running in the control device 12. The inflowing traffic flow takes place in vehicle pulse per revolution time of the signal time schedule SP ₁ . In accordance with the resulting traffic flow, detector raw data in the form of count values z _i and occupancy values b _i are detected at the measuring cross-section 3 by a vehicle detector 30, which is designed as an induction loop, for example, at equidistant time intervals i.

According to FIG. 2 be detected as raw data of the vehicle detector 30 so-called edge data, ie those times at which the occupancy state of the vehicle detector 30 of "occupied (value 1)" to "not busy (value 0)" changes and vice versa. At the times t ₁ , t ₂ , t ₃ and t _{4 of} the falling flanks, a vehicle F leaves the detection range of the vehicle detector 30. A time gap h ₁ , h ₂ or h ₃ arises until the next vehicle F enters the detection range of Vehicle detector 30 retracts. The subsequent occupancy time O ₁ , O ₂ or O ₃ then ends at the next falling edge t ₂ , t ₃ or t ₄ . Alternatively to the edge data 30 and every second resolution occupancy states of the vehicle can be used as raw data detector may be used, out of which also the time data t _i, h _i, O _i arise. Each interval [t _n ; t _{n + 1} ], with n = 1, 2, 3, two consecutive vehicles F are according to FIG. 3 a count z (n) = 1 / (t _{n + 1} -t _n ) and according to FIG. 4 assigned an occupancy value b (n) = o _n / (t _{n + 1} - t _n ). These are constant, macroscopic states of variable duration. To FIG. 5 Now the count value (dash-dotted line) and the occupancy value (solid line) are assigned to a discrete time axis with equidistant time intervals i of, for example, one second. The count value of z _i is the number of vehicles per second, in the time interval i, while the occupancy value b _i indicating the holding time per second in the time interval i.

The detector raw data are summarized according to the signal positions circulating for the incoming traffic flow, wherein the cycle time of the signal time schedule SP _{1 can be} for example 60 s or 90 s. At the exit cross-section 2 of the outflowing traffic flow is controlled by a traffic signal 20, which has a signal generator 21 and a controller 22, in which emitted by the signal generator 21 light signals are switched according to an expiring signal time plan SP ₂ .

Typically, the traffic signal systems 10 and 20 at the entrance cross-section 1 and exit cross-section 2 by adjustment the same cycle times of the signal time schedules SP ₁ and SP ₂ coordinated; The signal cycles of the signal time schedules SP ₁ and SP ₂ are offset in time according to the length of the road S ₁₂ and the typical driving speeds on the road S ₁₂ . The choice of the offset time is decisive for the quality of the coordination of the two traffic signal systems 10 and 20, respectively.

In a traffic control computer 40 is now a simulation of the traffic flow of the road section runs along S ₁₂ by means of a traffic model VM off by a Pulkprofil p 'of the incoming traffic flow model _to z' is estimated iteratively. Such a pulp profile p 'is in FIG. 6 shown. The pulp profile p 'indicates over a cycle time of, for example, 90 s, the time course of the proportion of vehicles F, which retract per period of time through the inlet cross-section 1. The entire pulp profile p 'is normalized by dividing by the total number of vehicles F passing the entrance cross section 1 during the circulation time. Over an examination period of, for example, one hour, it can be assumed that the pulse profile p 'is constant for every 40 time intervals or circulation times of 90 s in each case. The optimization with the help of the traffic model VM now follows as follows:

Starting from a Pulkprofil p 'are this imposed for each of the associated real Signalumläufe measured count values z _i multiplicative. This results as input to the transport model VM, the intensity distribution of the influent during a round trip time model for the traffic stream _to '.

By means of the traffic model VM, the movements of model vehicles of the incoming model traffic flow z are now simulated _to 'along a model road route, which generate model measurement data z _i ' or b _i 'when passing a model measurement cross section and which by a Departure light signal controlled model exit cross section. At this point In the traffic model VM, the real signal cycles of the signal time schedule SP ₂ of the traffic light control system 20 controlling the outflow enter. Such traffic models VM are known to those skilled in the art. In addition to macroscopic traffic models, microscopic traffic models aimed at tracking individual model vehicles are in use here. As a result, the traffic model VM provides model measurement data in the form of model counts z _i 'and model occupancy values b _i ', which are now compared to the measurement data z _i and b _i actually generated by the vehicle detector 30 in the respective time intervals i.

The distance measure d _i used is the Euclidean distance between the corresponding real and model-generated points in the fundamental diagram, in which for each time interval i the traffic volume q in vehicles per hour is plotted over the occupancy rate b in percent. The distance measure is now as follows: $${\mathrm{d}}_{\mathrm{i}}\mathrm{=}\sqrt{{\left({\mathrm{q}}_{\mathrm{i}}\mathrm{-}{\mathrm{q}}_{\mathrm{i}}\mathrm{\text{'}}\right)}^{\mathrm{2}}\mathrm{+}{\left({\mathrm{b}}_{\mathrm{i}}\mathrm{-}{\mathrm{b}}_{\mathrm{i}}\mathrm{\text{'}}\right)}^{\mathrm{2}}}$$

From this, an average value d is calculated from all the distance measures d _i for an examination period with N time intervals: $$\mathrm{d}\mathrm{=}\frac{\mathrm{1}}{\mathrm{N}}\mathrm{\cdot }{\displaystyle \underset{\mathrm{i}\mathrm{=}\mathrm{1}}{\overset{\mathrm{N}}{\mathrm{\Sigma }}}}{\mathrm{d}}_{\mathrm{i}}$$

The mean value d of the distance measures d _i is now compared with a threshold value D. As long as the mean value d exceeds the threshold value D, the pulse profile p 'is adapted using genetic algorithms GA and new model measurement data z _i ' or b _i 'is generated by means of the traffic model VM until the mean value d of the distance measures d _{i reaches} the threshold value D reaches or falls below. The iteration procedure can also be aborted if a predefined runtime is exceeded or if the mean value d only changed by small values. In this case, the optimized model traffic flow z _to 'was determined, which best simulates the real measured data z _i or b _i .

According to the invention, it is now assumed that there is a good correspondence between the generated model measured data z _i 'or b _i ' and the corresponding measured data z _i or b _i detected by the vehicle detector 30, so that the traffic information VI can be obtained from the simulated model. Traffic flow, which results from the optimized model traffic flow z _to 'can be determined. By statistical evaluation SA of the model vehicle movements of the optimized incoming model traffic flow z _to ', for example, movement lines, stops and waiting times of the model vehicles are determined. Furthermore, a quality characteristic value for the road S _{12 is} calculated. In addition, the optimal coordination between the traffic signal systems 10 and 20 at the exit section 2 or entrance cross section 1 can be determined.

This shows FIG. 7 a diagram of how the mean d of the distance measures d _i changes as a function of the change in the offset time between the signal time schedules SP ₁ us SP ₂ . The fictitious example assumes a given coordination of 60 s offset time. In the simulation by means of the traffic model VM, the coordination was varied over one revolution in steps of 10 s, whereby a clear minimum of the mean distance d at an offset of 60 s modeled in the traffic model VM is recognized. This shows the sensitivity for the coordination achievable with the method according to the invention even with a relatively short period of time of less than one hour.

Claims (21)

1. Method for ascertaining traffic information (VI) for a section of road (s₁₂) in a road network, wherein the section of road (s₁₂) has an entry cross-section (1) through which a stream of traffic requesting the section of road (s₁₂) enters, an exit cross-section (2) through which a stream of traffic controlled by means of a set of traffic lights (20) exits, and at least one measurement cross-section (3) which is arranged between the entry cross-section (1) and the exit cross-section (2) and on which a vehicle detector (30) captures measurement data (z_i, b_i) produced by passing vehicles (F), wherein the flow of traffic along the section of road (s₁₂) is simulated by means of a traffic model (VM), characterized in that an entering model stream of traffic (z_in') is taken as a basis for producing model measurement data (z_i', b_i') which are associated with the measurement data (z_i, b_i), wherein the entering model stream of traffic (z_in') is varied in terms of the temporal distribution of model vehicles entering the model section of road and is optimized in terms of a match between the respectively produced model measurement data (z_i', b_i') and the corresponding measurement data (z_i, b_i) captured by the vehicle detector (30), and wherein the traffic information (VI) is ascertained from the simulated model flow of traffic which arises from the optimized model stream of traffic (z_in').
2. Method according to Claim 1, wherein, over an examination period (T) which is divided into time intervals (i), a count (z_i) for detected vehicles (F) per time interval (i) and an engagement value (b_i) for the vehicle detector (30) per time interval (i) are captured as measurement data, wherein counts and engagement values (z_i, b_i) are ascertained from raw data from the vehicle detector (30).
3. Method according to Claim 1 or 2, wherein the entering model stream of traffic (z_in') is respectively referenced to a cycle time for a signal schedule (SP₁) running in a set of traffic lights (10) which controls the entering stream of traffic.
4. Method according to Claim 3, wherein the measurement data (z_i, b_i) captured during the time intervals (i) in the examination period (T) are associated with the individual signal cycles of the signal schedule (SP₁) as appropriate.
5. Method according to Claim 4, wherein the model stream of traffic (Z_in' entering during a cycle time is formed by multiplying the sum of the counts (z_i) of detected vehicles (F) that are associated with the cycle by a normalized group profile (p') which indicates a temporal distribution of vehicle components of an entering vehicle group within a cycle time, wherein the entering model stream of traffic (z_in') is varied by varying the underlying group profile (p').
6. Method according to Claim 5, wherein the formation of the model stream of traffic (Z_in' entering in the examination period (T) involves the same group profile (p') being taken as a basis for each cycle time.
7. Method according to one of Claims 1 to 6, wherein the flow of traffic is simulated by reproducing movements by model vehicles in an entering model stream of traffic (z_in') along a model section of road which produce the model measurement data (z_i', b_i') when passing through a model measurement cross-section and which exit through a model exit cross-section which is controlled by traffic lights.
8. Method according to Claim 7, wherein the entering model stream of traffic (z_in') is formed from model vehicles in various vehicle classes, each with an average vehicle length, wherein a composition for the model stream of traffic (z_in') from vehicle classes and the average vehicle lengths thereof are prescribed or varied.
9. Method according to one of Claims 3 to 8, wherein the repeating cycle time or a series of changing cycle times for the signal schedule (SP₁) running in the set of traffic lights (10) which controls the entering stream of traffic is varied, wherein the captured measurement data (z_i, b_i) are associated with the repeating cycle time or with the changing cycle times as appropriate.
10. Method according to one of Claims 1 to 9, wherein for at least one portion of the time intervals (i) a distance (d_i) between the respectively produced model measurement data (z_i', b_i') and the corresponding measurement data (z_i, b_i) captured by the vehicle detector (30) is calculated and the entering model stream of traffic (z_in') is optimized by minimizing a mean value (d) of the distances for the portion of the time intervals (i) in the examination period (T).
11. Method according to one of Claims 5 to 10, wherein the group profile (p') on which the entering model stream of traffic (z_in') is based is varied by applying genetic algorithms.
12. Method according to one of Claims 1 to 11, wherein for each variant of the entering model stream of traffic (z_in') a measure of quality which conveys the degree of the match between the model measurement data (z_i', b_i') produced by said model stream of traffic and the corresponding measurement data (z_i, b_i) captured by the vehicle detector (30) is ascertained, wherein a bandwidth of possible variation parameters is determined from the variants which exceed a prescribed threshold for the measure of quality.
13. Method according to one of Claims 1 to 12, wherein lines of movement and/or halt and/or waiting times for the model vehicles are ascertained from the optimized entering model stream of traffic (z_in') by statistically evaluating (AW) the model vehicle movements as traffic information (VI).
14. Method according to Claim 13, wherein the traffic information (VI) calculated is a quality parameter for the section of road (s₁₂).
15. Method according to Claim 13 or 14, wherein optimum coordination of the set of traffic lights (20) on the exit cross-section (2) with the set of traffic lights (10) on the entry cross-section (1) is ascertained.
16. Method according to Claim 15, wherein a discrepancy between current coordination and the optimum coordination of the sets of traffic lights (10, 20) is determined.
17. Method according to one of Claims 13 to 16, wherein halt and/or waiting times and/or coordination discrepancies have threshold values prescribed for them, the exceeding and undershooting of which are established during an analysis period and are reported as a quality analysis when the analysis period has elapsed.
18. Method according to one of Claims 1 to 17, wherein the section of road has a plurality of lanes, wherein at least one lane has at least one lane-related vehicle detector, wherein the entering model stream of traffic is varied in terms of the temporal distribution of model vehicles entering the respective model lanes on the model section of road and is optimized in terms of a match between the respectively produced model measurement data and the corresponding measurement data captured by the at least one lane-related vehicle detector.
19. Traffic computer (40) for ascertaining traffic information (VI) for a section of road (s₁₂) in a road network, which traffic computer is provided with a program code which contains control commands which prompt the traffic computer (40) to carry out a method according to one of Claims 1 to 18.
20. Machine-readable program code for a traffic computer (40) which contains control commands which prompt the traffic computer (40) to carry out a method according to one of Claims 1 to 18.
21. Storage medium having a machine-readable program code according to Claim 20 stored thereon.

Images