EP0740280B1 - Disturbance detection method for road traffic - Google Patents

Abstract

The system has a measuring point at both the entry and exit of a monitored sector, each providing measured data relating to the number and velocity of the vehicles passing through the sector, evaluated for determining the traffic level and the traffic speed. An expected traffic level is calculated from the number of vehicles entering the sector and compared with the actual traffic level at the end of the monitored sector, to determine the number of vehicles remaining within the latter, with a traffic hold-up when this number exceeds a threshold value.

Description

The invention relates to a method for fault detection in road traffic within of a sector to be monitored, with one measuring cross-section at the beginning of the sector and sector end as measured data the number and the speed of the Measuring cross sections passing vehicles continuously recorded, while finite, consecutively numbered measuring intervals collected and cyclically to average values the traffic volume and the speed are compressed and then evaluated, wherein each measurement cross section comprises all lanes that can be driven in one direction of travel.

Due to the constantly increasing road traffic, it is becoming more collective Control on multi-lane roads increasingly using influencing systems Variable message signs are used to control different methods can be used for fault detection in road traffic. The motivation for Investment in such systems and the development of associated procedures for fault detection is based on the fact that the variable message signs are feasible Warnings of the following traffic moving towards fault locations the risk in the form of secondary accidents can be significantly reduced. Significant is, in particular, the most reliable and rapid detection possible Disturbance, i.e. a reduction in the critical time between entry and protection a disturbance.

Traffic is currently recorded using measurement methods that are local Record measurement data from individual vehicles on so-called measurement cross-sections. It is customary to monitor individual, in each case by a measuring cross section on Beginning and end of limited sectors, i.e. Sections of a traffic line. In traffic engineering is the size "traffic volume" as the number of registered vehicles per Defines the time unit, from which, for example, [vehicles / min] results as its unit.

A method of the type described in the introduction is described in the publication "Automatic Accident detection on highways using fuzzy logic ", published in 1994 by the Siemens AG, Munich, plant engineering division, road traffic engineering division, described. The process essentially consists of the two Main parts of data preparation and fuzzy decision logic. In the data processing section are calculated from the average values averaged over a finite measurement interval the speed and traffic volume essentially the three as Basic parameters of accident detection are the indicators called "speed density difference" (VK-Diff), "trend factor" and "trend of traffic volume" calculated. These three indicators serve as input variables for the fuzzy decision logic, which provides an accident probability as an output variable. The fuzzy decision logic has a three-stage structure and consists of the three modules "Bulk detection", "Pre-accident investigation" and the actual core, the so-called "Accident detection". This core module provides a so-called output variable Accident probability, which is given in a range from 0 to 100%.

Due to the very complicated, but nevertheless makeshift definition of the The indicator "speed density difference" only works in known method certain traffic situations reliably without time delay. Especially in the case of faults that occur in the rear area of the monitored sector, the reliability is fault detection poor. Also with the approach, as a decision system for the combination of the indicators used Using a fuzzy decision logic could overcome these fundamental shortcomings of the procedure not resolved, but only weakened to a certain extent become.

Furthermore, methods are generally known in which the detection of a fault is essentially derived from the congestion of a measurement cross-section, i.e. a malfunction is detected when vehicles are already on a measurement cross-section stand or pass this at a very slow speed. These so-called Local methods, however, have the serious disadvantage that depending on the distance between the disturbance and the measurement cross section until the disturbance is detected valuable time is lost. Especially with a very long length of one to be monitored Sector, which is for cost reasons because of the then reduced number of Measuring cross sections should always be aimed at, elapses between the occurrence of a fault in short distance before a measurement cross-section and the time at which a Reached the next upstream measuring cross section, a very great period. As vehicles approaching the end of the traffic jam from behind during this system-related dead time cannot be warned, there is a considerable risk that rear-end collisions occur as a consequence of the primary fault.

In a brochure from ave Verkehrs- und Informationstechnik GmbH, Aachen, with The title "AVE - the complete solution for modern traffic management" will also be a method for fault detection in traffic described on the Determination and analysis of route-related traffic parameters based. With this on The so - called method of "pattern recognition" building approach the measured cross-section signals of the individual vehicles evaluated by characteristic Sample features are extracted, analyzed and standardized. The so prepared measurement data (sample features) are each downstream to the next the assigned measuring station assigned to the measuring cross-section and there continuously compared with the analyzed measurement data of the output cross section. With help suitable correlation methods, so-called link system functions are formed, from which the route-related traffic parameters "travel time" (and thus the average cruising speed) of the vehicle collectives under consideration in the route section between input and output cross-section and the "traffic density" in the section between the input and output measurement cross-section to let.

However, such methods place very high demands on the measuring technology and to the data transmission capacities, so that in addition to the data transmission between the route stations assigned to the respective measuring cross sections and one central evaluation point also direct connections between the neighboring route stations among themselves are required.

With the measuring, transmission and evaluation devices currently in operation to carry out the methods according to the prior art, these are Hardware requirements are usually not met, so very expensive retrofits would be necessary to use such route-related procedures can. Furthermore, all of them are currently in use or still under development current techniques for traffic data acquisition (i.e. for measuring local speeds, the traffic volume and the traffic composition), which in the essential of induction loops, microwave sensors (radar), laser scanners and video cameras with subsequent evaluation of the recorded data using special computer programs are represented, with relatively large uncertainties afflicted. The detection of cars in particular is prone to errors. In the Individual vehicles, e.g. due to lane changes in the area of measuring cross sections, not recorded. The failure is more lane-related or even cross-sectional detection devices possible. For this reason deliver methods based on route-related traffic data acquisition, in which the reliability of vehicle detection is of central importance, only unsatisfactory results.

Known methods of fault detection evaluate those that occur in the data center and average values determined from the recorded measurement data by cyclic compression in real-time data processing. However, the data situation is due of the vehicle detection, which is never 100% reliable.

To make decisions under uncertainty or based on uncertain data exist different approaches, of which in traffic engineering in addition to that already mentioned above described method of fuzzy logic according to the prior art also methods for signal conditioning and statistical methods.

As a disadvantage in signal conditioning, mostly in the form of an exponential Smoothing is carried out, it appears that for each measuring cross section and for all individual data parameters must be managed and set, which is a very high parameterization effort results. Statistical methods, however, indicate that Disadvantage that they because of the missing population in the individual decision are not significant.

The invention has for its object to propose a method with which independently of the traffic situation with as little time loss as possible and with little Data processing and measurement technology expenses Disruptions in road traffic detected can be.

• daß aus den am Sektoranfang ermittelten Durchschnittswerten der Verkehrsstärke und der Geschwindigkeit der Fahrzeuge, der Länge des Sektors, einer Annahme für die zeitliche Verteilung der Fahrzeuge innerhalb des Meßintervalls und einer Annahme für den Verlauf der Geschwindigkeit der Fahrzeuge beim Durchfahren des Sektors zyklisch ein Prognosewert der Verkehrsstärke am Sektorende berechnet wird, wobei
• a) aus der Länge des Sektors und der Annahme für den Verlauf der Geschwindigkeit der Fahrzeuge beim Durchfahren des Sektors eine Durchfahrtszeit ermittelt wird,
• b) aus dieser Durchfahrtszeit die Nummern derjenigen Meßintervalle bestimmt werden, in denen die im jeweiligen Meßintervall am Sektoranfang erfaßten Fahrzeuge den Meßquerschnitt am Sektorende passieren,
• c) unter Berücksichtigung der Annahme für die zeitliche Verteilung der Fahrzeuge Anteilsfaktoren für die Aufteilung des Durchschnittswertes der Verkehrsstärke auf die gemäß b) bestimmten Meßintervalle ermittelt werden und
• d) der Prognosewert der Verkehrsstärke durch Summierung der Produkte aus den Anteilsfaktoren und den zugehörigen Verkehrsstärken berechnet wird, wobei über alle vorangegangenen Meßintervalle und das aktuelle Meßintervall summiert wird;
• daß zyklisch ein Vergleich des Prognosewerts der Verkehrsstärke und des aus den am Meßquerschnitt am Sektorende erfaßten Meßdaten ermittelten Durchschnittswerts der Verkehrsstärke am Sektorende durchgeführt und die jeweilige Differenzverkehrsstärke bestimmt wird,
• daß eine zyklenübergreifende Summierung der Werte der Differenzverkehrsstärke durchgeführt und fortlaufend die Anzahl der im zu überwachenden Sektor zusätzlich verbleibenden Fahrzeuge bestimmt wird sowie
• daß bei Überschreitung eines Grenzwertes für die Anzahl der zusätzlich im Sektor verbleibenden Fahrzeuge eine Störungsmeldung ausgelöst wird.

Starting from a method of the type described at the outset, this object is achieved by

• that from the average values of the traffic volume and the speed of the vehicles, the length of the sector, an assumption for the temporal distribution of the vehicles within the measurement interval and an assumption for the course of the speed of the vehicles as they pass through the sector, cyclically a forecast value of the traffic intensity is calculated at the end of the sector, where
• a) a transit time is determined from the length of the sector and the assumption for the course of the speed of the vehicles when passing through the sector,
• b) from this transit time, the numbers of those measuring intervals are determined in which the vehicles recorded in the respective measuring interval at the beginning of the sector pass the measuring cross section at the end of the sector,
• c) taking into account the assumption for the temporal distribution of the vehicles, proportion factors for the distribution of the average value of the traffic volume over the measurement intervals determined in accordance with b) are determined, and
• d) the forecast value of the traffic volume is calculated by summing the products from the proportion factors and the associated traffic volumes, totaling over all previous measurement intervals and the current measurement interval;
• that a cyclical comparison of the forecast value of the traffic volume and the average value of the traffic volume determined at the end of the sector on the measurement cross section is carried out and the respective differential traffic volume is determined,
• that a cross-cycle summation of the values of the differential traffic strength is carried out and the number of additional vehicles remaining in the sector to be monitored is continuously determined and
• that a fault message is triggered if a limit value for the number of additional vehicles remaining in the sector is exceeded.

Due to the continuous, dynamic balancing of the vehicles entering the monitored sector can be very reliable and fault detection largely independent of the respective traffic situation achieve. In a trouble-free traffic situation, the can in the monitored sector incoming traffic taking into account the average speeds, of driving behavior, sector length and a distribution assumption find at the sector exit.

When determining the forecast value of traffic volume for the sector end passing vehicles is taken into account according to the inventive method, what proportion of the vehicles measured during a detection cycle the monitored Sector in the same cycle, only in the following cycle or in one the later cycle. For this purpose, the proportion factors are determined, with the help of the vehicle collective of an acquisition cycle recorded at the beginning of the sector divided between the two cycles during which its shares Sector is expected to be left again.

The disadvantages of methods that use the measurement data of the measurement cross sections at the beginning of the sector and only consider the sector end at the same time and regardless of the route topology and with which only one, even when constructing complex analysis criteria relatively unreliable fault detection is possible by the invention Procedure safely avoided. The data processing effort is included the method of the invention at a very low level, the most likely comparable to that in the case of accident detection methods based on local measured variables is. Their serious disadvantages that an accident detection only when congestion a measurement cross section is possible, but is avoided by the invention, because the dynamic balancing possible disruptions very quickly and almost independently are detectable from their location in the monitored sector.

In comparison to methods known from the prior art, in which a Determination and analysis of route-related traffic parameters is carried out and in which therefore because of the correlation of vehicle groups at the beginning of the sector and Sector end using a pattern identifier an enormously high data processing effort this effort is very low in the method according to the invention. The Complex indicators, such as the speed density difference (VK-Diff) or special trend factors is not necessary. This also applies to the Analysis of the measurement data for the possible presence of a vehicle group, which is the result for route-related procedures without its detection and adequate consideration would falsify.

A comparison with known methods, e.g. the procedure based on the three indicators "Speed density difference", "trend factor", "trend of traffic volume", clearly shows that traffic disturbances more reliable by means of the method according to the invention and can be recognized more quickly. With it are also very in long sectors high security in fault detection achievable, and the results will be very little affected by the location of the disruption within the sector, leading to the known method does not apply.

The method according to the invention can also be used very economically since one additional direct connection between those assigned to the measuring cross sections Line stations, as used in methods based on route-related traffic parameters is absolutely necessary, can be avoided and because of the high reliability of the accident detection the distances between two measuring cross sections, i.e. the sector length can be very large.

The calculation of the number of additional remaining in the monitored sector Vehicles on which the fault report is based can be identified by the following Formulas illustrate:

1. Calculation of transit times

TD_i,z:

die Zeit, die Fahrzeuge zum Durchfahren des Sektors benötigen

v(s):

angenommene Geschwindigkeit eines Fahrzeugs am Ort s innerhalb des Sektors

v_i,z

über den Streckenverlauf des Sektors S_i gemittelte Geschwindigkeit

S_i:

Länge des Sektors zwischen den Meßquerschnitten i und i + 1

s:

Längenvariable mit s ∈ 0... S_i

und den Indizes

i:

Index für Meßquerschnitte, in Fahrtrichtung aufsteigend

z:

Index Index für Erfassungszyklen, 1 = aktueller Zyklus, in die Vergangenheit aufsteigend

The transit time TD _{i, z} that a vehicle needs to pass through the monitored sector is: TD i, e.g. = S i / v i, e.g. wherein: v i, e.g. = (1 / S i ) * ∫ s = 0 .. Si v (s) ds With

TD _{i, z} :

the time it takes vehicles to drive through the sector

v (s):

assumed speed of a vehicle at location s within the sector

v _{i, z}

Speed averaged over the course of the sector S _i

S _i :

Length of the sector between the cross sections i and i + 1

s:

Length variable with s ∈ 0 ... S _i

and the indices

i:

Index for measuring cross-sections, ascending in the direction of travel

z:

Index Index for acquisition cycles, 1 = current cycle, ascending in the past

vm_i,z:

Gemessener Durchschnittswert der Geschwindigkeit der im Zyklus z am Meßquerschnitt i erfaßten Fahrzeuge

For v (s) an assumption must be made that should take into account the current traffic condition and the route topology. For example, with steep inclines - especially for weakly motorized vehicles - a drop in speed when driving through is to be expected. The same applies to very winding routes, the opposite for gradients. The dependence of v (s) on the traffic condition can be determined by the measured speeds vm _{i, z} , the route topology is taken into account by a correction factor. In it

vm _{i, z} :

Measured average value of the speed of the vehicles recorded in cycle z on the measurement cross-section i

The assumption is a constant driving speed and no special topology of the route (no curves, inclines, descents), then v (s) = const = vm _{i, z can be} assumed. Other assumptions are possible.

2. Allocation of the registered vehicles

n_i,z:

der zum aktuellen Zyklus gehörige Anteilsfaktor; dieser Anteil an den insgesamt während des aktuellen Zyklus in den Sektor eingefahrenen Fahrzeuge verläßt nach der Durchfahrtszeit TD_i,z den Sektor in demselben Zyklus z am Meßquerschnitt i+1 wieder.

T:

Länge des Erfassungszyklus

The number of vehicles on the measurement cross-section i is counted within the detection cycle z. The transit time TD _{i, z} can be used to determine when vehicles which enter the sector at the measurement cross section i in the current time cycle can be expected at the measurement cross section i + 1. With a discrete view of the time recording, it must be determined in which recording cycle which proportion of the vehicles leaves the sector. The proportion factors n _{i, z} result for the current cycle z = 1 and the past cycles z> 1 from: n i, e.g. = Max (0, (z - TD i, e.g. / T)) for TD i, e.g. <z * T n i, e.g. = Max (0, (1-Max (0, (z - TD i, e.g. / T))) for TD i, 1 ≥ z * T With

n _{i, z} :

the proportion factor associated with the current cycle; after the passage time TD _{i, z,} this portion of the total number of vehicles that have entered the sector during the current cycle leaves the sector again in the same cycle z at the measurement cross section i + 1.

T:

Length of the acquisition cycle

3. Determination of the forecast value of the traffic volume

From the counted vehicles, an assumption about the time distribution of the detected vehicles over the detection cycle T and the determined Proportional factors of the vehicles can now be the forecast value of the traffic volume determine.

A distribution assumption is made for the vehicles recorded during the cycle x = P (t) based on the time t ∈ 0 ... T. Usually can be worked with a simple uniform distribution of the vehicles.

PQ_i+1:

Der aktuelle Prognosewert (z=1) der Verkehrsstärke am Meßquerschnitt i+1

Q_i,z:

Anzahl der am Meßquerschnitt i im Erfassungszyklus z gezählten Fahrzeuge

P(n_i,z):

der Verteilungsannahme ermittelte Anteil an Fahrzeugen Q_i,z, die im Zyklus z am Meßquerschnitt i+1 ausfahren, z.B. ist bei Annahme einer Gleichverteilung P(n_i,z) = n_i,z

The forecast value of the traffic volume is given as PQ i + 1.1 = Σ P (n i, e.g. ) * Q i, e.g. (z = 1..Z)
In it

PQ _{i + 1} :

The current forecast value (z = 1) of the traffic volume at the measurement cross section i + 1

Q _{i, z} :

Number of vehicles counted at the measurement cross section i in the acquisition cycle z

P (n _{i, z} ):

the distribution assumption determined proportion of vehicles Q _{i, z} that extend in the cycle z on the measurement cross section i + 1, for example, assuming a uniform distribution P (n _{i, z} ) = n _{i, z}

4. Determination of differential traffic volume

DQ_i,z:

Differenzverkehrsstärke am Meßquerschnitt i im Erfassungszyklus z

PQ_i,z:

Prognosewert der Verkehrsstärke am Meßquerschnitt i im Erfassungszyklus z

Q_i,z:

Anzahl der am Meßquerschnitt i im Erfassungszyklus z gezählten Fahrzeuge

The differential traffic strength DQ _{i, z is} determined by: DQ i, e.g. = PQ i, e.g. -Q i, e.g. With

DQ _{i, z} :

Difference traffic strength at the measurement cross section i in the detection cycle z

PQ _{i, z} :

Forecast value of the traffic volume at the measurement cross section i in the detection cycle z

Q _{i, z:}

Number of vehicles counted at the measurement cross section i in the acquisition cycle z

5. Determination of excess vehicles

BDQ_i:

Anzahl an Fahrzeugen im Sektor zwischen den Meßquerschnitten i-1 und i

DQ_i,z:

Differenzverkehrsstärke am Meßquerschnitt i im Erfassungszyklus z

Z:

Anzahl der Zyklen, innerhalb der die Summierung durchgeführt wird

The number of BDQi of the additional vehicles remaining in the monitored sector is determined from: BDQ i = Σ DQ i, e.g. (z = 1 ... Z)
or recursively BDQ i = BDQ i, old + DQ i, e.g. With:

BDQ _i :

Number of vehicles in the sector between the measurement cross-sections i-1 and i

DQ _{i, z} :

Difference traffic strength at the measurement cross section i in the detection cycle z

Z:

Number of cycles within which the summation is carried out

Extensions through a calibration to 0 (a negative number of vehicles is only conceivable due to start-up behavior or acquisition problems) or in recursive Damping procedures prevent the summation of Measurement errors.

If the size BDQ _i exceeds a certain limit value (for example, depending on the traffic condition), a fault message is triggered. The fault message can vary depending on the severity of the fault, ie the level of BDQ _i , in order to prompt the following vehicles to behave appropriately in the special case.

A particularly advantageous embodiment of the invention is that the calculation the forecast value of the traffic volume separately for each lane Direction of travel is carried out and a cyclical comparison between the sum of the Forecast values of all lanes in one direction and the one on the measurement cross-section on Sector end certain value of traffic volume is carried out. The reliability of the forecast value can be increased with the help of this individual consideration of the lanes because there are more precise assumptions for driving behavior for a single lane and the time distribution of the vehicles can be taken. Individual forecasts for each lane with specific driving behavior (right lane: Truck traffic, left lane (overtaking traffic) and subsequent summation thus deliver better results than a sum formation before the forecast leads to a loss of information.

According to a further embodiment of the method according to the invention, it is proposed that that fuzzy logic is used to trigger the fault message, in addition to the number of vehicles remaining in the sector at least an input variable describing the traffic condition is used.

In this context, traffic conditions include (possibly over several cycles average) speed of vehicles, traffic volume and traffic density (= Traffic volume / speed), the standard deviation of the speed (as Measure of the "unrest" in the traffic flow) as well as topological or meteorological variables to understand, being in the definition of the descriptive of the traffic condition Input size does not necessarily fall back on all of the above sizes must become. As an advantage of taking the traffic condition into account when triggering the fault report results in a significant increase in the reliability of the Procedure. In this way, a particularly simple adaptation of the invention Perform the procedure at different locations and conditions, since there is no need for time-consuming setting and calibration of rigid limit values.

The same effect of an increase in reliability also occurs if in another Refinement of the method according to the invention, the describable uncertainty, fuzzy logic takes into account when errors are detected in a measurement cross-section and is therefore used when a fault message is triggered.

A further development of the method according to the invention also consists in that the fuzzy logic is supplied with an output variable that describes the type of fault.

In this way, depending on the severity of the fault, different fault reports can be made spend (for example, "slow-moving traffic" or "complete standstill" of vehicles). The traffic moving towards a fault location can thus effectively warned and prompted for appropriate driving behavior.

It is also particularly advantageous if the course of the speed of the Vehicles when driving through the sector using fuzzy logic is described.

As a result, when determining the forecast value of the traffic volume on Sector end of the describable uncertainties, how the drivers of the vehicles their Behavioral scope in their driving style, depending on the traffic situation exploit, be taken into account in an advantageous manner. The effect is one further increase in the reliability of the method according to the invention.

Developing the invention further, it is provided that only the Traffic volume and the speed of the trucks evaluated for fault detection provided that the share of trucks in the total traffic volume is a certain one Limit exceeds.

Experience has shown that the speeds of the trucks are subject to significantly lower fluctuations than those of the cars, which means that this measure can further increase the reliability of accident detection. With the introduction of the future EU speed controller for trucks, their
Maximum speed limited to a maximum of 88 km / h, an intensification of this positive effect is expected. Furthermore, compared to cars, trucks are recorded with much greater certainty using the detection technology that is common today. At times when the truck share of the total traffic volume falls below a limit value required for reliable fault detection, it is sensible to switch over to the recording of all vehicles, ie both the car and the truck.

Furthermore, it is particularly advantageous if those detected at the sector end of a sector Measurement data simultaneously as measurement data at the beginning of the sector of a subsequent sector be used.

In this way, sectoral observation can be carried out in a simple and very inexpensive way to fully monitor an entire of several Extend sectors composite line. Each measurement cross section represents at the same time the measuring cross-section at the sector end of sector i and at the beginning of the sector sector i + 1.

Claims (8)

1. A procedure for identifying hold-ups in road traffic within a sector to be monitored, wherein at each measuring cross-section at the sector start and sector end the number and the speed of the vehicles passing through the measurement cross-sections are continuously recorded as measured data, whilst finite, continuously numbered measuring intervals are collected and are cyclically compressed to form mean values of the traffic density and of the speed and are then evaluated, wherein each measuring cross-section comprises all the traffic lanes which can be used in one direction of travel, characterised in that

   a prognosis value for the traffic density at the sector end is calculated cyclically from the mean values of the traffic density and of the speed of the vehicles as determined at the sector start, from the length of the sector, from an assumption for the time-dependent distribution of the vehicles over the measuring interval and from an assumption for the progression of the speed the vehicles whilst travelling through the sector, wherein

   a) a time of passage is determined from the length of the sector and from the assumption for the progression of the speed of the vehicles whilst travelling through the sector,

   b) from this time of passage the numbers of those measuring intervals are determined in which the vehicles which are recorded in the measuring interval in question at the sector start pass through the measuring cross-section at the sector end,

   c) proportioning factors for apportioning the mean value of the traffic density to the measuring intervals determined as in b) are determined taking into consideration the assumption for the time-dependent distribution of vehicles, and

   d) the prognosis value for the relevant measuring interval of the traffic density is calculated by summation of the products of the proportioning factors and the associated traffic densities, wherein the summation is effected over all the preceding measuring intervals and the current measuring interval;

   that a comparison is made cyclically between the prognosis value of the traffic density and the mean value of the traffic density at the sector end as determined from the measured data recorded at the measuring cross-section at the sector end, and the differential traffic density in question is determined,

   that a summation which covers the cycles is effected of the values of the differential traffic density and the number of vehicles which additionally remain in the sector to be monitored is continuously determined, and

   that if a limiting value for the number of vehicles which additionally remain in the sector is exceeded a hold-up message is released.
2. A procedure according to claim 1, characterised in that the calculation of the prognosis value of the traffic density is performed separately for each traffic lane of a direction of travel and a cyclic comparison is made between the sum of the prognosis values of all the traffic lanes of a direction of travel and the value of the traffic density as determined at the measuring cross-section at the sector end.
3. A procedure according to claims 1 or 2, characterised in that fuzzy logic is employed for the release of the hold-up message, in which fuzzy logic at least one input quantity which describes the traffic situation is used in addition to the number of vehicles which additionally remain in the sector.
4. A procedure according to claim 3, characterised in that describable uncertainties due to faulty determinations on a measuring cross-section are taken into consideration in the fuzzy logic.
5. A procedure according to claims 3 or 4, characterised in that an output quantity which gives the type of hold-up is supplied by the fuzzy logic.
6. A procedure according to any one of claims 1 to 5, characterised in that the progression of the speed of the vehicles whilst travelling through the sector is described with the aid of fuzzy logic.
7. A procedure according to any one of claims 1 to 6, characterised in that it is merely the traffic density and the speed of heavy goods vehicles which are evaluated as measured data for the identification of a hold-up, provided that the proportion of heavy goods vehicles in relation to the total traffic density exceeds a defined limiting value.
8. A procedure according to any one of claims 1 to 7, characterised in that the measured data recorded at the sector end of a sector are used simultaneously as the measured data at the sector start of a following sector.