EP1276085B1 - Method for determining a traffic jam index and for determining tailback lengths - Google Patents

Abstract

The method is for determining a value characterizing blocking at an operating station for dispatching individual units, where the station operates in a stop-go mode whereby unit movement is stopped or units are allowed through. The values is determined used a detector for measuring the filling time between the beginning of a block or stop phase and the continuous loading of the detector and comparison with a reference filling time. The characteristic value is set at one value when the reference filling time is exceeded and another otherwise. <??>Independent claims are also made for methods for determining the filling time requirement and the backwash length.

Description

The present invention relates to a method for determining a congestion index δ and the resulting self-calibrating methods for estimating backflow lengths at operator stations for handling individually moved units, such as Light signal systems or locks, with a detector in front of it. The variables determined in this way and characteristic values derived from them can be used to control the Light signal systems or locks used or to display the traffic condition be used in higher-level institutions.

State of the art

An important concern of road traffic technology is the determination of backflow lengths on traffic signal systems in order to obtain information about the traffic flow. Knowing the backflow lengths can also be used to control the signal systems (Bernhard Friedrich, Methods and Potentials of Adaptive Methods for Traffic Signal Control, Road Traffic Engineering 9/1996). According to Joos Bernhard, Thomas Riedel, Detection of traffic jams with short loop detectors, road traffic technology 7/1999, traffic jams in traffic light systems can only be recognized or calculated between the stop line and the detector. The same applies to congestion at any operator station for the handling of individually moving units with alternating blocking and passage phases.

A major disadvantage of this known method is that there are no jam lengths to be able to determine the greater than the distance between the operator station and Are detector.

The object of the invention is therefore to provide a method with a determination of the backflow length at operator stations for dispatch individually moving units not only between operator station and detector is enabled to with the help of this backflow length or derived values such as Waiting times to control a traffic light system or lock or traffic conditions in parent institutions.

Description of the invention

This problem is solved by a method for determining a congestion index δ according to claim 1, with a simple determination of the backflow length can be done. In addition, this congestion number can also be used for other the system control relevant parameters, such as the saturation time requirement, determine. Procedure for determining the backflow length using the congestion index are the subject of claims 4 and 16.

In particular, the present invention provides a method for determining a congestion index δ at operator stations ready for handling individually moving units, whereby Each check-in phase consists of a blocking and a pass phase and is in front the operator station is located by measuring the time (filling time) between Blocking begins or a time linked to the beginning of blocking and permanent occupancy of the detector and subsequent comparison with a reference filling time, where δ if the reference filling time is exceeded, a first value and otherwise a second value is assigned.

In addition to the start of the lock, one can also start at the start of the filling time Transition time before the start of the blocking phase coupled time can be selected. at The yellow phase could be considered as a transition period for light signals.

If the reference filling time is undershot, the distance between the operator station becomes and detector filled up faster than in the reference time, so you can avoid a traffic jam out. Otherwise the units are in free flow.

The reference filling time can be taken, for example, from simulative tests or empirical studies. The reference filling time is advantageously dependent on the geometry of the inflow area, for example on the distance between Detector and operator station, the track width, etc., and / or the pass time of the Operator station selected.

With the congestion index δ determined in the manner described above, a Most relevant for a throughput optimization or a traffic status display Determine parameters.

A first method for estimating the backflow length L and _n with the aid of the congestion number determined according to the invention in the nth handling phase is based on the assumption that L and _n as a linear function of a smoothed congestion number δ and _n , which is derived from the congestion number δ _n taking into account the ( n -1) th smoothed congestion index δ and _{n -1 is} determined, there is: L n ( δ n ) = m δ n . where δ and _n can no longer take only two but several values. With a given m , the backflow length for given δ and _n results from equation (1). The congestion figure is smoothed to avoid large jumps in the congestion figure from one handling phase to the next.

This method is characterized in that for determining the backflow length no speed measurements are required.

The slope is advantageously readjusted in every nth handling phase. To do this, determine the traffic volume q _n . This results, for example, from an estimate or from the measured number of units that pass the detector during the nth handling phase. The traffic volume can be used to calculate how many units were available at least before the operator station during the n- th blocking phase; a lower bound L 0 / n for the backflow length is thus obtained. On the other hand, the backflow length function of the previous clearance step L and _{n -1} (δ and _n ) = m _{n -1} δ and _n with δ and _n and a suitably chosen m _{n -1 gives} an estimate of the actual backflow length in the current clearance phase. By comparing L 0 / n and L and _{n -1} (δ and _n ) one can calibrate m _n and thus L and _n .

The slope of the ( n -1) th clearance phase is advantageously obtained by recursively using the method just described with suitable initial values for δ and ₀ and m ₀ . This method is therefore self-calibrating.

The congestion index is preferably smoothed by forming a convex combination of the current congestion index and the smoothed congestion index of the previous clearance: δ n = αδ n + (1 - α) δ n -1 , α ∈ [0,1].

The traffic volume q _n is preferably measured with the detector located in front of the operator station.

In an advantageous embodiment, the lower limit of the accumulation length L 0 / n is given as a linear function of q _n , since this simple form is a good approximation. The slope of this straight line preferably depends on the time in which the detector is permanently occupied during a section of the dispatch phase. If this dependency is taken into account, the correspondence with the real data improves.

It is advantageous to change the slope m _n in the nth step only if either δ _n has assumed the second value and L 0 / n > L and _{n -1} ( δ and _n ) = m _{n -1} δ and _n or if δ _{n has taken} the first value and L 0 / n < L and _{n -1} ( δ and _n ) = m _{n -1} δ and _n . In the first case, δ _n indicates a traffic jam at a distance of at least L 0 / n from the operator station, on the other hand the estimate of the traffic jam length L and _{n -1} ( δ and _n ) is less than L 0 / n . In the second case, δ _n does not indicate a congestion of length L 0 / n , but according to the estimate L and _{n -1} ( δ and _n ), the congestion is even longer than L 0 / n . In both cases, a calibration of the slope m _n is therefore appropriate. If, on the other hand, the value of the congestion index and the estimated accumulation length do not contradict each other, the slope is maintained: m _n = m _n-1 .

A smooth accumulation length L ' _n can be used to adjust the slope m _n , which results as a combination of L 0 / n and L and _n -1 ( δ and _n ): L ' n = βL 0 n ( q n ) + (1-β) L n -1 ( δ n ) β> 0th

The congestion index determined according to the inventive method described above δ can also be used to determine the saturation time requirement; in this connection is the average time required for a unit with saturated (no longer free) flow during the pass phase. The saturation time requirement is a measure of the performance of the operator station. On the other hand, can it can also be used to estimate the backlog length using a queue model.

berechnet werden, wobei t g / n die Durchlaßzeit im n-ten Abfertigungsschritt bezeichnet.To determine the saturation time requirement t B / n in the nth handling step, the congestion index δ is first determined using the method according to the invention and the traffic volume q _{n is} measured or estimated. The saturation time requirement can then be exceeded using a suitable initial condition for t B / 0

can be calculated, where t g / n denotes the pass time in the nth dispatch step.

To avoid large changes in the saturation time requirement from a handling step to the next, you only have a predetermined maximum change Δ t B / max> can preferably in each step 0 the saturation time requirement to. If the t B / n obtained from equation (4) is one of the inequalities Δ t B : = t B n - t B n -1 > Δ t B Max or .delta.t B <-Δ t B Max a modified saturation time requirement t and B / n is advantageously met t B n = t B n -1 + Δ t B Max or t B n = t B n -1 - Δ B Max calculated.

It is advantageous to measure the traffic volume q _n with the detector in front of the operator station.

As an alternative to the method according to the invention described above, the backflow length can be determined with the aid of a queue model which, as parameters to be calibrated, contains a model's own saturation time requirement τ B / n with a suitably chosen initial value. Such a procedure can consist of the following steps in every nth handling operation:

First, the actual saturation time requirement t B / n is determined according to the inventive method described above. If there is a change from the saturation requirement value of the last dispatch phase by Δt ^B , the model's own saturation requirement value τ B / n becomes τ B n = τ B n -1 + c d Δ t B adapted, where c _d denotes a suitably chosen damping constant. In particular, the model's own saturation demand value is included τ B n = τ B n -1 + c d SNG (Δ t B ) Min {| Δ t B |, Δ t B Max } adjusted if only a maximum change allowed by Δ t B / max for the actual saturation demand value, where sgn (Δ t ^B), the sign of At ^B, respectively. A lower bound for the length of traffic L 0 / n is calculated from the traffic volume. With these variables, a queue model is used to calculate a first estimate of the backflow length L " _n . Then L" _n and L 0 / n are compared, analogous to the above method for backflow length estimation. If L " _n > L 0 / n and δ _n has assumed the first value or if L " / n < L 0 / n and δ _n has assumed the second value, the model's own saturation time requirement must be modified. With the calibrated model saturation time requirement, a calibrated estimate of the backlog length is then calculated with the queue model.

This method is characterized in that for determining the backflow length no speed measurements are required.

Furthermore, disturbances in the discharge and in the queue model can advantageously be taken into account a correspondingly modified traffic volume can be used.

In a favorable embodiment of the interference compensation, q _{n is} only modified if it is smaller than the second largest value max _10.2 ( q ) of the last ten q values. In this case, one selects a time interval during the dispatch phase to calculate the interference compensation and counts predetermined, shorter time intervals, for example the full seconds, in which the detector is permanently occupied in the entire interval. The entire interval preferably begins a few seconds after the start of the pass phase and ends a few seconds after the end of the pass phase. If the number obtained in this way is divided by the length of the entire interval, the degree of occupancy b ∈ [0.1] of the detector is obtained. If b falls below a lower limit u , the value 0 is assigned to a fault code s. If b exceeds an upper bound o, s is assigned the value 1. If u ≤ b ≤ o , s results as s = bu ou ,

wobei p _komp eine Konstante ist, mit der die Stärke der Störungskompensation eingestellt werden kann.The modified traffic volume q '/ n is then taken

where p _{comp is} a constant with which the strength of the interference compensation can be adjusted.

(entspricht einem Sägezahn-Integrierglied) und

(entspricht einem Differenzierglied). Falls δ _n L " / n ≥ δ_nL 0 / n ist

und der Sättigungszeitbedarf wird nicht verändert. Andernfalls definiert man die Hilfsvariable A = tB n tg n (L" n - L 0 n ). The calibration of the model's own saturation time requirement is advantageously carried out using a feedback method based on a classic PID controller (proportional-integral-differential controller). For this purpose, δ _n should be assigned -1 as the first value (if there is no congestion) and 1 as the second value (if there is no congestion). The calibration uses two variables:

(corresponds to a sawtooth integrator) and

(corresponds to a differentiator). If δ _n L "/ n ≥ δ _n L 0 / n

and the saturation time requirement is not changed. Otherwise the auxiliary variable is defined A = t B n t G n ( L " n - L 0 n ).

und

wobei t_d eine geeignet zu wählende Konstante ist. Daraus ergibt sich dann der kalibrierte Sättigungszeitbedarf für das Warteschlangenmodell.

wobei p_p , p_i und p_d die Parameter des Reglers bezeichnen.To avoid over-correcting the saturation time requirement, you can A '= sgn ( A ) min {| A | 1} define, where sgn ( A ) denotes the sign of A. You choose now

and

where t _{d is} a suitable constant to choose. This then results in the calibrated saturation time requirement for the queue model.

where p _p , p _i and p _d denote the parameters of the controller.

It is advantageous to smooth the calculated backflow length by forming a convex combination of L 0 / n and L "/ n : L n = γ L 0 n + (1-γ) L " n , γ ∈ [0.1].

This prevents over-correction of the backflow length.

Figur 1

die berechnete Steigung m_n der Rückstaulängenfunktion in Abhängigkeit der Zeit aus Verfahren 1,

Figur 2

den geschätzten Rückstau (in Fahrzeugen) in Abhängigkeit des explizit gemessenen, geglätteten Rückstaus aus Verfahren 1,

Figur 3

die Schätzung des Zeitbedarfswerts t B / n in Abhängigkeit der Zeit aus Verfahren 2.

Two methods according to the invention for determining the backflow length estimate with the aid of the method according to the invention for determining the congestion number are described below with reference to the drawing. The drawing shows:

Figure 1

the calculated slope m _{n of} the backflow length function as a function of time from method 1,

Figure 2

the estimated backflow (in vehicles) depending on the explicitly measured, smoothed backflow from method 1,

Figure 3

the estimate of the time requirement value t B / n as a function of time from method 2.

Procedure 1

The application of the backflow length estimation method and its verification at a driveway of a heavily used traffic light system (into the city of Landsberger / Trappentreustraße, Munich) with strongly fluctuating green times (passage times) shown.

The detector is located 30 m or approx. 5 vehicles from the stop line. As Reference fill time for this distance is taken 22 seconds.

If the reference filling time is exceeded, δ is assigned the value 0 and otherwise the value 1. The congestion index is smoothed by δ and _n = αδ _n + (1 - α) δ and _{n -1} , where α is typically between 0.05 and 0.2 and δ ₀ = δ and ₀ = 0.

The lower bound is calculated over L 0 n = q n 1-min (γ 1 , bγ 2 ) + α 1 γ i ≥ 0, where a ₁ takes into account the vehicles between the detector and stop line and therefore assumes the value a ₁ = 5. In this exemplary embodiment, γ ₁ = 0.9 and γ ₂ = 1.2 are selected. The degree of occupancy b of the detector is obtained by counting the full seconds between 5 s after the start of passage and 15 s after the end of passage in which the detector is permanently occupied and then divided by the total length of this time interval; thus b ∈ [0.1].

und

wobei

In this example, the slope m _n is written as m _n = m '/ n / m "/ n , where m' / 0 = 10 and m " / 0 = 0.5 form suitable initial values. The slope is modified using a smoothed value L '/ n = βL 0 / n ( q _n ) + (1 - β ) L and _{n -1} (δ and _n ) with β = 0.7. It is

and

in which

Suitable values for a quick but stable estimate are k ₀ = 10 and K = 1000.

Figure 1 shows the calibration of the slope m _n . The arbitrarily specified value of approx. 20 increases on the first day to the value that corresponds to the traffic characteristics of the lane. Subsequently, only slight adjustments are made. The control behavior is stable and robust.

FIG. 2 shows the comparison of the estimated, smoothed backlog length with manually ascertained, slightly smoothed backflow length values. The measured backflow L real / n was with L real n = 0.3 L real n + 0.7 L real n -1 smoothed. A squared correlation coefficient of R ² = 0.7748 indicates a good relationship between the estimated and real backflow length.

Procedure 2

As an application of the method, the determination of the backflow length on the in the above Example of access to a traffic light system using a queue model described.

To calculate the saturation time requirement, a maximum change of Δ t B / max = 0.02, is allowed. The change is additionally dampened for the use in the queue model with the factor c _d = 0.9.

FIG. 3 shows the determination of the time requirement value t B / n as a function of time with an initial value of t B / 0 = 2 s . It can be seen that in addition to the settling process, fluctuations of t B / n occur several times within the two working days. These fluctuations can be explained, among other things, by changing traffic compositions and the time-dependent driving behavior of road users.

Disruptions in the drain are compensated for using the occupancy rate known from the above example. The fault index s results from equation (9), with u = 0.2 and o = 1.1 being used for the barriers. This choice guarantees that s always remains less than 1.

und

wobei C = 0.6 die statistischen Schwankungen beim Abfluß charakterisiert.The macroscopic queue model is taken from RM Kimber and EM Hollis, Traffic queues and delays at road junctions , TRRL Laboratory Report 909, Berkshire, 1979 in this example. The model equation for the backflow length L is L = 1 2 ( A 2 + B - A ) With

and

where C = 0.6 characterizes the statistical fluctuations in the discharge.

Suitable parameters for the calibration of the saturation time requirement analogous to a PID controller are p _d = 0.003, p _i = 0.01, p _d = 0.01 and t _d = 1.2.

The backlog length estimation is smoothed with γ = 0.6.

Claims (20)

1. Method of determining a tailback characteristic factor δ at operating stations for despatching individually moving units having alternating barrier and release phases and having a detector upstream of the respective operating station by measuring the filling time between the barrier start or a time instant tied to the barrier start and continuous occupancy of the detector and subsequent comparison with a reference filling time, in which method a first value is assigned to the tailback characteristic factor δ if the reference filling time is exceeded and a second value is assigned if the reference filling time is not exceeded.
2. Method according to Claim 1, in which the reference filling time is chosen as a function of the geometry of the inflow region of the operating station.
3. Method according to one of the preceding claims,in which the reference filling time is chosen as a function of the release time.
4. Method of determining the tailback length L and_n in the nth despatch phase by

   (a) determining the nth tailback characteristic factor δ _n according to Claim 1,

   (b) calculating a smoothed tailback characteristic factor δ and _n using the (n-1)th smoothed tailback characteristic factor δ and_n-1 ,

   (c) determining the tailback length L and_n (δ and_n ) = mδ and_n with suitably predetermined slope m.
5. Method according to Claim 4, wherein the slope m_n is determined in the nth despatch phase by

   (a) determining the traffic level q_n ,

   (b) calculating a lower limit L_n ⁰ for the tailback length as a function of q_n ,

   (c) determining the slope m_n by comparison of L_n ⁰ with L and_{n -1}(δ and_n ) with a suitably predetermined slope m_{n -1}.
6. Method according to Claim 5, in which the slope m_{n -1} is determined by recursive application of the method according to Claim 5 with suitable starting conditions for m ₀ and δ and ₀.
7. Method according to one of Claims 4-6 , in which the smoothed tailback characteristic factor δ and_n is calculated as a convex combination of δ_n and δ and_{n -1} in accordance with δ and _n = αδ _n + (1-α) δ and_n-1 , α ∈ [0,1].
8. Method according to one of Claims 5-7, in which the traffic level q_n is measured with a detector situated upstream of the operating station.
9. Method according to one of Claims 5-8, in which the lower limit L 0 / n of the tailback length is predetermined as a linear function of q _n .
10. Method according to Claim 9, in which the slope L_n ⁰ (q_n) is predetermined as a function of the time, in which the detector is continuously occupied during a portion of the despatch phase.
11. Method according to one of Claims 5-10, in which the slope m_n is altered with respect to m_n-1 if the second value is assigned to δ_n and L _n ⁰ > L and _n-1(δ and _n)=m_n-1δ and_n or if the first value is assigned to δ_n and L _n ⁰ < L and_{n -1}(δ and_n ) = m_{n -1} δ and_n and otherwise m_n =m_{n -1} is set.
12. Method according to one of Claims 5-11, in which the slope m_n is adapted by means of a smoothed value L_n' = βL 0 / n (q_n ) + (1 - β) L and_n-1 (δ and _n) where β > 0.
13. Method of determining the saturation time requirement t_n ^B , which corresponds to the average time requirement of a unit with saturated flow during the release phase, by

    (a) determining the tailback characteristic factor according to one of Claims 1-3,

    (b) determining the traffic level q_n ,

    (c) determining the saturation time requirement t _n ^B using the release time t _n ^g and a suitable starting condition for t ₀ ^B in accordance with

    

    if δ_n = δ _n-1 is equal to the second value, otherwise.
14. Method according to Claim 13, in which the saturation time requirement t B / n is altered in each nth despatch phase by not more than a predetermined maximum value compared with the saturation time requirement of the (n-1)th despatch phase.
15. Method according to Claim 13 or 14, in which the traffic level q_n is measured with the detector upstream of the operating station.
16. Method of determining the tailback length L"_n by

    (a) determining the saturation time requirement t ^B _n according to one of Claims 13-15,

    (b) determining the inherent model saturation time requirement τ ^B _n in accordance with τ _n ^B = τ ^B _n-1 + c_d (t ^B _n - t ^B _n-1 using the (n-1)th model saturation time requirement τ ^B _n-1 and with a suitably chosen c_d ,

    (c) calculating a lower limit of the tailback length L ⁰ _n as a function of q_n ,

    (d) calculating a tailback length estimation with a queue model using the inherent model saturation time requirement,

    (e) calibrating the inherent model saturation requirement by comparing the tailback length estimation with the lower limit L _n ⁰,

    (f) calculating the tailback length L_n " with a queuing model using the calibrated inherent model saturation time requirement.
17. Method according to Claim 16, in which the tailback length calculation is made with a modified traffic level that takes account of faults in the outflow.
18. Method according to Claim 17, in which the flow compensation is calculated by counting in a time interval during the despatch phase predetermined time intervals, in particular complete seconds, in which the detector is continuously occupied.
19. Method according to one of Claims 16 to 18, in which the inherent model saturation time requirement is calibrated with a method along the lines of a classic PID controller.
20. Method according to one of Claims 16-19, in which the tailback length estimation is smoothed by forming a convex combination of L _n ⁰ and L _n ^" in accordance with L_n = γL _n ⁰ + (1-γ)L _n ^", γ ∈ [0,1].

Images