Method and device for controlling traffic flows at intersections

Abstract

Method for controlling traffic flows at intersections by means of a signaling system and at least one control unit (4), wherein the signaling system has a parameterizable cycle time c and the control unit (4) calculates control parameters for the signaling system based on the loss times of road users and transmits them to the signaling system, wherein the loss times are determined by sensors, characterized in that an average loss time di over a time interval Δtupdate is determined for each approach network edge by means of the sensors, wherein in the control unit (4) average approach traffic volumes qi are determined from the determined average loss times and converted into the control parameters, wherein the cycle time c is a multiple smaller than the time interval Δtupdate, wherein the average approach traffic volumes qi are determined by means of the Equationdi(t)=f(qi(t);gi(t))=c(1−λi(t))22(1−qi(t)s)+(qi(t)λi(t)s)22qi(t)(1−qi(t)λi(t)s)withλi(t):=gi(t) / c can be determined, where S is a known saturation traffic volume, where the release times gi(t+1) for the next time slice are determined by gj(t+1)=qj(t)∑qk(t)⋅c with j∈{1,2,…} or the average inflow traffic volumes q1, q2 are determined by means of the coupled, non-linear System of equationsd1(t)=f(q1(t);q1(t)q1(t)+q2(t)c)=c(1−q1(t)q1(t)+q2(t))22(1−q1(t)s)+(q1(t)+q2(t)s)22q1(t)(1−q1(t)+q2 (t)s)d2(t)=f(q2(t);q2(t)q1(t)+q2(t)c)=c(1−q2(t)q1(t)+q2(t))22(1−q2(t)s)+(q1(t)+q2(t)s)22q2(t)(1−q1(t)+q2(t)s)determined are, whereby the release times gj(t) respectively can be determined by gj(t)=qj(t)q1(t)+q2(t)⋅cwithj∈{1,2}.;

Description

[0001] The invention relates to a method and a device for managing traffic flows at intersections by means of a signaling system.

[0002] Traffic flow at intersections within a road network is typically controlled by traffic signal systems (TSS), with these intersections usually being junctions. Fixed-time control or adaptive control of the blocking and release times are common.

[0003] A good overview of the state of the art for adaptive traffic control using traffic signal systems, especially those based on stationary detection, can be found in DE 10 2009 033 431 B4. DE 10 2009 033 431 B4 itself (similar to DE 10 2010 027 327 B3) further describes a method for controlling a traffic signal system based on the so-called loss times of individual vehicles. Loss times occur whenever a vehicle has to reduce its speed (possibly to a standstill) due to a traffic jam at the signal. With each second below the clear speed, the vehicle-specific loss time increases continuously. The aim of the described control method is to extend the signal's clearance time, within certain limits, until all vehicles with accumulated loss time have passed the stop line and the traffic jam has cleared. The crossing traffic flow is then released.Communication between the appropriately equipped vehicles and the signaling system takes place in the sense of vehicle-infrastructure integration – also called car-to-infrastructure communication (C2I) – as described, for example, in EP 1 628 274 B1. DE 10 2011 107 663 B4 extends this approach to include the determination of a main traffic direction based on the so-called loss time traffic volume, which is defined as the traffic volume of vehicles with accumulated loss time.

[0004] The basic requirement in any case is the availability of appropriately equipped vehicles similar to the floating car principle known from the literature, with the additional possibility of direct communication between vehicle and infrastructure (i.e., signaling system).

[0005] In accordance with current technology, adaptive traffic control at intersections is primarily achieved using signal systems based on stationary detectors installed specifically for this purpose. Newer technologies, such as the aforementioned floating car approaches, are conceptually much more flexible, as they operate not only locally but across a wider area and also have the potential to integrate additional driver assistance functions (e.g., traffic-dependent navigation, traffic light assistance, dynamic green wave). However, even in the most favorable case, the loss-time-based approaches of DE 10 2009 033 431 B4, DE 10 2010 027 327 B3 or DE 10 2011 107 663 B4 require an equipment rate of at least approximately 10% of communication-enabled vehicles to achieve meaningful control results (cf. Oertel, 2011: Loss-time-based traffic signal control of a single node, Straßenverkehrstechnik 9 / 2011, pp. 561–568).

[0006] In addition to the aforementioned floating car approach, DE 10 2010 018 815 A1 discloses a method for generating traffic information within a spatial area, wherein a detector determines its spatial position at least with respect to the spatial area. The detector detects at least one signal from at least one first transmitting unit, and the detector identifies an identifier of the first transmitting unit. The detector transmits at least its spatial position and the identifier of the first transmitting unit to a central unit for generating traffic information, wherein the central unit determines traffic information within the spatial area from at least the transmitted data. The detector, which is preferably a detector of WLAN signals and / or Bluetooth signals and / or GSM signals, can be mobile or stationary.In general, the detector picks up signals from wireless devices.

[0007] From Webster, FV (1958): Traffic Signal Settings, Road Research Technical Paper No. 39, Her Majesty's Stationery Office, London, a calculation of loss times from inflow traffic volumes is known.

[0008] The invention is based on the technical problem of providing a method for controlling traffic information at intersections and creating a device by means of which adaptive control of traffic flows is possible with less sensor and computing effort.

[0009] The solution to the technical problem is achieved by the method with the features of claim 1 and the device with the features of claim 5. Further advantageous embodiments of the invention are set forth in the dependent claims.

[0010] The method for controlling traffic flow at intersections of a road network is implemented using a traffic signal system and at least one control unit. The traffic signal system is preferably a traffic light system, and the intersection is preferably a junction. The traffic signal system has a configurable cycle time. This cycle time is equal to the sum of the green and blackout times for one direction of traffic. The control unit calculates control parameters for the traffic signal system based on the loss times of road users. The loss time can be measured directly or, for example, determined from travel times. The loss times are detected by sensors, or the sensors collect traffic information from which the loss time can be calculated.

[0011] Using the sensors, an average loss time is calculated for each inflow edge over a time interval Δt. updatedetermined, where the orbital period is a multiple smaller than the time interval Δt update For example, the orbital period is one minute and the time interval is Δt. update The process lasts fifteen minutes. In the control unit, average traffic flow rates are then calculated from the average loss times and converted into the control parameters (blocking and release times) of the signaling system.

[0012] This approach offers numerous advantages. Since the time interval is relatively long, direct communication between the vehicles of the traffic signal system is unnecessary. Instead, the data can be collected centrally and then transmitted to the signal system. Another advantage is that, for example, when using FCD vehicles as sensors, significantly lower equipment rates are required, as only moderate loss times are needed. Equipment rates between 1% and 2% are sufficient, which is already achievable today. Various methods exist for converting loss times into incoming traffic volumes.

[0013] Preferably, the allocation of release times at traffic-controlled intersections is analogous to optimal fixed-time control (see Webster, 1958). The idea is to base blocking and release times on the proportions of the incoming traffic volumes of the individual intersection flows, but to recalculate these regularly according to the update interval of the mean loss times (e.g., every 15 minutes). For this purpose, the measured mean loss times are first converted into the respective incoming traffic volumes using common waiting time formulas for traffic congestion at signal systems (see Webster, 1958), from which the allocation of blocking and release times is then derived.

[0014] This means that – unlike, for example, in DE 10 2009 033 431 B4 – given the low equipment rates in practice for floating cars, the method according to the invention does not achieve vehicle-specific traffic adaptability of the signaling system. However, it does create a new, optimal fixed-time control regularly, according to the update interval of the average loss times (e.g., 15 minutes), and thus also achieves a form of traffic adaptability. Since the control dynamically follows the actual traffic demand, the method according to the invention also offers a conceptual advantage over the so-called time-dependent signal program selection frequently used in practice. In the latter, assuming regularly recurring traffic patterns, various fixed-time control settings are not directly dependent on traffic but solely on the day of the week and time of day.

[0015] Regarding the determination of blocking and release times according to the inventive method, two alternative approaches are possible.

[0016] In a first embodiment, which can also be described as a forward approach, the average inflow traffic volumes q are i using the equation with λ (t) i ≔ g (t) i / c determined, whereby the release times g i (t+1) simplified for the next time period by be determined.

[0017] Here, c is the cycle time and s is a saturation traffic volume assumed to be constant. Furthermore, for a typical intersection, j ∊ {1, 2}.

[0018] The release times are g j (t+1) that is, based on the time interval (update interval) Δt after the end of the time interval. update The quantities known at time t were determined, namely the determined average loss times and the previously defined release times g. j(t) based on the inflow traffic volumes determined in a previous update interval q i (t-1) The advantage is the very simple calculation.

[0019] In a second embodiment, which can also be described as a reverse approach, the average inflow traffic volumes q1 are (t) , q2 (t) by means of the coupled, non-linear system of equations determined, whereby the release times g j (t) each by be determined.

[0020] Formally, the release times are g j (t) for the update interval here retrospectively based on the estimated inflow traffic volumes q i (t)These times are determined from the average downtimes in the current update interval. The advantage of this more complex calculation compared to the forward step is the direct, formulaic link between release times and average downtimes in the same time interval, which tends to lead to greater control stability.

[0021] The sensors could, for example, be FCD vehicles, since only a few vehicles are sufficient for a mean loss time. Alternatively or additionally, wireless devices based on WLAN, Bluetooth, GSM, etc., can also be used. Camera-based sensors can also be employed.

[0022] The invention is explained in more detail below with reference to a preferred embodiment. The figures show:

[0023] Fig. 1 a schematic block diagram of a device for controlling traffic flows at junctions of a traffic network,

[0024] Fig. 2. A representation of the loss time versus the inflow traffic volume,

[0025] Fig. 3. A representation of the intersecting inflow traffic volumes,

[0026] Fig. 4 a representation of the bounded solution space according to Fig. 3,

[0027] Fig. 5 a representation of a contour diagram in the solution plane according to Fig. 3,

[0028] Fig. 6. A representation of another contour diagram in the solution plane according to Fig. 3,

[0029] Fig. 7 a graphical representation of the existence of at least one non-degenerate solution,

[0030] Fig. 8 a graphical representation of the uniqueness of the solution and

[0031] Fig. 9 a graphical representation of a missing solution when the given conditions are violated.

[0032] Fig. 1 shows a device ( 1 ) for carrying out the method according to the invention. This is a four-arm, two-phase controlled intersection with indicated time-dependent traffic volumes. q (t) 1 , q (t) 1' , q (t) 2 , q (t) 2' the intersecting traffic flows. The depicted floating cars 2 In accordance with a common floating car approach, they send their positions to an FCD central unit. 3 , which, using known algorithms and methods, calculate average travel times at regular time steps (update interval e.g. 15 min). ((Δt) (t) 1 , (Δt) (t) 1' , (Δt) (t) 2 , (Δt) (t) 2' ) transferred per inflow network edge. By subtracting the so-called free travel time according to with length L i of the respective inflow arm and the respective permitted maximum speed or the normally observed free-running speed v max,iFor i ∊ {1, 1', 2, 2'}, the mean loss times are thus obtained in each update interval. d (t) 1 , d (t) 1' , d (t) 2 , d (t) 2' for the four inflow arms. These are forwarded to a control unit. 4 (e.g. within a traffic control center), which adaptively determines the switching parameters (i.e., in particular the blocking and enabling times) of the traffic signal system according to the inventive method and transmits these via a conventional, connected signal controller. 5 to the signal transmitters 6.1 – 6.4 the intersection.

[0033] The crucial factor is how the signal parameters are determined from the mean loss times. It is known (see Webster, 1958) that the loss times d (t) i with i ∊ {1, 1', 2, 2'} given a cycle time c and a known saturation traffic volume s (assumed constant here for all inflow arms!) as a function f of the respective inflow traffic volume q (t) i and the associated (effective) release time g (t) i ≤ c can be described. Specifically, the following applies: with λ (t) i ≔ g (t) i / c. It should be noted that f for q (t) i < λ (t) i s, as in Fig. 2 for positive values ​​of q (t) i outlines a continuous, strictly monotonic function with respect to q (t) i with pole station at λ (t) i s is, i.e., in particular

[0034] Since, in the case of a 2-phase control of the traffic signal system in the embodiment described here, fundamentally g (t) 1 = g (t) 1' and g (t) 2 = g (t) 2' This holds true, and therefore, due to the assumption of a constant saturation traffic volume s for all four inflow arms, it follows that q (t) 1 ≥ q (t) 1' This is fulfilled precisely when d (t) 1 ≥ d (t) 1' . The analogous relationship also applies, of course, to q (t) 2 and q (t) 2' . Thus, despite unknown q (t) i solely based on the measured average loss times per inflow arm (d (t) i ) determine which is the so-called relevant traffic volume for each phase, i.e., the larger value of q (t) 1 and q (t) 1' or q (t) 2 and q (t) 2' is. The following applies without restriction: q (t) 1 ≥ q (t) 1' and q (t) 2 ≥ q (t) 2' (otherwise the indices will be swapped).

[0035] Assuming constant saturation traffic volumes s for all four inflow arms, an optimal fixed-time control can then be derived from the ratio of q (t) 1 and q (t) 2 (see Webster, 1958). Given the adaptive adjustment of the signal parameters to be carried out in each update interval according to the method of the invention, two different variants for determining the (effective) release times result. g (t) 1 and g (t) 2 – and thus also the (effective) release times g (t) 1' and g (t) 2' and the (effective) curfew times r (t) i = c – g (t) i with i ∊ {1, 1', 2, 2'}. Option 1:

[0036] The release times g (t+1) j For each subsequent time step (t + 1), a "forward approach" (cf. variant 2 below in contrast) is used solely based on the quantities known after the end of the update interval t, i.e., the measured values. d (t) j and the one in the previous time step based on the estimated relevant traffic volumes q (t-1) 1 and q (t-1) 2 determined g (t) j determined for j ∊ {1, 2}. The required values ​​of q (t) 1 and q (t) 2 According to (2), the following solutions to the equation are obtained for j ∊ {1, 2}

[0037] Provided that q (t) j < λ (t) j s = q (t-1) j s / (q (t-1) 1 + q (t-1) 2 ) For j ∊ {1, 2}, the following holds:

[0038] As can also be seen from the strict monotonicity of the function f in conjunction with (3), a unique, non-degenerate solution of (5) exists if and only if a solution exists in the respective open interval. ]0, λ (t) j s[ with j ∊ {1, 2}, if applies. In all other cases, where one of the mean loss times d (t) j The violation of condition (9) follows under the natural assumption that at least d (t) j ≥ 0 According to (6), a non-positive solution applies, i.e., a decisive traffic volume. q (t) j ≤ 0. This is obviously unrealistic, so in this case a traffic volume is assumed which, according to the release times set in (4), leads to a minimum release in accordance with common guidelines. In general, from a practical point of view, minimum release times should be observed for both phases of the signal control, which in individual cases may mean a deviation from the definition of the release time in (4). Furthermore, if both d (t) 1 as well as d (t) 2 If condition (9) is not met, the traffic volumes may q (t) 1 and q (t) 2 with the consequence of equal release times g (t+1) 1 and g (t+1) 2 In (4), for example, they can be chosen identically. Alternatively, a setting according to the ratio of d (t) 1 to d (t) 2 conceivable. Option 2:

[0039] The second variant of the method according to the invention is, in a sense, a "backwards approach". Formally, the release times are g (t) j for the update interval t with j ∊ {1, 2}, namely subsequently based on the estimated traffic volumes q (t) 1 and q (t) 2 which in turn are determined according to (2) from the measured mean loss times, which are only known at the end of the time interval t d (t) 1 and d (t) 2 can be derived. The basic assumption here is the existence of a stationary traffic situation, in which there is always This applies regardless of the release times that should actually be switched. g (t) j at the switching time in interval t, the values ​​are not yet known. In practice, one will get around this by simply using the values ​​calculated under the assumption of steady state in the update interval t. g (t-1) j as actually switched release times (t) j , dh (t) j ≔ g (t-1) j to be used for j ∊ {1, 2}. The difference to the first variant of the method according to the invention therefore lies in the possible discrepancy between g (t) j and (t) j . Furthermore, the quotient of release time and cycle time defined above depends on (2), i.e. λ (t) j ≔ g (t) j / c, in this case – deviating from the first option – directly from q (t) 1 and q (t) 2 from and is therefore no longer a constant in the solution of equation (2). In particular, unlike in (5), this does not result in two separate equations, but rather in the significantly more complex, coupled non-linear system of equations. with the unknown q (t) 1 and q (t) 2 . As before, a solution is sought that q (t) j ∊ ]0, λ (t) j s[ for j ∊ {1, 2} or because λ (t) j = q (t) j / (q (t) 1 + q (t) 2 ) equivalent to this:

[0040] The space of feasible solutions is therefore an open triangle in the first quadrant of the q1-q2 plane (cf. Fig. 3), where, as in the following, the time index (t) is suppressed for all variables appearing in (12) for the sake of clarity.

[0041] For all admissible pairs (q1, q2) according to (13) the following then applies:

[0042] Similar to (9), for a non-degenerate solution at least the two conditions must be met simultaneously. must be fulfilled. In the non-degenerate case with q j For j ∊ {1, 2} to be greater than 0, at least d1, d2 > 0 must hold. Furthermore, (15) provides the equivalent inequalities.

[0043] Equation (16), together with d1, d2 > 0, thus directly implies a necessary condition for the existence of a non-degenerate solution of the system of equations (12), which depends only on the measured mean loss times d1 and d2, namely

[0044] In practical terms, this sometimes means a reduction of the solution space in the sense that non-degenerate solutions are only possible within a sub-area of ​​the feasible triangle. Fig. 3 are possible. The additional boundary lines result directly from (16) and are in Fig. Figure 4 is shown schematically. A measured mean loss time d1 < c / 2 always implies a reduction of the solution space from the direction of the q2-axis, and a value d2 < c / 2 implies a reduction from the direction of the q1-axis. Mean loss times d j≥ c / 2 with j ∊ {1, 2}, however, do not lead to any additional restriction of the solution space with respect to non-degenerate solutions, since the respective inequality in (16) is automatically satisfied in this case. It should also be noted that the reduced solution space becomes the empty set for certain constellations of values ​​d1 and d2, namely precisely when the additional upper and lower boundary lines in Fig. 4 cross or equivalently condition (17) is violated.

[0045] Therefore, let d1, d2 > 0 such that (16) is satisfied. Otherwise, if any solutions exist at all, they are only degenerate solutions of the system of equations (12), and a default approach with suitable minimum release times should be applied analogously to the first variant.

[0046] The goal is therefore to find a solution (q1, q2) of (12) in the possibly reduced solution space according to Fig. 4. To do this, one first considers the function values ​​of the function f from (12) for all relevant pairs (q1, q2), independently of d1 and d2, where in the case of the first equation from (12) the in Fig. The contour diagram shown in section 5 is obtained. The second equation from (12) also yields a completely symmetrical diagram (see below). Fig. 6) The contour lines in the solution space according to (13) with can be described as solutions of a 4th-degree polynomial, as can be seen by rearranging the equations in (12). The complete determining equation for the function values ​​of In this context, it is for all d j > 0.

[0047] One can prove the unequivocal existence of However, it can also be demonstrated with less effort.

[0048] It should be noted that for j, k ∊ {1, 2} with j ≠ k the function in the interval [0, s[ strictly monotonically increasing in q j is with

[0049] The equivalent transformation of the equation f0(q j ) = d j further shows that f0 has the value d j > 0 exactly at that point assumes that this is a non-trivial zero point of the contour line on the boundary of the solution space according to (13). For q j ∊ [0, s[ therefore f0(q j ) < d j or f0(q j ) > d j exactly when q j < n(d j ) or q j > n(d j ). Now is for any, but fixed, q j ∊]0, s[ because in the solution space according to (13) strictly monotonically increasing in q k -direction with

[0050] For reasons of continuity and monotonicity, therefore, for all q, there exists j ∊]0, n(d j )[ each unique so that This applies. For q j ∊]n(d j ), s[ does such a thing exist because f0(qj ) > d j However, not with. This confirms the unambiguous existence of the contour line as a function with function values ​​such that always This is shown. From the continuous differentiability of the function f from (26) it even follows that the differentiability of on the interval ]0, n(d j )[ as well as the continuity at point n(d j ).

[0051] Fig. 5 and Fig. 6 further imply that the continuous extensions of all contour lines for q j ↓ 0 passes through the origin. To prove this, let j, k ∊ {1, 2} with j ≠ k, as before. Furthermore, let d j > 0 arbitrary. Because of (23) and the strict monotonicity of the function f0 from (21), there then exists an ε~ > 0 such that for all ε > 0 with ε < ε~ the inequality This holds true. Then, because of (32), for every ε > 0 with ε < ε~, there exists an additional δ > 0 such that for all δ > 0 with δ < δ, the estimate This results in... Furthermore, considering the limit for any fixed δ > 0... Thus, from the continuity of the function f, from (26) together with (33) the existence of a q~ follows. j ∊]0, ε[ such that applies. The contour line h dj It therefore always runs through the rectangle and the border crossing ε, δ → 0 shows that d j > 0 was chosen arbitrarily, as claimed for all d j > 0.

[0052] The limit the slope of the contour line at the origin for a given d j < c / 2 with j ∊ {1‚ 2}, given its existence, is, moreover, identical to the slope of the corresponding boundary line of the non-degenerate region of the solution space according to (16). In general, for j ∊ {1, 2}:

[0053] To prove this, let j, k ∊ {1, 2}, such that j ≠ k. For the contour line This then automatically applies due to its course in the first quadrant of the q1-q2 plane (assuming the existence of the limit).

[0054] The two cases d must be distinguished below. j ≥ c / 2 and d j < c / 2. i) Let d be the first j ≥ c / 2. For any m ≥ 0, the following then holds for q k = mq j

[0055] The straight line q k = mq j Therefore, it is not tangent to the contour line. in the origin, and it follows

[0056] From (38) the claim for d follows. j ≥ c / 2, i.e. ii) In the second case, let d j < c / 2. With and q k = mq j This then applies analogously to (39) and due to the strict monotonicity of the function m ↦ m 1 + m the inequality

[0057] Again, the straight line q k = mq j no tangent to the contour line in the origin, and according to (38) it follows

[0058] Conversely, with accordingly

[0059] Together with (44) therefore, as claimed

[0060] If one then considers the zeros of the corresponding continuous contour lines for given d1, d2 > 0 and (including the origin), as (37) proves according to the schematic drawing in Fig. 7 the existence of at least one intersection point of the two contour lines and thus the existence of a non-degenerate solution of the system of equations (12) with q1, q2 > 0, which, because of (16), automatically lies within the reduced solution space, whenever d1 ≥ c / 2 or d2 ≥ c / 2 or otherwise This holds true (see (17)). It follows that (17) is therefore not only a necessary but also a sufficient condition for the existence of a solution to the system of equations (12). Systematic numerical test calculations further demonstrate that the contour lines in the entire triangle of the solution space according to (13) strictly convex functions in the variable q j are. Overall, this ultimately results in the following: Fig. 8 outlines the situation regarding the location of the solution to the system of equations (12). The strict convexity of the contour lines implies the important uniqueness of the solution, since otherwise the control parameters (i.e., blocking and enabling times) according to (10) would not be uniquely defined. Fig. Figure 9, incidentally, schematically shows why no (non-degenerate) solution exists if condition (17) is violated. Due to their described properties, there is no intersection point of the corresponding contour lines in the open triangle of the solution space according to (13) or (18). Fig. 3.

[0061] In summary, condition (17) is thus a very simple criterion that can be used to decide, based solely on the measured mean loss times, whether it is possible to determine the release times of the signaling system to be controlled by solving the system of equations (12), or when a default approach based on predefined minimum release times is required. Furthermore, the benign structure of the non-linear problem from (12) allows for a reliable solution using standard numerical methods. QUOTES INCLUDED IN THE DESCRIPTION

[0062] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0063] DE 102009033431 B4 [0003, 0003, 0005, 0014] DE 102010027327 B3 [0003, 0005] EP 1628274 B1

[0003] DE 102011107663 B4 [0003, 0005] DE 102010018815 A1

[0006] Cited non-patent literature

[0064] Oertel, 2011: Loss time-based traffic signal control of a single node, Straßenverkehrstechnik 9 / 2011, pp. 561–568

[0005] Webster, FV (1958): Traffic Signal Settings, Road Research Technical Paper No. 39, Her Majesty's Stationery Office, London

[0007] Webster, 1958

[0013] Webster, 1958

[0033] Webster, 1958

[0035]

Claims

[1] Method for controlling traffic flows at intersections by means of a signaling system and at least one control unit ( 4 ), wherein the signaling system has a parameterizable cycle time c and the control unit ( 4 ) Control parameters for the signaling system are calculated based on the loss times of road users and transmitted to the signaling system, whereby the loss times are determined by sensors, characterized by that the sensors determine the average loss time d i over a time interval Δt update is determined for each inflow network edge, whereby in the control unit ( 4 ) from the determined average loss times, average inflow traffic volumes q i are determined and converted into control parameters, where the orbital period c is a multiple smaller than the time interval Δt update is. [2] Method according to claim 1, characterized by that the average inflow traffic volumes q iusing the equation with λ (t) i ≔ g (t) i / c to be determined, whereby the release times g i (t+1) for the next time segment by be determined. [3] Method according to claim 1, characterized by that the average inflow traffic volumes q1, q2 can be determined using the coupled, non-linear system of equations to be determined, whereby the release times g j (t) each by be determined. [4] Method according to any of the preceding claims, characterized by that the mean loss times using FCD vehicles ( 2 ) and / or by means of at least one camera and / or signals from wireless devices. [5] Device ( 1 ) for controlling traffic flows at junctions of a traffic network, comprising a signaling system, at least one control unit ( 4) and sensors for determining the loss times of road users, wherein the signaling system has a parameterizable cycle time c, wherein the control unit ( 4 ) is designed in such a way that these control parameters for the signaling system are calculated based on the loss times of road users and transmitted to the signaling system, characterized by that the sensors or an evaluation unit is designed in such a way that a mean loss time d can be calculated based on the sensor data. i over a time interval Δt update is determined for each approach edge of the transport network, whereby the control unit ( 4 ) is designed in such a way that average inflow traffic volumes q can be derived from the determined average loss times i are determined and converted into control parameters, where the orbital period c is a multiple smaller than the time interval Δt update is. [6] Device according to claim 5, characterized bythat the control unit ( 4 ) is designed in such a way that the average inflow traffic volumes q i using the equation with λ (t) i ≔ g (t) i / c to be determined, whereby the release times g i (t+1) for the next time segment by be determined. [7] Device according to claim 5, characterized by that the control unit ( 4 ) is designed such that the average inflow traffic volumes q1, q2 are determined using the coupled, non-linear system of equations to be determined, whereby the release times g j (t) each by be determined. [8] Device according to any one of claims 5 to 7, characterized by that the sensors are used as FCD vehicles ( 2 ) and / or at least one camera and / or are designed as wireless end devices.

Images