EP2161698B1 - Method for coordinating light signal-controlled nodes in a street network - Google Patents

Description

The invention relates to a method for the coordination of light signal-controlled nodes of a road network according to the preamble of claim 1.

In inner-city road networks vehicle traffic at junctions, hereafter referred to as node, controlled by traffic lights. A traffic signal system comprises signal generators grouped for signal groups for different traffic flows for emitting light signals to the road users. Typically, at a node there is a main direction signal group for controlling the traffic along a main traffic direction at the node and a sub direction signal group for controlling the traffic turning in or out of a side traffic direction. The traffic signal system further comprises a control device in which a signal program runs to switch on the signal groups according to specific signal times. The signal times comprise, for each signal group, the green times, defined by the times green start and end of green within one round trip time, as well as a phase sequence of the vehicle traffic blocking red and releasing green phases. Basically, a distinction fixed-time signal controls with specified, for example, time-of-day, signal times without impact opportunities by road users and traffic-dependent signal controls in which the road user can influence the signal program. In partially or fully traffic-dependent controllers, the signal program is specified as a frame signal plan whose phase transitions are immutable while maintaining intermediate times, but whose durations can be expanded or compressed as required within predefinable permission ranges.

The Green Wave is of particular importance for controlling the traffic flow through traffic lights over several nodes. It is achieved by coordinating the signal programs of adjacent nodes, in which the majority of vehicles can pass several nodes without stopping, while maintaining a certain speed. In this case, the green times in the direction of travel of successive nodes are matched by offsetting the signal programs. The Green Wave on the road or in the street is mainly used to reduce the sum of all personal travel times, to improve driving comfort, to reduce fuel consumption and to minimize the impact of the environment on noise and pollutants. For this purpose - as well as to increase traffic safety - the aim is to keep the scattering of the speeds of the individual vehicles and the number of stops of all vehicles as small as possible. In the road network, a total optimization is to be striven for. Green waves for motor vehicle traffic are recommended for distances between traffic signals of up to 750 m, in special cases up to 1000 m. At longer distances, vehicle spools dissolve to such an extent that it is no longer sensible to coordinate the traffic lights.

For the coordination of traffic light systems in addition to the offset time, the selection of the phase sequence as a manipulated variable available. The optimization task is a problem of the complexity class NP, short for: non-deterministic with polynomial computation time. There is no known optimization technique that solves the original optimization task while guaranteeing the global minimum.

One approach to dealing with this dilemma is to determine the offset times rule-based. This is used in particular for the offline planning of green waves. Taking into account the travel times between The nodes are used in pairs as wide as possible Grünband overlaps. An optimization in the true sense does not take place. An impact model for determining waiting times and holding is not required.

Another approach is to use such heuristic optimization techniques that are known to avoid local optima, such as genetic algorithms, to arrive at an optimum without changing the original optimization task. Usually, additional constraints are introduced that narrow the potential solution space.

Another approach reformulates the optimization task so that it is no longer NP-heavy. Please refer to the brochure "Offset optimization in the road network: VERO", published 11/1994 by Siemens AG, order no. A24705-X-A367 - * - 04, a method for optimizing the coordination of traffic light systems in a road network known, which emanates from the intensity distributions of the individual tributaries to a traffic signal, ie the distribution of each arriving at the end of the access traffic intensity. An optimization plan defines an order of the nodes - and thus an optimization direction - in which the nodes are coordinated. Optimum skew times are determined between the signal time schedules of the node currently to be coordinated, referred to below as the main node, and the already coordinated adjacent node, referred to hereinafter as the last or last pre-node. For this purpose, an objective function in the form of a weighted sum of waiting times and numbers of vehicle stops is minimized in and against the optimization direction between the last Vorknoten and the main node moving vehicle body. The waiting times and numbers of stops thereby depend on the phase sequences of the signal time schedules of the main node and the last predecode, on the offset time between these signal timetables and on the vehicle pulse-modeling intensity distributions. Each signal group becomes one Intensity distribution of a vehicle body approaching the signal group and an intensity distribution of a transmitted from the signal group vehicle pulse modeled. As intensity distributions, rectangular profiles with a high and low intensity section, parabolic approximated profiles or any profile shapes obtained from measured data are used.

As described in more detail below, this known coordination method suffers from the disadvantage that the modeled intensity distributions before and after the signal groups of the main node do not simulate the real vehicle pulse well enough and therefore lead to suboptimal coordination results.

The invention is therefore based on the object to provide a method of the type mentioned, which compared to the cited prior art provides improved coordination results.

The object is achieved by a generic coordination method with the features specified in the characterizing part of claim 1. In determining the optimal offset time for the main node, it is evaluated how well an intensity distribution modeled for a vehicle cluster approaching the signal group of the last predecode against the optimization direction agrees with the intensity distribution that is involved in coordinating the last predecode for one of the signal group of the last predecode against the optimization direction to the penultimate preselected transmitted vehicle pulse was modeled. Alternatively or simultaneously, it is evaluated how well an intensity distribution, which is modeled for a vehicle cluster approaching the signal group of the main node in the optimization direction, agrees with the intensity distribution that results when a next sequence node, that is, one of the main nodes, is coordinated adjacent node and following in the optimization plan, will be modeled for a vehicle cluster sent by the signal group of the main node in the optimization direction. In this way, the problem is encountered that arrival distributions are not yet known by themselves against the optimization direction or departure distributions of the vehicle pulse moving in the optimization direction because of the not yet coordinated signal groups at the main node. As a result, intensity distributions for vehicle pulse trains can be simulated, which come closer to reality and lead to better coordination results.

In an advantageous embodiment of the coordination method according to the invention, the intensity distribution which was modeled for a vehicle pulse transmitted from a main direction signal group of the penultimate node to the last precursor is taken into account for the signaling of the main direction signal group of the last forward node for a vehicle pulse transmitted from the last precursor node to the main node weiterpropagiert. The intensity distributions are thus calculated away in the optimization direction over the main node to be coordinated or propagated unchanged on arrival of the vehicle body during the green time, which corresponds graphically in a time-path diagram pushing through the main direction vehicle pulse. The vehicle spools are used as they were sent by the Vorknoten. This avoids the modeling inadequacy that resulted from the assumption that the shape of the approaching vehicle pulse was assumed to be equally distributed at each primary node to be co-ordinated. The shape of the vehicle pulse is therefore no longer dependent only on the green time distribution of the transmitting signal group of the last predecessor, but also on the penultimate or further precursor. Upon arrival of the vehicle pulse during the red period, this is further propagated by the signal group with two-stage intensity distribution from the green start.

In a preferred embodiment of the coordination method according to the invention, the intensity distribution which was modeled for a vehicle pulse sent by a pitch signal group of the penultimate forecast node and approaching the last pitch node during its red phase is sent as part of the main node from the main direction signal group of the last forward node Vehicle Pulse propagated. The secondary directional vehicle pulse accumulating in the optimization direction at the preselected node is further propagated from the start of green with high intensity, where it merges with the main direction vehicle pulse sent from the pre-node to the main node.

In a further advantageous embodiment of the coordination method according to the invention, two-stage intensity distributions with two sections of constant intensity of different heights or one-stage intensity distributions with only one section of constant intensity are provided for modeling the vehicle pulse moving against the optimization direction. Thus, a choice can be made between the known two-stage intensity distributions for main directional pulses, which are preferably used in accesses at the edge of the network, and an alternative one-level intensity distribution which better corresponds to the expected vehicle shape in a coordinated road and is therefore preferably used for such coordination units in the network. The width of the pulse shape is not dependent on the green time distribution, which is the case for both the high and low intensity sections when using the two-step shape.

Preferably, the intended intensity distributions differ by a time interval by which the constant-intensity section in a single-level intensity distribution or by the higher-intensity section of constant intensity in a two-level intensity distribution within the green period of Main direction signal group is shifted from green start. By taking into account coordination not only at green start and at end-of-turn which already provide good coordination results, but also intermediate solutions, the solution space of possible intensity distributions is advantageously extended, preferably for main direction and coordination pulse moving in the direction of optimization.

In a further advantageous embodiment of the coordination method according to the invention, the objective function is varied over a large number of possible phase sequences and / or over a plurality of intensity distributions with possible time intervals, in order to determine the optimum phase sequence and / or the intensity distribution with optimum time interval, in each case with optimal offset time. In addition to planning specifications, these intensity distributions can advantageously be applied automatically in order to determine an optimized coordination. Thus, the invention can be used not only to optimize the offset times alone, but also to optimize the phase sequences and / or the intensity distributions.

In another preferred embodiment of the coordination method according to the invention, the intensity distribution for a vehicle group transmitted by a signal group is modeled with at least one section of constant, maximum intensity whose value is less than the maximum possible intensity, in order to take into account a dispersion and / or deflections of a vehicle group.

The reduced intensity value corresponds more closely to the real or in reality expected, maximum intensity, which can be determined from the real or expected vehicle pulse approaching this signal group on the basis of real time pulse lengths. This allows a dispersion of in and against the particular Optimizing direction moving vehicle pulse and by turning data related effects of reality are modeled, which is particularly necessary for a correct offset time measurement. Otherwise there is a risk of modeling to narrow green bands.

FIG 1

die Topologie eines zu koordinierenden Straßenzugs,

FIG 2

Intensitätsverteilungen eines Fahrzeugpulks an einer Nebenrichtungs-Signalgruppe gemäß Stand der Technik,

FIG 3

Intensitätsverteilungen eines Fahrzeugpulks an einer Hauptrichtungs-Signalgruppe gemäß Stand der Technik,

FIG 4

ein Zeit-Weg-Diagramm mit gegen die Optimierungsrichtung koordinierten Knoten gemäß Stand der Technik,

FIG 5

ein Zeit-Weg-Diagramm mit in Optimierungsrichtung koordinierten Knoten gemäß dem erfindungsgemäßen Verfahren,

FIG 6

eine alternative Intensitätsverteilung eines Fahrzeugpulks an einer Signalgruppe bei angenommenem koordinierten Knoten,

FIG 7

Intensitätsverteilungen eines Fahrzeugpulks an einer Signalgruppe bei zu Grünbeginn und zu Grünende koordiniertem Knoten,

FIG 8

Intensitätsverteilungen eines Fahrzeugpulks an einer Hauptrichtungs-Signalgruppe bei zu Grünbeginn und zu Grünende koordiniertem Knoten,

FIG 9

ein Zeit-Weg-Diagramm mit gegen die Optimierungsrichtung koordinierten Knoten unter Verwendung der Intensitätsverteilungen nach FIG 7,

FIG 10

Intensitätsverteilungen eines sich gegen die Optimierungsrichtung bewegenden Fahrzeugpulks mit Berücksichtung einer Pulkdispersion,

schematisch veranschaulicht sind.Further features and advantages of the coordination method according to the invention will become apparent from an embodiment explained in more detail with reference to the drawings, in which

FIG. 1

the topology of a street to be coordinated,

FIG. 2

Intensity distributions of a vehicle body on a prior art line signal group;

FIG. 3

Intensity distributions of a vehicular pulse to a main direction signal group according to the prior art,

FIG. 4

a time-distance diagram with coordinated against the optimization direction nodes according to the prior art,

FIG. 5

a time-distance diagram with nodes coordinated in the direction of optimization according to the method according to the invention,

FIG. 6

an alternative intensity distribution of a vehicle body on a signal group assuming a coordinated node,

FIG. 7

Intensity distributions of a vehicle pulse at a signal group at green start and at the end of green coordinated node,

FIG. 8

Intensity distributions of a vehicle pulse at a main direction signal group at green start and at green end coordinated node,

FIG. 9

a time-path diagram with nodes coordinated against the direction of optimization using the intensity distributions FIG. 7 .

FIG. 10

Intensity distributions of a vehicle pulse moving against the direction of optimization, taking into account a pulse dispersion,

are illustrated schematically.

The coordination method according to the invention is not restricted to a specific network topology or to special signaling at the node. For the sake of simplicity, however, the following explanations will refer to the topology of a network section FIG. 1 basis, according to which the motor vehicle traffic along a road is controlled by at three consecutive nodes 1, 2, 3 arranged light signal systems. The main traffic direction runs along this road, on the drawing sheet from bottom to top or from top to bottom, with side branches defining secondary traffic directions at the nodes 1, 2, 3 turn vehicles in the main traffic direction and turn it. Light signal systems with two- or three-pronged signal generators, which are grouped into signal groups and controlled by a running in a node control unit signal program with a traffic-dependent or daytime-dependent signal timing plan, are known per se and in FIG. 1 not fully illustrated.

At the first node 1, the traffic in the main traffic direction is controlled by a signal group H1 having a signal time plan SZP _H1 and in the secondary traffic direction by a signal group N1 having a signal time plan SZP _N1 . Accordingly, the traffic at the second node 2 or at the third node 3 is controlled by signal groups H2 and H3 with signal time schedules SZP _H2 and SZP _H3 for the main traffic direction and signal groups N2 and N3 with signal time plans SZP _N2 and SZP _N3 for the secondary traffic directions. In the following, a uniform circulation time t _{U is} assumed for all signal time schedules, so that the best possible offset times between their signal time schedules, ie their relative start time shifts, are to be determined in order to coordinate the traffic light systems.

For this purpose, a so-called optimization plan is first defined, which specifies the order in which the nodes 1, 2, 3 are included in the optimization. For example, the optimization plan "1 - 2 - 3" can be specified: After optimization of node 1, the optimization of node 2 and then finally the optimization of node 3. Each node 1, 2, 3 is optimized individually.
To optimize the offset time, the procedure is as follows: For example, if the node 1 is already optimized, ie the optimal timing position of the signal time plan SZP _H1 or SZP _N1 set, so is an offset time-dependent objective function, for example, enter the respective holding and waiting times, by variation minimizes the offset time between the coordinated signal time plan SZP _H1 and the signal timing plan SZP _H2 to be coordinated.

According to the prior art described at the outset, the effects are determined only at the immediately adjacent, already coordinated, preliminary nodes 1. When optimizing the main node 2, therefore, on the one hand, the effects on the Vorknoten 1 by the main node 2 to Vorknoten 1 - contrary to the optimization direction OPT - moving vehicles and on the other hand, the effects on the Hautpknoten 2 through which the Vorknoten 1 to the main node 2 - in the optimization direction OPT - Moving vehicles considered.

In order to determine the effects of an offset time on the movements of the vehicles, distributions of the traffic intensities are modeled via a signal circulation of a vehicle pulse approaching a signal group and a vehicle pulse transmitted by the signal group. The vehicle pulse from the Vorknoten 1 to the main node 2, for example, one or more signal groups H1, N1 sent to the Vorknoten 1. In the above-described prior art is according to FIG. 2 modeled for a Fahrzeugpulk from the secondary directions N a constant intensity during the green time of the transmitting signal group N and a negligible intensity during the red time, namely for the intensity distributions of both before i _N ^- and after i _N ⁺ of the signal group N. In the signal schedule SZP _N a signal circulation is shown which starts with the time unit t at 0 and ends with the circulation time t _U ; the green period is marked by an empty bar between green start t _Gb and end of green t _Ge , the rest of the red period by a bar with a dash. The intensity distribution i _H ^{- of} the approaching in the main traffic direction of the signal group H vehicle pulse is in accordance with FIG. 3 is modeled as constant throughout the orbital period t _U ; the intensity distribution i _H ⁺ after the signal group H is modeled with a high intensity i _h at the level of the saturation traffic intensity starting at green start t _Gb and then with a lower intensity in for the remaining green time until end t _Ge .

In the time-distance diagram according to FIG. 4 the intensity distributions for vehicle pulse resulting in each case from an offset time optimized according to the prior art are shown against the optimization direction OPT. The intensity distribution of a vehicle body is modeled from node 2 to node 1. Here, the two-level intensity distribution for a main directional pulse is determined according to FIG. 3 the hatched area marks the high intensity portion i _h , the unshaded area the low intensity portion i _n . According to this model, the offset time of the signal time chart SZP _{H2 is} optimized for the signal time schedule SPH1. Subsequently, the intensity distribution of a vehicle body is modeled from node 3 to node 2 in order to optimize the offset time of the signal time schedule SZP _H3 to the signal time schedule SP _H2 . The vehicle cluster from node 3 to node 2 has a different location at node 2 than the vehicle cluster from node 2 to node 1. Assuming that node 3 is the first in the street, that corresponds to Modeled intensity of the vehicle pulse from node 3 to node 2 largely reality, but not the modeled intensity of the vehicle body from node 2 to node 1. Extend mentally the high-intensity component i _{h of} the vehicle body from node 3 to node 2, this yes during Green time at node 2 happens to node 1, so you can tell that most of the vehicles in this vehicle cluster arrive at node 1 during its off-time - ie have to stop. This problematic modeling of the vehicle pulse is used in the aforementioned prior art both in and against the optimization direction.

According to the invention, this error modeling in the optimization direction is avoided by "shifting" the intensity distribution: after optimizing the offset time at node 1, the intensity distributions of the vehicle pulse are modeled from node 1 to node 2. After optimization of the offset time at node 2, the intensity distributions of the vehicle pulse are then modeled from node 2 to node 3, the intensity distribution no longer being present in front of signal group H2 at node 2 (as in FIG FIG. 3 ) is assumed to be equally distributed over the orbital period t _U , but according to the intensity distribution modeled from node 1. Accordingly remains FIG. 5 the main directional pulse from node 1 through node 2 to node 3 is unchanged in the form of its intensity distribution. The main directional pulse of node 1 with its high intensity i ₁ from green start and its low intensity i ₂ to green end passes node 2 in green and remains unchanged in shape from node 2 to node 3 with the two stepped areas of high intensity i ₄ and then low intensity i ₃ . The tributary pulse of node 1 of low intensity i _{5 appears} at node 2 during its red-time and is then propagated there as a high-intensity region i ₆ from node 2 to node 3. The intensity ranges i ₃ , i ₄ and i ₆ merge into a new main direction pulse, only its intensity components i ₃ and i ₄ from the main direction come. At node 3, the accumulated portions of the main direction pulse i ₃ , i ₄ and i ₆ merge with the intensity i _{7 of} the secondary direction pulse from node 2 to node 3 to a new vehicle pulse with an intensity distribution, not shown. The individual intensity profiles are folded to illustrate each as a cross section to the left in the drawing plane.

The intensity distributions of the vehicle pulse can not be "pushed through" against the optimization direction OPT. The intensity distributions of the incoming vehicle pulse are not known, because the nodes of the required signal groups are not yet optimized. Accordingly, uniformly distributed intensities in front of the signal groups continue to be used here.

The accuracy of the pulp modeling is improved according to the invention by evaluating, when optimizing a node, how well the intensity distributions of the vehicle pulse arriving at the adjacent, already optimized nodes match those intensity distributions previously used for their optimization. In the illustrated embodiment, the main node 3 is optimized so that the intensity distributions of the vehicle node arriving from the main node 3 at the last Vorknoten 2 as possible correspond to the intensity distributions of the vehicle pulse previously modeled for vehicle pulse, which were sent from the last Vorknoten 2 to its Vorknoten 1 to to optimize the offset time at node 2.

A simple embodiment for compliance with the intensity distribution of incoming vehicle pulse is a modification of the target function PI. According to the prior art mentioned at the outset, this consists of a weighted sum of waiting times w _s (t) and holding h _s (t) per signal group s at the node with a total of S signal groups, for example H and N, over a signal revolution t from 0 to t _U -1, where α _s (eg 0.01) and β _s (eg 0.8) are weighting factors with, for example, the values given in parentheses: $$\mathrm{PI}\mathrm{=}\underset{\mathrm{t}\mathrm{=}\mathrm{0}}{\overset{{\mathrm{t}}_{\mathrm{U}\mathrm{-}\mathrm{1}}}{\mathrm{\Sigma }}}\underset{\mathrm{s}\mathrm{=}\mathrm{1}}{\overset{\mathrm{S}}{\mathrm{\Sigma }}}\left({\mathrm{\alpha }}_{\mathrm{s}}\mathrm{\cdot }{\mathrm{w}}_{\mathrm{s}}\left(\mathrm{t}\right)\mathrm{+}{\mathrm{\beta }}_{\mathrm{s}}\mathrm{\cdot }{\mathrm{H}}_{\mathrm{s}}\left(\mathrm{t}\right)\mathrm{+}{\mathrm{\gamma }}_{\mathrm{s}}\mathrm{\cdot }\left({\mathrm{G}}_{\mathrm{s}}\left(\mathrm{t}\right)\mathrm{\cdot }{\stackrel{\mathrm{~}}{\mathrm{H}}}_{\mathrm{s}}\left(\mathrm{t}\right)\mathrm{+}{\mathrm{H}}_{\mathrm{s}}\left(\mathrm{t}\right)\right)\right)$$

_s(t) gibt die Anzahl von Fahrzeugen zur Umlaufzeiteinheit t an der Signalgruppe s an, die ohne Halt die Signalgruppe s durchfahren kann, sinngemäß also die Zahl der Fahrzeuge, die nicht halten müssen: $${\stackrel{\mathrm{‾}}{\mathrm{h}}}_{\mathrm{s}}\left(\mathrm{t}\right)\mathrm{=}{\mathrm{c}}_{\mathrm{s}}\left(\mathrm{t}\right)\mathrm{-}{\mathrm{h}}_{\mathrm{s}}\left(\mathrm{t}\right)$$

Hierbei ist c_s(t) die Anzahl der zur Umlaufzeiteinheit t an der Signalgruppe s ankommenden Fahrzeuge. g_s(t) modelliert eine linear steigende Funktion, die bei Grünbeginn t_Gb den Wert 0 und bei Grünende t_Ge den Wert 1 annimmt, wenn bei Grünbeginn t_Gb koordiniert wird, und eine entsprechend linear fallende Funktion, wenn bei Grünende t_Ge koordiniert wird.According to the invention, the target function PI is extended by an additional term multiplied by the weighting factor γ _s (eg 0.8) per signal group s and circulating time t, which indicates the modeling accuracy of the respective vehicle pulse. ${\stackrel{\mathrm{~}}{\mathrm{H}}}_{\mathrm{s}}\left(\mathrm{t}\right)$

_s (t) indicates the number of vehicles at the orbital period t at the signal group s, which can pass through the signal group s without stopping, analogously therefore the number of vehicles that do not have to stop: $${\stackrel{\mathrm{~}}{\mathrm{H}}}_{\mathrm{s}}\left(\mathrm{t}\right)\mathrm{=}{\mathrm{c}}_{\mathrm{s}}\left(\mathrm{t}\right)\mathrm{-}{\mathrm{H}}_{\mathrm{s}}\left(\mathrm{t}\right)$$

In this case, c _s (t) is the number of vehicles arriving at the signal group s for the circulating time unit t. g _s (t) models a linearly increasing function, which assumes the value 0 at green start t _Gb and the value 1 at green end t _Ge , if t _{Gb is} coordinated at green start, and a correspondingly linear falling function, if co-ordinates at end of green t _Ge becomes.

Alternatively, instead of a linear function, an inverted intensity distribution of the pulse intensity distribution to be maintained can be taken: $${\mathrm{G}}_{\mathrm{s}}\left(\mathrm{t}\right)\mathrm{=}{\mathrm{i}}_{\mathrm{Max}\mathrm{.}\mathrm{s}}\mathrm{-}{\mathrm{i}}_{\mathrm{s}}{}^{\mathrm{+}}\left(\mathrm{t}\right)$$

Here, i _{max, s are} the maximum intensity at the signal group s, which corresponds to the main direction pulse of the high intensity i _h at the green start t _Gb . i _s ⁺ (t) indicates the intensity distribution according to the signal group s.

A further advantage results if, instead of the standard profile, a profile with only one pulse intensity i _{k is} optionally selected for the direction of optimization FIG. 6 is used. The underlying vehicle pulse is called a coordination pulse because its intensity distribution better corresponds to the expected shape of a coordinated node where the intensity before and after the transmitting signal group from green start t _{Gb has} a constant value i _k for a certain time, which then - still during green time - drops to zero when all the vehicles in the Coordination Pack have passed the node. For a given traffic volume, the width of the coordination pulse is not dependent on the length of the blocking period. In contrast, when using the intensity distribution for the main directional pulse according to FIG. 3 the widths of both the high-intensity region i _h and the low-intensity region i _{n are} dependent on the green-time distribution.
The Coordinate Pulse and the Main Direction Pulse will normally start at green start t _Gb , as in FIG. 7 respectively. FIG. 8 each in the upper picture of the associated intensity distributions i _{Gb is} shown. However, this greatly restricts the solution space, since only solutions in which the coordinated vehicle cluster passes through _Gb at the start of greening, if possible, are allowed against the optimization direction OPT. According to the invention, it can therefore be provided to allow intensity distributions for the coordination pulse and for the main direction pulse, which have different time intervals Δt of green start t _Gb . The lower pictures in FIG. 7 respectively. FIG. 8 show intensity distributions i _{Ge of} the coordination pulse and the main directional pulse, in which the intensity distribution or the high-intensity region are shifted to the end of the t _Ge . Merely the selection of these two types of intensity distributions i _Gb and i _Ge provides very good coordination solutions. In principle, however, a plurality of intensity distributions with time intervals Δt of different lengths can be included in the solution space. At the second resolution there are at most as many different time intervals Δt, such as the existing green time minus the time width of the coordination pulse or the high-intensity region of the main directional pulse, measured in seconds. The application of the different intensity distributions can either be done via planning specifications or automated, in that the optimization method evaluates all permitted time intervals Δt for all phase sequences at the best offset time and that selects the best combination of phase sequence and time shift.

In the time-distance diagram according to FIG. 9 FIG. 3 shows a coordination structure from node 3 via node 2 to node 1, in which there is always a coordination pulse with an intensity distribution according to FIG FIG. 7 has been used. At node 2, a coordination was carried out with an intensity distribution i _Ge shifted to the end of the earth t _Ge . While the known from the prior art solution according to FIG. 4 for most vehicles the main traffic direction gives a stop, the generated in FIG. 9 solution shown does not stop against the optimization direction OPT. Also in optimization direction OPT must after FIG. 5 keep the fewest vehicles.

The intensity of a vehicle cluster coordinated over several nodes decreases from traffic signal to traffic signal system, since a dispersion of the vehicle body can be detected by vehicles moving at different speeds and by vehicles turning off the road to be coordinated. In the optimization direction OPT both effects are imaged when "pushing through" the intensity distribution of the vehicle pulse. The intensity distributions of the vehicle pulse are again and again set against the optimization direction OPT, whereby initially the maximum intensity is always assumed. So that the pulse dispersion can also be imaged here, the maximum intensity i _{max is determined} according to FIG. 10 percentage - for example, to 75% - reduced to the actual Reproduce pulse lengths. The reduction of the maximum intensity i _max results in a time extension of the vehicle body. This avoids to model the intensity distributions of the vehicle pulse compared to the reality spatially and temporally too short. Thus, the green bands are not determined too short. The determination of the intensity reduction can be carried out automatically from the perspective, for example by taking into account the turning rates or by planning specifications.

Claims (7)

1. Method for coordinating light-signal-controlled nodes (1, 2, 3) in a road network, wherein the vehicle traffic at a node (1, 2, 3) is controlled by at least one signal group (H, N) of a light-signal system, wherein a phase sequence of red phases stopping and green phases releasing the vehicle traffic is specified in a signal timing plan (SZP_H, SZP_N) of the light-signal system for the signal group (H, N) within a cycle time (t_U) by predefinition of green start (t_Gb) and green end (t_Ge), wherein an order of the nodes (1, 2, 3) and hence an optimisation direction (OPT) are specified in an optimisation plan, with the nodes (1, 2, 3) being coordinated in said optimisation direction (OPT), wherein an optimal offset time between the signal timing plans (SZP_H, SZP_N) of the main node (2, 3) to be coordinated in each case and a previously coordinated preceding node (1, 2) is determined by minimising a target function in the form of a weighted sum of wait times and numbers of stops of vehicles of vehicle platoons moving in and/or against the optimisation direction (OPT) between the last preceding node (1, 2) and the main node (2, 3), wherein the waiting times and numbers of stops are dependent on the phase sequences of the signal timing plans (SZP_H, SZP_N) of the main node (2, 3) and the last preceding node (1, 2), on the offset time between said signal timing plans (SZP_H, SZP_N), and on intensity distributions modelling the vehicle platoons, wherein an intensity distribution (i_H ^-, i_N ^-) of a vehicle platoon approaching the signal group (H, N) and an intensity distribution (i_H ⁺, i_N ⁺) of a vehicle platoon sent by the signal group (H, N) are modelled per signal group (H, N),
   characterised in that in the course of determining the optimal offset time for the main node (3) an assessment is made as to how well an intensity distribution which is modelled for a vehicle platoon approaching the signal group (H2) of the last preceding node (2) against the optimisation direction (OPT) corresponds to the intensity distribution which was modelled in the course of coordination of the last preceding node (2) for a vehicle platoon sent by the signal group (H2) of the last preceding node (2) against the optimisation direction (OPT) to the next-to-last preceding node (1), and/or how well an intensity distribution which is modelled for a vehicle platoon approaching the signal group (H3) of the main node (3) in the optimisation direction (OPT) corresponds to the intensity distribution which will be modelled in the course of coordination of a next succeeding node for a vehicle platoon sent by the signal group (H3) of the main node (3) in the optimisation direction (OPT).
2. Coordination method according to claim 1, wherein the intensity distribution which was modelled for a vehicle platoon sent by a main direction signal group (H1) of the next-to-last preceding node (1) to the last preceding node (2) is propagated further while taking into account the signalling of the main direction signal group (H2) of the last preceding node (2) for a vehicle platoon sent to the main node (3).
3. Coordination method according to claim 1 or 2, wherein the intensity distribution which was modelled for a vehicle platoon sent by a subsidiary direction signal group (N1) of the next-to-last preceding node (1) and approaching the last preceding node (2) during its red phase is propagated further as part of the vehicle platoon sent by the main direction signal group (H2) of the last preceding node (2) to the main node (3).
4. Coordination method according to one of claims 1 to 3, wherein two-stage intensity distributions having two sections of constant intensity (i_h, i_n) of different height or one-stage intensity distributions (i_k) having only one section of constant intensity are provided for modelling moving vehicle platoons.
5. Coordination method according to claim 4, wherein the intensity distributions provided differ by a time interval (Δt) by which the section of constant intensity in a one-stage intensity distribution or by which the section of constant intensity of higher value in a two-stage intensity distribution is shifted from green start (t_Gb) within the green time of the main direction signal group (H).
6. Coordination method according to claim 5, wherein the target function is varied over a multiplicity of possible phase sequences and/or over a multiplicity of intensity distributions with possible time intervals (Δt) in order to determine the optimal phase sequence and/or the intensity distribution with optimal time interval at an optimal offset time in each case.
7. Coordination method according to one of claims 1 to 6, wherein in order to take account of a dispersion and/or of right-/left-turning vehicles leaving a vehicle platoon the intensity distribution (i⁺) for a vehicle platoon sent by a signal group is modelled with at least one section of constant maximum intensity (i_max, _red) whose value is set lower than the maximum possible intensity (i_max).

Images