EP2492886A1 - Method and light signal assembly control system for controlling light signal assemblies - Google Patents

Abstract

The method involves performing a distribution of green times to signal groups e.g. main signal group (SG1), of signalized points by a candidate selection process according to a number of predetermined decision criteria. The distribution of the green times is carried out at predetermined time intervals in seconds. The predetermined decision criteria are processed in accordance with associated priorities during the distribution of the green times. A timing element and/or hysteresis element are provided during the determination of a node-circulation time and/or control range rotation time. Independent claims are also included for the following: (1) a light signal assembly control system for traffic control at signalized nodes in a control region (2) a computer program product having a set of instructions for performing a method for controlling a light signal system for road traffic on signalized nodes in a controllable region.

Description

The invention relates to a method and a traffic signal control system for controlling traffic light systems at one or more signalized nodes in a control area. In particular, it relates to a method in which all road users, such as private transport, public and private transport, as well as pedestrians and cyclists can be taken into account depending on the traffic situation, and allows an optimization of the traffic flow at the signaled nodes in a control area with regard to the selection of the relevant Manipulated variables, such as the phase sequence, the rotation time selection, the green time distribution and the offset time calculation. However, the method can also be used for the planning optimization of such methods on the basis of previously determined traffic data.

Traffic control systems of the more complex type are now used on busy roads, especially on highways and in highly frequented inner-city routes such as ring roads and major city streets. They usually have a large number of traffic control actuators. This is understood as meaning all traffic control instruments which pass on traffic rules and traffic information to traffic participants by means of signaling. In particular, this includes traffic light systems, traffic signs that variably or statically display such rules and instructions, but also instruments such as traffic information or traffic control via remote control of navigation systems.

There are clear differences between the local application and the non-local application. Municipal traffic control systems are mainly on it via an urban traffic management system to effectively control or direct urban traffic flows via a traffic signal system. In contrast, off-site, especially motorway traffic control systems, usually have a significantly larger number of different actuators for controlling traffic flows. These include, but are not limited to, variable traffic control indicators that may represent speed limits, overtaking bans, speed bids, safety alerts, weather information, and other distance or environmental indicators.

Inner and extra urban traffic control systems also often have directional arrows that can lock or unlock certain lanes for traffic or indicate a necessary lane change. In addition to this, diversions can be signposted and congestion can be seen, which can include mobile displays in addition to permanently installed scoreboards.

Methods for the control of traffic light systems as a function of the traffic volume are known and are becoming increasingly important. In this case, the green time is often measured, ie a needs-based adjustment of the release time (= green time) determined. In this case, traffic-dependent release time adaptation has hitherto mainly dealt with time gaps between the vehicles and occupancy levels in the access roads to the junctions, with fixed threshold values frequently being defined. Exceeding or undershooting these threshold values, or even combinations of these input variables, leads to switching off or extending the current traffic flow. In the release time adjustment by means of time gap measurement, the time intervals between successive vehicles of a vehicle current are measured as a time gap via a detector in the node approach. The release time is adjusted to the current needs of the incoming vehicles after the selected minimum release time or after reaching an earliest time in circulation. The release time may be extended until the measured time gap is at least as large as a predetermined timeout threshold or until the longest committed release time or time limit is reached in the recirculating signal.

With the release time adjustment via occupancy rate measurement, threshold values for canceling the release time are also defined. The evaluation of the measured values is done separately for each lane. The original occupancy rate is smoothed using a compensation procedure. If necessary, a compensation factor for rising and falling tendencies of the original values is used.

In the so-called "Volume Density Method", the number of vehicles standing in front of red is compared with the time gaps in current traffic. A limit curve is defined, which assigns a fixed time gap value to a fixed number of waiting vehicles, up to which the current release time is still extended. In an advanced method, the green-time calculation can be done on an urgency basis. The number of vehicles standing in front of red is compared here with the time gaps and additionally with the occupancy in current traffic. Again, thresholds are used.

In the WO 97/34274 A1 , an equivalent to that EP 0 886 845 A1 , a method for traffic-dependent control is described in which the current traffic condition is assessed by means of fuzzy logic. After weighing the different interests of the competing traffic flows, a decision is made regarding the extension of the current phase or the termination and a selection of the next phase. This is done according to the method described in this document depending on the given possibilities by the current signal / master plan. In this case, the use of fuzzy logic in the area of node control allows the previously used hard threshold values to be replaced by sliding transitions and thus For example, a more differentiated consideration of the input values is possible (cf. Xiaohong, Peng et al., "Twostage fuzzy logic controller based on variable phase sequence for urban traffic intersection" in IEEE International Conference on Measuring Technology and Mechatronics Automation, pp. 610-613, April 2009 ).

US 2009/0322561 A1 describes a method for controlling signaling systems based on data from the current traffic situation and its development over time. In this case, the method includes the calculation of the green time required to resolve a backlog.

WO 2010/040649 A2 describes a traffic adaptive network controller and method for optimizing the control parameters based on a genetic algorithm.

Huang, Bo et al., "Modeling of urban traffic systems based on fluid stochastic petri nets," IEEE Fourth International Conference on Natural Computation, pp. 149-153, October 2008 describe a hybrid model for traffic system controls, and more particularly, an enhanced Petri-Net model based traffic light control.

DE 10 2005 023 742 B4 describes a method for controlling a traffic network using a selection of signal groups, a variation of green times, a variation of switching points and a selection of control strategies.

A simple example of a model-based method is a green time estimation method in which the utilization level of a current signal group is compared with the utilization level of an enemy red signal group.

However, all of these methods have their limitations, especially in terms of algorithmic approaches due to missing or too simple control models. Further limits result from a strong restriction of traffic relevant control scenarios, eg that a signal group, if it occurs in several phases one behind the other, can only be extended in the last phase, but not in the first phase, among which possibly other signal groups would need the second most important additional green do not get this.

More recent approaches optimize the green time distribution, round-trip time selection and offset time optimization as well as possibly the selection of the phase sequence in a closed control model simultaneously or in parallel. The disadvantage of this is that the optimization problem NP is complete, i. with known non-heuristic optimization methods, the system optimum can not be found in finite time. Therefore, it is necessary to resort to heuristic optimization methods, such as "Simulated Anealing" or taboo search or genetic algorithms as described by Braun in "A Real-Time Evolutionary Algorithm for Network-Wide Optimization of Traffic Light Control".

However, these methods also have their limitations, in particular with regard to the required computing time and the number of manipulated variables to be taken into account, as well as algorithmic limits due to the complexity of the methods and the fact that finding the system optimum can not be guaranteed. Thus, in the methods listed above, the respective manipulated variables, such as the phase sequence, the round trip time selection, the green time distribution and the offset time calculation, are optimized jointly or separately from one another.

In contrast to the parallel or simultaneous optimization in an optimization method, the advantage of a separate optimization in independent optimization processes, which are generally consecutive in time, is that the optimization complexity of the individual optimizations is reduced to such an extent that they are no longer NP-complete and the system optimum of the optimization respective individual task can not be found in finite time.

An example of such a process consisting of individual optimizations which uses a gradient descent method for green time optimization is the method developed by Siemens, which is also known under the name "Motion 4.x". However, since several manipulated variables, namely the green-time components of the phases, are optimized in the green-time distribution, there is still a multidimensional solution space. The solution must therefore be optimized in several process steps by means of iterations. This generally leads to a long computing time. In addition, the process is not very transparent (so-called black box character) and boundary conditions can be considered only to a limited extent in the form of additional target function terms.

Besides this method, model-based methods for green time optimization are known which perform the green time distribution in units of time, e.g. second increments. This reduces the complexity of the optimization to only one dimension, namely the green time to be distributed. These can in turn be divided into methods that require a number of iterations to find a solution in order to approximate the optimal solution taking into account boundary conditions, and methods that achieve the optimal green time distribution in an iteration. An iteration here means that each time unit to be distributed has been distributed once. An iteration therefore consists of a number of sub-iterations corresponding to the number of time units to be distributed.

Examples of methods with multiple iterations are the so-called lock group method, in which enemy signal groups are searched, and the method according to the Manual for the design of road traffic situations (HBS). However, these methods with multiple iterations require considerable and previously unidentifiable computing time, are also not very transparent (black box character), boundary conditions can usually only be considered implicitly, and the procedures Often also subject to algorithmic restrictions, eg with regard to the supported signal configurations.

Procedures for green time distribution with only one iteration for green time distribution are, for example, the "Motion 2.x" and "Motion 3.x" methods. Due to the simple calculation method, these require a lower computing time and boundary conditions can be met more easily, but they also have disadvantages. In particular, they can not handle all signal constellations, e.g. Signal overlaps including an inner multi-phase, optimally take into account, and compliance with the rules is not transparent, i. This procedure also has a black box character. An inner multi-phase is given when two signal groups are released together and also one of the two signal groups in the flow and the other of the two signal groups is enabled in the wake.

In addition to the above-described methods for green time distribution in such traffic control systems, the round trip time selection in the signaled nodes and the entire control range affects the performance of the signaled nodes. Thus, in principle, a change in the cycle time is associated with switching losses, if the traffic control systems have a mixed centralized and decentralized architecture. This is due to the fact that in the control units of these systems must be switched between different signal plans. The switching losses result from the fact that the signaled nodes must be synchronized. However, due to the synchronization, they change their wave position. This usually leads to additional holding of road users during the switchover.

Thus, various methods are known for determining the circulation time, which will be briefly explained here for understanding.

In an exemplary method, the round trip time can be determined by means of an evaluation of waiting times and holding in the coordinated overall system in accordance with a holistic optimization. The disadvantage of this is that the choice of the orbital period is less comprehensible and subject to fluctuations due to the dependence on the coordination.

In another method, the round trip time selection based on the critical signal group, ie the signal group in a control range, which has the highest utilization. The orbital time selection is thus interpretable, but still subject to strong fluctuations. This is partly due to the fact that the traffic demand on the signal groups can vary greatly, and secondly because the most heavily used signal group in the control area can change rapidly. This method is used for example in "Motion 4.x".

It is therefore an object of the present invention to provide an improved alternative to the previous traffic control methods and a traffic signal control system therefor.

This object is achieved on the one hand by a control method according to claim 1 and on the other hand by a traffic signal control system according to claim 14.

In the method according to the invention for controlling traffic light systems at one or more signalized nodes in a control area, the distribution of green times to the respective signal groups of the individual signalized nodes is carried out by means of a candidate selection method according to a number of predetermined decision criteria, preferably several different decision criteria.

1. 1. Berechnung der minimalen Umlaufzeit je Knoten bezüglich Randbedingungen und Auslastung für alle Knoten des Regelbereichs;
2. 2.Auswahl der gemeinsamen Umlaufzeit des Regelbereichs;
3. 3. Grünzeitverteilung für alle signalisierten Knoten für die gewählte Umlaufzeit des Regelbereichs gemäß dem vorstehend erläuterten Verfahren mit einer Iteration.

For this purpose, the method according to the invention can be based on a general method for green-time distribution, in which a common cycle time of a control area, the so-called control area round trip time, is taken as the basis for the green time distribution of each signaled node. The basic procedure of this general method consists of the following three steps:

1. 1. Calculation of the minimum round trip per node in terms of boundary conditions and utilization for all nodes of the control area;
2. 2.selection of the common cycle time of the control area;
3. 3. Green time distribution for all signalized nodes for the selected round trip time of the control range according to the method explained above with an iteration.

An advantage of this general method is that the round trip time depends on the utilization levels of all signal groups of a signaled node. This is a spatial averaging.

According to the invention, a node is defined as an area in traffic on which hostile traffic flows meet, in particular intersect or terminate. Examples of such nodes are intersections, pedestrian crossings, junctions, etc. One speaks of a signalized node when one or more traffic streams are controlled at this node by a signaling system, in particular a traffic signal system. Here, the term light signal is not limited to light signals, but it is also possible that the light signals are supplemented by other signals such as acoustic signals. Therefore, these are also named as signalers below. Especially in the field of pedestrian signals acoustic and optical signals are often used in combination, for example, to facilitate the use of visually impaired people or to achieve by a double signaling increased attention among road users.

A control area usually comprises more than one signaled node and serves to control the traffic streams at two or more nodes consecutive in the traffic flow. In particular, the control areas serve to increase the traffic flow in this. Often, a "green-wave circuit" is then spoken of, which basically represents an optimal case of traffic control. A simple embodiment of such a traffic control selects the pre-defined signal programs and thus also the cycle time in a control area daytime-dependent, since the traffic flows usually vary greatly with the time of day. For example, there is often a heavy traffic in the morning in the direction of the city center, while in the evening the traffic flows in the opposite direction, ie out of town. A control area can include an entire city, a part of a city or just main traffic arteries or their parts.

In order to achieve an improvement or optimization of the traffic flow in a control range, a traffic-dependent distribution of the green times to the respective lane groups or signal groups of the individual signalized nodes is sought in the inventive method instead of the daytime-dependent selection of signal programs. Here, the distribution of green times, which will be explained in more detail below, a method is used, which can respond to variable traffic flows relatively quickly. As a signal group is understood according to the invention a group of light signals, each having identical signaling. For example, in multi-lane roadways next to side-mounted signalers often signalers are mounted on the lanes, each representing a signal generator on the road with a side signal generator a signal group. Generally speaking, one always speaks of signal groups, even if there is only one signal generator. Lane groups are groups of lanes that have homogeneous signaling, that is each lane is controlled by the same signal groups. Not signal Disseminated Lanes of a node access are also combined to lane groups to which no signal is assigned. Signal groups can also be assigned as main signal groups to several lane groups. In addition to a main signal group, a lane group can also be assigned an additional signal, so that the lane groups then do not differ in the main signal group but in the additional signal group. By way of example, a node access may have three lanes, two of which are straight ahead and one right. The two lanes straight ahead is a dreifeldiges main signal with full discs to be sorted, the Rechtsabbiegefahrstreifen the same dreifeldige main signal and in addition a zweeldiges additional signal with Rechtsabbiegepfeilen. This results in two lane groups. The first consisting of two lanes straight from which only the main signal is assigned. The second consisting of the lane to the right, to which the main signal and the additional signal is assigned. The lane groups are not explicitly defined in the German-language literature, but they are defined in the "Highway Capacity Manual", which among other things defines traffic signal planning and care in the USA.

The green time distribution is performed by a specific candidate selection method in which a candidate, for example a specific phase of a signaled node or a special signal group, is specifically selected for the respective green time interval to be awarded on the basis of predetermined decision criteria in a simple and, in particular, comprehensible manner. A phase in the sense of the invention is a specific signaling state of a specific node, which is defined by released signal groups. According to the invention, the selection is made for each decision criterion with the aid of an exclusion method, in which more and more candidates fall out due to their previously determined decision value by a predetermined grid. If no clear selection is possible when going through the procedure for a decision criterion, the procedure becomes for a further decision criterion on the basis of the remaining candidates continued until only one candidate for the green time interval to be distributed remains. This is then assigned the green time and the process starts again from the beginning for the next green time interval to be distributed.

By selecting a single candidate in the candidate selection method used according to the invention by means of the above-explained targeted selection on the basis of one or more decision criteria, an improvement of the traffic flow to be controlled, i. an improvement of the traffic quality at the nodes, can be achieved. In addition, this candidate selection process has a high degree of transparency and traceability compared to the previous methods. In particular, the targeted selection of a single remaining candidate based on one or more decision criteria increases the transparency and traceability of the control in comparison to the previously used procedures, which usually regulated the green-time distribution on the basis of complex decision trees. Often, however, due to a lack of systematics, no clear selection of a single candidate was possible.

• eine Datenübermittlungsschnittstelle zur Übernahme von Verkehrsdaten aus einer Verkehrsüberwachungseinrichtung,
• eine Verkehrsdaten-Verarbeitungseinheit zur Ermittlung von Lichtsignalanlagen-Steuerungsdaten,
• eine Datenübergabeschnittstelle für die Lichtsignalanlagen-Steuerungsdaten an eine Lichtsignalanlagen-Steuerungseinheit. Die Verkehrsdaten-Verarbeitungseinheit ist dabei so ausgebildet, dass eine Verteilung von Grünzeiten auf Signalgruppen von einzelnen signalisierten Knoten unter Verwendung eines Kandidatenauswahlverfahrens gemäß vorgegebenen Entscheidungskriterien erfolgt.

An inventive traffic signal control system for traffic control at one or more signalized nodes in a control area accordingly has at least the following components:

• a data transmission interface for transferring traffic data from a traffic monitoring device,
• a traffic data processing unit for detecting traffic signal control data,
• a data transfer interface for the traffic signal control data to a traffic signal control unit. The traffic data processing unit is designed so that a distribution of green times on signal groups of individual signalized node below Use of a candidate selection process according to predetermined decision criteria takes place.

In such a traffic signal control system, the data transmission interface, the traffic data processing unit and the data transfer interface both as a standalone hardware and / or software components executed as well as be integrated together within an electro-technical processor module. They may be implemented in whole or in part on a computer of a traffic control system. In addition, the Datenübernahme- or the data transfer interface can be designed both as hardware in the form of input and output sockets or wireless interfaces of a device as well as in the form of software or as a combination of hardware and software components. Interfaces, for example, in the form of pure software interfaces and directly take over data from a control system, for example, if the light signal control device is arranged on the same computer as the control system. The interfaces can continue to be combined as an input / output interface.

The structure of the traffic signal control system in the form of software has the advantage of a quick and cost-effective implementation. For carrying out the method according to the invention, therefore, a computer program product which can be loaded directly into a processor of the computer device and has program code means in order to carry out all the steps of the inventive method described above is preferably used.

Such a software component can not be implemented only in a central control center. Instead, it can also be present, for example, in a decentralized conductor, ie a control unit which controls a plurality of adjacent nodes at a higher level, for example as a master control unit, in particular if no control center is present. In addition, the local use in the control unit is possible if the circulation time of the control areas is specified from the outside, or if the control unit may run in any round trip.

Further particularly advantageous embodiments and developments of the method according to the invention will become apparent from the dependent claims and the following description. In this case, the traffic signal system control system according to the invention or the computer program product can also be designed according to the dependent claims for the method.

The entire green time interval to be distributed can basically be assigned completely to a signal group. In a preferred embodiment of the method according to the invention, however, the distribution of green times in predetermined periods of time, particularly preferably in seconds, so that appropriate green-time or release intervals are obtained. A second green-time distribution is advantageous in terms of computational requirements and integration into existing systems.

The method according to the invention allows the use of several decision criteria, which can be taken into account in the candidate selection. In this case, it is advantageous to assign priorities to the individual decision criteria in order then to systematically process the decision criteria in accordance with the priorities assigned to them in the method for distributing the green times. By using a plurality of decision criteria and by prioritizing it, the method according to the invention has, among other things, the advantage over the previous methods that it can be decided relatively quickly and easily which candidate receives the green time to be distributed.

According to a preferred embodiment of the method according to the invention, the green times can be phase-oriented or barrier-group-oriented be distributed. "Phase-oriented" means, for example, that it is determined which phase or the signal groups assigned to it are enabled for a long time, ie "green". "Lock group-oriented" means that in a first step, it is calculated which lock group is needed most of the time and, in a second step, which of the signal groups contained in the lock group requires the most time. A blocking group consists of time-sequentially released signal groups, which are hostile to each other.

In the method, one or more decision tuples, preferably 2-tuples, are determined for each decision criterion for each phase or blocking group. A decision tuple consists of a reference object of the decision criterion, for example a specific signal group, and an associated decision value, which can be determined for example by predetermined arithmetic operations for each reference object. With the help of the decision tuple, the method can make the distribution of the green times on the individual phases or blocking groups by selecting the most favorable phase or blocking group by the above-explained candidate selection process is run so long until in a decision criterion only one candidate is left.

In the above procedure, for example, for each decision criterion, the decision tuples for each phase or lock group may be input to a decision matrix. This decision matrix can then be processed in rows or columns until the phase or blocking group is found which receives the respective green time interval. As a result, in particular the black box character of the previous method is resolved, which is often a major obstacle for customers to use adaptive traffic control method. It is also advantageous that the decision matrices of the individual decision steps are logged for the purpose of transparency and relatively easily evaluated and can be interpreted. Users can later evaluate the recorded or logged matrices line by line.

Another advantage of this method is that the computation time compared to the previous methods, for example compared to the gradient descent method, is significantly shorter.

As a possible alternative to the decision matrix, a decision tree could also be used in which the decision tuples are represented as branches in the decision tree. This is then progressively processed according to the same principles as the matrix. For example, in each decision node of the decision tree, the remaining candidates are decided. A simple embodiment comparable to a decision matrix would have a simple chain of decision nodes at which the remaining candidates are determined. Each decision node corresponds to one row of the decision matrix. As a result, a selection procedure would also be possible here in which, after several branching steps, only one candidate would remain, to which the respective green time interval would then be allocated.

In the above methods, the minimum phase duration, minimum signal group duration, maximum phase duration, maximum signal group duration, lane group performance or free green distribution, number of released signal groups, or instantaneous phase duration are preferably used as predetermined decision criteria. The decision criteria are particularly preferably prioritized in the sequence given above, in order to achieve a suitable adaptive traffic control of the signalized nodes in a control range, in which boundary conditions can be met selectively and likewise different signal constellations can be taken into account.

For example, in order to favor pedestrian phases, it may therefore be appropriate to use the number of released signal groups of a phase as a decision criterion. The instantaneous phase duration can also be used, for example, to achieve the most uniform possible green-time distribution for arbitrary multiple applications, such as double-strokes, in which a signal group in one revolution receives "green" twice or more times.

If the performance of the lane groups is a decision criterion, it may preferably be weighted differently for the allocation of the green time for the distribution of the green time for the allocation of the free green, as soon as the performance of the respective node is reached. This can be achieved for certain lane groups a greater release time. This increased release time, in turn, can be used to advantage by offset time optimization to enhance the green wave to major corridors, e.g. to prevent traffic from shifting to residential areas.

CAlternatively, the performance of the lane groups may also be determined by means of a degree of utilization or a performance index of the signal groups assigned to the lane groups, for example according to Webster (1958). Webster uses the wait time d as the optimization goal to be minimized. This is calculated according to Webster: $$\mathrm{d}\mathrm{=}\frac{\mathrm{C}\left(\mathrm{1}\mathrm{-}\mathrm{<figure-callout\; id="2"\; label="\lambda "\; filenames="imgf0002.png,imgf0003.png"\; state="\{\{state\}\}">\lambda </figure-callout>}\right)}{\mathrm{2}\left(\mathrm{1}\mathrm{-}\mathrm{\lambda x}\right)}\mathrm{+}\frac{{\mathrm{x}}^{\mathrm{2}}}{\mathrm{2}\mathrm{q}\left(\mathrm{1}\mathrm{-}\mathrm{x}\right)}\mathrm{-}\mathrm{0.65}{\left(\frac{\mathrm{<figure-callout\; id="2"\; label="C\; q"\; filenames="imgf0002.png,imgf0003.png"\; state="\{\{state\}\}">C\; </figure-callout>}}{{\mathrm{q}}^{}}\right)}^{{}^{\mathrm{1}}{\mathrm{/}}_{\mathrm{3}}}{\mathrm{x}}^{\left(\mathrm{2}\mathrm{+}\mathrm{5}\mathrm{\lambda }\right)}$$

C

C is the orbital period, q the traffic volume, x the load factor and λ the green time component of the orbital period. Those skilled in the art are aware of further loss time related quality criteria based on queuing models. These can be roughly distinguished into deterministic and stochastic queue models. Also approximate solutions, both Consider deterministic behavior as well as stochastic behavior are known, such as from Kimber, RH & Hollis, M., 1979. Traffic Lanes and Delays at Road Interjunctions, TRRL Report LR 909, Transport and Road Research Library, Crowthorne ,

In one embodiment of the method according to the invention, in a first step, a green time distribution for determining a node cycle time of the respective signalized nodes in a control range is made. Thereafter, in a second or later step, the previously determined node cycle times are used to select a control range cycle time of the entire control range.

In this procedure, a trend of a minimum node round trip time selection and / or control range round trip time selection can be taken into account for determining the node or control range round trip time. Thus, the control area round trip time can be adjusted in time, especially for a rapid increase in traffic volume, e.g. during the morning or evening traffic peaks.

Alternatively, or in combination, a timer may be considered. A timer is a counter that counts the number of times an event should or should be changed again until the node cycle time and / or the rule area cycle time are changed. Such an event may be, for example, that at a node because of an overload actually a longer node cycle time should be present, as it currently exists, or that could be switched back to a shorter node cycle time, ie it is a kind of "orbital change -Anforderungsereignis ". There are inhibiting counters that indicate whether it is allowed to change into a new control range round trip time, and there are blocking counters that indicate whether it is permissible to switch out of a current control range round trip time.

For example, in determining the node round trip time and / or the closed loop round trip time, a timer may be taken into account, during which a lower node and / or closed loop roundtrip time must at least be present for it to be implemented in the next steps of the control method, i. before switching to the new node round trip time or closed loop round trip time. As a result, it may be achieved that an adaptation of the corresponding round trip time may only take place if a lower round trip time has elapsed, at least during a predetermined time interval, e.g. during the last n minutes (n = 10-20, 12-18, or 14-16, especially n = 15) was sufficient. A similar timer can be used for higher round trip times. Other timers may be used to ensure that a round trip time, i. a node round-trip time or a closed-loop cycle time must be present for a certain time before it can be changed to a higher or a lower node round-trip time or closed-loop cycle time. This may also depend on whether the orbital period in the meter intervals has previously increased or decreased.

Alternatively or additionally, a hysteresis element can be used. For a hysteresis member, predetermined or arbitrary limits are used in combination with or in place of the timing elements for decision making.

If, for example, a higher control range round trip time has been selected when a minimum node round trip limit value is exceeded, this closed loop round trip time can only be switched to a lower closed loop round trip time if the minimum node roundtrip limit is significantly lower than that which must have been exceeded previously.

The advantage is that the usual fluctuations in traffic demand do not lead to fluctuations in the orbital period. Therefore, a simpler overall vote in the control area be achieved, which usually increases the flow of traffic.

As an alternative to the above procedure, a blocking group method can also be used for the control range circulation time selection. Thus, the trend is even better to consider than working with the phase-oriented selection process. This is due to the fact that the minimum possible node cycle times in the blocking group method are significantly lower than by means of the phase-oriented green-time distribution, because the blocking group method neither takes into account the duration of the phase transitions nor minimum phase duration nor other boundary conditions, except the intermediate times between hostile signal groups. This combination of a phase-based green time distribution method for determining the respective node cycle times and the blocking group method for determining the trend of the minimum node cycle times, a prediction value for the minimum node cycle time is determined, resulting in proactively higher round-trip times for the control area just in the morning peak. In addition, even with this extension, a sufficient level of transparency is maintained by the forecast for each node and its phase sequences can be logged with their characteristics. Thus, a good acceptance of the traffic planners is to be expected.

In the selection of the control range round trip time, the number of signaled nodes to be considered in a control range can be chosen to be planning, so that for the traffic subordinate or uncritical nodes, for example minor intersections or junctions on secondary lines, are irrelevant, since the non-selected nodes not included in the orbital time selection. Another example of an uncritical node would be, for example, a pedestrian crossing with a double attack, which would become computationally relevant if both accusations were active in each round. However, this is not the case in times of low pedestrian traffic.

FIG 1

ein Beispiel für einen signalisierten Knoten mit einer Hauptsignalgruppe SG1 und einer Zusatzsignalgruppe SG2 für die jeweiligen Fahrstreifengruppen FG1 bzw. FG2,

FIG 2

ein Schaltbild eines erfindungsgemäßen Lichtsignalanlagen-Steuerungssystems 1 mit einer Datenübermittlungsschnittstelle 10, einer Verkehrsdaten-Verarbeitungseinheit 20 und einer Datenübergabeschnittstelle 30,

FIG 3

ein Beispiel für eine phasenbasierte Entscheidungsmatrix mit einem Entscheidungskriterium, hier einem Block zur Verteilung von Grünzeitreserven C_res von Fahrstreifengruppen, welche in dem erfindungsgemäßen Verfahren ausgewertet wird, wobei im ersten Schritt des Verfahrens in die Entscheidungsmatrix für jede der fünf Phasen P1 bis P5 Entscheidungstupel ET aus einem Entscheidungswert W und dem Bezugsobjekt, hier der dazugehörigen Signalgruppe SG1 bis SG7, eingetragen werden,

FIG 4

die gemäß den Richtlinien des erfindungsgemäßen Verfahrens im zweiten Schritt neu sortierte Entscheidungsmatrix aus FIG 3,

FIG 5

ein weiteres Beispiel einer Entscheidungsmatrix, in der neben dem Block zur Verteilung der Grünzeitreserven C_res der Fahrstreifengruppen mit fünf Phasen P1 bis P5 und sieben Signalgruppen SG1 bis SG7 zwei weitere Blöcke mit den Entscheidungskriterien Zahl der Signalgruppen C_ped(n) in jeder Phase bzw. die augenblickliche Phasendauer C_double zur Entscheidungsfindung herangezogen werden,

FIG 6

ein weiteres Beispiel einer Entscheidungsmatrix, in der neben dem Block zur Verteilung der Grünzeitreserven C_res der Fahrstreifengruppen mit fünf Phasen P1 bis P5 und sieben Signalgruppen SG1 bis SG7 ein vorrangiger Block mit Randbedingungen (C_max(i,n)) zur Entscheidungsfindung herangezogen wird,

The invention will be explained in more detail below with reference to the accompanying figures with reference to embodiments. The same components are provided with identical reference numerals in the various figures. Show it:

FIG. 1

an example of a signalized node with a main signal group SG1 and an additional signal group SG2 for the respective lane groups FG1 and FG2, respectively,

FIG. 2

a circuit diagram of a traffic signal system control system 1 according to the invention with a data transmission interface 10, a traffic data processing unit 20 and a data transfer interface 30,

FIG. 3

an example of a phase-based decision matrix with a decision criterion, here a block for the distribution of green time reserves C _res of lane groups, which is evaluated in the inventive method, wherein in the first step of the method in the decision matrix for each of the five phases P1 to P5 decision tuple ET from a decision value W and the reference object, here the associated signal group SG1 to SG7, are entered,

FIG. 4

the decision matrix newly sorted according to the guidelines of the method according to the invention in the second step FIG. 3 .

FIG. 5

a further example of a decision matrix, in addition to the block for the distribution of the green time reserves C _{res of} the lane groups with five phases P1 to P5 and seven signal groups SG1 to SG7 two more blocks with the decision criteria number of signal groups C _ped (n) in each phase or the instantaneous phase duration C _{double is used} for decision-making,

FIG. 6

another example of a decision matrix, in addition to the block for the distribution of the green time reserves C _{res of} the lane groups with five phases P1 to P5 and seven signal groups SG1 to SG7 a priority block with boundary conditions (C _max (i, n)) is used for decision-making,

In FIG. 1 an example of a signaled node is shown in a control area. This signalized node is an example of a node with two lane groups FG1 (straight ahead traffic) and FG2 (right turn) from the south. The lane group FG1 for the straight-ahead traffic is signaled by means of the main signal group SG1. The right-turn lane group (lane group FG2) is also signaled by means of the main signal group SG1 and by means of an additional signal group SG2. The signaling of this lane group thus comprises two signal groups. The main signal group is usually executed as a three-field signal generator with full discs, the additional signal for right turn as doubly signal with arrows to the right.

This schematic representation of a signaled node is shown in detail only for the lane groups from the south. The lane groups from the north as well as from the east and west can be executed analogously or otherwise signaled. For example, it would be possible for two lane groups with the main signal or additional signal to be signaled from the north, while only one lane group with one signal group is signaled from the east and west. Further traffic routes with one or more lane groups and corresponding signal groups are known to the person skilled in the art.

A control area may include one or more such signalized nodes, each having different lane groups FG and signal groups SG. In a control range, this makes it possible not only to control the traffic light systems at one node, but also to tune the traffic light system control at several successively signaled nodes of a control area in such a way that that an improvement in the flow of traffic throughout the control area is achieved. It would be optimal if a so-called "green wave circuit" could be achieved on the routes with the highest traffic volume.

Optimizing the offset times is crucial for this. The offset time is the time at which the phase sequences of signal groups are offset from one another at two consecutive signaled nodes. This calculation is supported by means of the method according to the invention for controlling traffic light systems at one or more such signalized nodes in a control range. More specifically, in the coordination of two nodes, the phase length of certain phases may be targeted by a certain amount of time, e.g. a few seconds, to improve the traffic flow and its degree of coordination over both signaled nodes.

• eine Datenübermittlungsschnittstelle 10 zur Übernahme von Verkehrsdaten aus einer Verkehrsüberwachungseinrichtung 15,
• eine Verkehrsdaten-Verarbeitungseinheit 20 zur Ermittlung von Lichtsignalanlagen-Steuerungsdaten, insbesondere zur Ermittlung der minimalen Knoten-Umlaufzeit, der Regelbereich-Umlaufzeit und der Versatzberechnung für alle signalisierten Knoten des Regelbereichs, und
• eine Datenübergabeschnittstelle 30 zur Übergabe der zuvor ermittelten Lichtsignalanlagen-Steuerungsdaten an eine Lichtsignalanlagen-Steuerungseinheit 35.

For this purpose, in a traffic signal system control system 1 according to the invention, as shown in the FIG. 2 as an example, the following facilities are provided:

• a data transmission interface 10 for transferring traffic data from a traffic monitoring device 15,
• a traffic data processing unit 20 for determining traffic signal control data, in particular for determining the minimum node cycle time, the control area cycle time and the offset calculation for all signalized nodes of the control area, and
• a data transfer interface 30 for transferring the previously determined traffic signal control data to a traffic signal control unit 35.

The traffic data processing unit uses the inventive method for determining the light signal control data. In the following, an exemplary embodiment of such a method will be explained in more detail. In particular, here is an improved and very transparent method for the distribution of green times on the respective phases explained. However, as already explained above, such a phase-oriented green time distribution method can also be replaced by a blocking group-oriented method, for example.

In the method, a plurality of computation steps or arithmetic operations are carried out successively in which, for example, an interval determined by the method according to the invention is extended by a specific time interval, here for example one second, before the procedure is run through again for the next time interval. These repetitions are performed until a previously determined closed-loop time, which is the same for all the signalized nodes in the control area, is reached. Then the cycle is completed and another cycle begins in which the same computation steps or arithmetic operations are also repeated again. As a result, no phase shift occurs.

In order to be able to determine which phase is extended, a decision matrix is set up in which all phases of a signaled node are plotted in columns. In the FIGS. 3 to 6 exemplary decision matrices are shown which are used in the method according to the invention. The candidate selection process is divided into several steps:

Step 1: Fill the decision matrix with decision tuples.

How out FIG. 3 can be seen, there are two columns per phase. The left column contains the value of the decision criterion W, the right column the reference object of the decision criterion. The two values of the columns of a phase are also called decision tuples, as they remain connected in all steps of the selection process.

In the in FIG. 3 decision matrix shown, the reference object is the number of the identifying signal group SG, if the decision criterion lane group oriented (eg performance) or signal group oriented (eg minimum signal group periods), or the number of the respective phase, if the decision criterion is phase-oriented. Which decision criteria lane group-oriented and which are phase-oriented, results from the further description of the embodiment.

C = Zahl der Spalten einer Entscheidungsmatrix,
P = Zahl der Phasen.A lane group is a group of lanes that are identically signaled. Is a lane group next to a main signal still assigned an additional signal, such as in the lane group FG2 in FIG. 1 , the number of the additional signal is used for referencing. The recognition number EK references the type of the decision criterion. As a result, the number of columns of a decision matrix is calculated according to the following formula: $$\mathrm{C}\mathrm{=}\mathrm{2}\mathrm{<figure-callout\; id="1"\; label="P\; +"\; filenames="imgf0002.png,imgf0003.png"\; state="\{\{state\}\}">P\; </figure-callout>}\mathrm{+}\mathrm{}\mathrm{1}$$

C = number of columns of a decision matrix,
P = number of phases.

The decision matrix used in this embodiment is constructed of blocks of different types of decision criteria, which are sorted according to their given priority. These are entered line by line in the matrix. Initially, the block with the highest priority decision criterion is entered, followed by the blocks with the lower priority decision criteria in descending order. Each block can be divided into one or more lines. The number of lines depends on the respective decision criterion and, if necessary, is explained in more detail in the description of the individual decision criteria. In the FIGS. 3 to 6 are not all Blocks shown, because the matrices are otherwise too big. Therefore, for reasons of clarity, all blocks have been omitted, which need not be taken into account in each case specifically explained in the figure decision.

Minimum phase duration

C_min =

minimale Phasendauer, [s]

d_act =

aktuelle Grünzeit, [s]

d_min =

minimale Grünzeit, [s]

n =

augenblicklicher Zeitschritt

p =

Phase.

For example, the first block may consist of a row with the decision criterion C _min for the minimum phase durations (in the FIGS. 3 to 6 not shown). The block for minimum phase durations gets the recognition number 1 and consists of one line. The minimum phase duration is determined according to the following equation: $${\mathrm{C}}_{\min }\left(\mathrm{p}\mathrm{n}\right)\mathrm{=}{\mathrm{d}}_{\mathrm{act}}\left(\mathrm{p}\mathrm{n}\right)\mathrm{-}{\mathrm{d}}_{\min }\left(\mathrm{p}\right)$$

C _min =

minimum phase duration, [s]

_dact =

current green time, [s]

d _min =

minimal green time, [s]

n =

instantaneous time step

p =

Phase.

If the minimum criteria are not met, the result for the instantaneous minimum phase duration C _min (p, n) is a negative value, which is then entered in the matrix in the column of the respective phase. Otherwise, a positive value would result, but instead the default value 99 is entered in the matrix.

Minimum signal group duration

C_min(s,n) =

minimale Signalgruppendauer, [s]

d_act =

aktuelle Grünzeit, [s]

d_min =

minimale Grünzeit, [s]

s =

Nummer der Signalgruppe

n =

augenblicklicher Zeitschritt.

For example, the second block may consist of up to S lines for minimum signal group durations (in the FIGS. 3 to 6 not shown). S is the number of signal groups. The decision criterion of the minimum signal group duration gets the recognition number 2 and is calculated as follows. $${\mathrm{C}}_{\min }\left(\mathrm{s}\mathrm{n}\right)\mathrm{=}{\mathrm{d}}_{\mathrm{act}}\left(\mathrm{s}\mathrm{n}\right)\mathrm{-}{\mathrm{d}}_{\min }\left(\mathrm{s}\right)$$

C _min (s, n) =

minimum signal group duration, [s]

_dact =

current green time, [s]

d _min =

minimal green time, [s]

s =

Number of the signal group

n =

instantaneous time step.

If the minimum criteria are not met, the result for the current minimum signal group duration C _min (s, n) is a negative value, which is then entered in the matrix in the column of the respective phase and in the decision tuple for the respective signal group. Otherwise, a positive value would result, but instead the default value 99 is entered in the matrix.

Maximum phase duration

C_max =

maximale Phasendauer, [s]

d_act =

aktuelle Grünzeit, [s]

d_max =

maximale Grünzeit, [s]

n =

augenblicklicher Zeitschritt

p =

Phase.

The third block may consist of a line with the decision criterion C _max for the maximum phase durations (in the FIGS. 3 to 6 not shown). The detection number for the maximum phase duration is 3. The maximum phase duration C _max is calculated as follows: $${\mathrm{C}}_{\mathrm{Max}}\left(\mathrm{p}\mathrm{n}\right)\mathrm{=}{\mathrm{d}}_{\mathrm{act}}\left(\mathrm{p}\mathrm{n}\right)\mathrm{-}{\mathrm{d}}_{\mathrm{Max}}\left(\mathrm{p}\right)+100$$

C _max =

maximum phase duration, [s]

_dact =

current green time, [s]

d _max =

maximum green time, [s]

n =

instantaneous time step

p =

Phase.

If the maximum criteria are not or only just kept, the maximum phase duration C _max (p, n) results in a value greater than or equal to 100, which is then taken over into the matrix. Otherwise, a value of less than 99 would result, but instead the default value 99 is entered in the matrix.

Maximum signal group duration

C_max(s,n) =

maximale Signalgruppendauer, [s]

d_act =

aktuelle Grünzeit, [s]

d_min =

minimale Grünzeit, [s]

s =

Nummer der Signalgruppe

n =

augenblicklicher Zeitschritt.

The fourth block (see eg FIG. 6 ) again consists of up to S lines for maximum signal group durations. S is the number of signal groups. The decision criterion of the maximum signal group duration gets the recognition number 4 and is calculated as follows. $${\mathrm{C}}_{\mathrm{Max}}\left(\mathrm{s}\mathrm{n}\right)\mathrm{=}{\mathrm{d}}_{\mathrm{act}}\left(\mathrm{s}\mathrm{n}\right)\mathrm{-}{\mathrm{d}}_{\mathrm{Max}}\left(\mathrm{s}\right)+100$$

C _max (s, n) =

maximum signal group duration, [s]

_dact =

current green time, [s]

d _min =

minimal green time, [s]

s =

Number of the signal group

n =

instantaneous time step.

If the maximum criteria are not or just kept, the maximum signal group duration C _max (s, n) results in a value greater than or equal to 100, which is then entered in the matrix in the column of the respective phase and in the decision tuple for the respective signal group becomes. Otherwise, a value of less than 99 would result, but instead the default value 99 is entered in the matrix.

Green time reserve

C_res =

Grünzeitreserve

d_act =

aktuelle Grünzeit, [s]

d_nec =

notwendige Grünzeit, [s]

1 =

Fahrstreifengruppe

n =

augenblicklicher Zeitschritt.

The fifth block (see FIGS. 3 to 6 ) contains up to L lines for the green time reserves of lane groups. Here, L is the number of separately signaled lane groups 1. This decision criterion is used first to ensure efficiency. The decision criterion C _res for ensuring the performance is calculated using the following formula: $${\mathrm{C}}_{\mathrm{res}}\left(\mathrm{l}\mathrm{n}\right)\mathrm{=}\frac{{\mathrm{d}}_{\mathrm{act}}\left(\mathrm{l}\mathrm{n}\right)\mathrm{-}{\mathrm{d}}_{\mathrm{nec}}\left(\mathrm{l}\mathrm{n}\right)}{{\mathrm{d}}_{\mathrm{nec}}\left(\mathrm{l}\mathrm{n}\right)}$$

C _res =

Green time reserve

_dact =

current green time, [s]

d _nec =

necessary green time, [s]

1 =

lane group

n =

instantaneous time step.

d_nec =

notwendige Grünzeit, [s]

q(l) =

effektive Verkehrsaufkommen, [Fahrzeuge/h]

tc(n) =

augenblickliche Umlaufzeit, [s]

cap(l) =

Kapazität der Fahrstreifengruppe 1

sat(l) =

Auslastungsgrad der Fahrstreifengruppe 1, [Fahrzeuge/h].

The necessary green time d _{nec is} calculated as: $${\mathrm{d}}_{\mathrm{nec}}\left(\mathrm{l}\mathrm{n}\right)\mathrm{=}\frac{\mathrm{q}\left(\mathrm{l}\right)\mathrm{\cdot }\mathrm{tc}\left(\mathrm{n}\right)}{\mathrm{cap}\left(\mathrm{l}\right)\mathrm{\cdot }\mathrm{sat}\left(\mathrm{l}\right)}$$

d _nec =

necessary green time, [s]

q (l) =

effective traffic, [vehicles / h]

tc (n) =

current orbital period, [s]

cap (l) =

Capacity of lane group 1

sat (l) =

Degree of utilization of lane group 1, [vehicles / h].

If a lane group has more than one signal group, the green times of the signal groups are added. The obtained values for the decision criterion of the green time reserve C _res are in turn entered as decision tuple ET for each of the phases in the respective rows of the matrix. The recognition number for the decision criterion is 5 (see decision matrix in FIG. 3 ).

However, the decision criterion of the green time reserve C _{res can} also be omitted and replaced by the decision criterion for free green C _free (not shown in the figures). This is used instead of the green time reserve decision criterion if the decision values of the performance for all lane groups or the green time reserve are greater than or equal to zero. The free green of a signaled node can be defined as the difference between the control range round trip time and the minimum necessary node round trip time of the phase sequence of the respective signaled node. The detection number for the free green is 6.

C_free =

Freies Grün, [s]

d_act =

aktuelle Grünzeit, [s]

d_nec =

notwendige Grünzeit, [s]

fac (p) =

Phasenfaktor für Phase p

l =

Fahrstreifengruppe

n =

augenblicklicher Zeitschritt.

The allocation of the free green C _free is calculated according to the following equation: $${\mathrm{C}}_{\mathrm{free}}\left(\mathrm{l}\mathrm{n}\right)\mathrm{=}\frac{{\mathrm{d}}_{\mathrm{act}}\left(\mathrm{l}\mathrm{n}\right)-\mathrm{fac}\left(\mathrm{p}\right)\cdot {\mathrm{d}}_{\mathrm{nec}}\left(\mathrm{l}\mathrm{n}\right)}{\mathrm{fac}\left(\mathrm{p}\right)\cdot {\mathrm{d}}_{\mathrm{nec}}\left(\mathrm{l}\mathrm{n}\right)}$$

C _free =

Free Green, [s]

_dact =

current green time, [s]

d _nec =

necessary green time, [s]

fac (p) =

Phase factor for phase p

l =

lane group

n =

instantaneous time step.

The phase factor fac (p) can be parameterized per phase. If a lane group is released in several phases, then the phase-related green time reserve of the lane group is different in different phases. The same applies to competing or enemy lane groups. As a result, the free green can be assigned to specific phases by means of parameterization of the phase factor fac (p).

Number of signal groups

The sixth block (see eg FIG. 5 ) consists of a line for preferential distribution of the green to phases with more signal groups. As a result, if there are, for example, two phases which do not differ with respect to the signal transmitters for the lanes, so-called vehicle signals, the phase which includes the pedestrian signals is preferably extended. The recognition number for the decision criterion of the number of signal groups is 7.

C_ped =

Entscheidungskriterium für die Zahl der Phasen

S(p) =

Zahl der in Phase p freigegebenen Signalgruppen

n =

augenblicklicher Zeitschritt.

This decision criterion for the number of signal groups, also called decision criterion for the pedestrian phases, is marked C _ped and can be calculated as follows: $${\mathrm{C}}_{\mathrm{ped}}\left(\mathrm{n}\right)\mathrm{=}\mathrm{100}\mathrm{-}\mathrm{S}\left(\mathrm{p}\right)$$

C _ped =

Decision criterion for the number of phases

S (p) =

Number of signal groups released in phase p

n =

instantaneous time step.

Instantaneous phase duration

C_double =

Entscheidungskriterium für die augenblickliche Phasendauer, [s]

d(p,n) =

augenblickliche Phasendauer der Phase p, [s]

n =

augenblicklicher Zeitschritt.

The seventh block (see eg FIG. 5 ) consists of a line with the current phase durations. This decision criterion serves to evenly divide the green time into phases for multiple scores, here double scores. The decision criterion C _{double is} calculated as: $${\mathrm{C}}_{\mathrm{stand-in}}\left(\mathrm{n}\right)\mathrm{=}\mathrm{d}\left(\mathrm{p}\mathrm{n}\right)$$

C _double =

Decision criterion for the current phase duration, [s]

d (p, n) =

instantaneous phase duration of phase p, [s]

n =

instantaneous time step.

Overall, from the seven blocks of a decision matrix used according to the invention, up to R rows result, with R = 2S + L + 4.

Decision values that are not relevant or do not exist for a phase are entered with the value 99. Thus, in each column (phase) for each row corresponding decision tuples are set, which are then used in the next step to divide the green time to a phase.

step 2

In the next step of the selection process on the basis of a decision matrix, as described in FIG. 3 for the case of block 5, the entries of the columns within the blocks for each phase are sorted separately in ascending order. That is, the decision tuples in each phase are resorted, starting with the lowest value and ending with the highest value. Lines that are then only assigned the values 99 are deleted. The resorted decision matrix FIG. 3 is in FIG. 4 shown. How to see there is, the decision matrix simplifies significantly, as some of the lines have already been eliminated.

When reordering the decision tuples, it should be noted that the order of the blocks is not changed, but this re-sorting is performed separately for each block only.

step 3

The order of the blocks themselves determines the sequence and importance, i. the priority, the different types of decision criteria.

This means that primarily the minimum phase durations and signal group durations are adhered to, which are fixed boundary conditions. For example, pedestrians can be given enough time to reach the opposite side of the road. Also, the required clearance times and switching times for the signalized node are maintained here.

The next priority is the maximum phase durations and signal group durations. Consequently, it is also possible that by observing the minimum conditions, the maximum conditions of individual phases or signal groups can be violated.

Another priority is the performance and the free green. By observing the minimum and maximum conditions, individual signal groups may not reach their capacity.

If, thereafter, the phase is not yet clear, the phases with the highest number of signal groups (e.g., pedestrian phases) are preferred.

As another - in the embodiment explained here the last - priority, the phases with the lowest Phase duration are preferred. This is a decision criterion, especially for multiple applications.

In the third step, the above decision criteria are processed sequentially, ie in the order of their priorities. Since the lines in which all decision values were 99 have been deleted, it may happen that some blocks are omitted before the decision-making and are no longer to be considered in this step. Therefore, it may also be that all boundary conditions such as the minimum conditions and maximum conditions of the first four blocks are not used for the decision. This is for example in the in FIG. 4 The decision matrix shown in the case. Here is the first, relevant block the green time reserve.

Decision-making is then done line by line, searching for the smallest value of the line. If there is only one phase in the first line, which alone has been assigned the smallest decision value, then the decision-making process is completed after the first line.

On the other hand, if there are two phases assigned the smallest value, then the decision making in the next row proceeds on the basis of these two phases. Analogously, this is done in three or more equivalent phases. If in the next line one of these phases is the one with the lowest value, the decision-making process ends in this phase.

If no decision has yet been found, go to the next line until only one phase has been found that has been assigned the lowest decision value on any of the following lines. If all the lines of a decision criterion have been processed without a decision having been made on this selection procedure, In the next block, the decision criterion of lower priority is continued according to the same scheme.

If there are still several candidates after the last line, the second receives the phase that comes first within the phase sequence. Thus, the method results in a clear candidate selection over a decision path with decreasing priority of the decision criteria. The special feature of this procedure is the clear transparency, which is given by the high system of the candidate selection procedure. So you can even read on the basis of the logged matrixes even at a later time, which phase has received on the basis of which decision criterion the respective green time interval.

1. 1. Initialisiere die Signalgruppendauern etc. entsprechend der Länge der Phasenübergänge und der minimalen Phasendauern entsprechend der Versorgung.
2. 2. Iterativ je zu verteilender Zeiteinheit, in dem Beispiel je Sekunde:
   • 2.1 Berechne die notwendige Grünzeit und Reserve je Fahrstreifengruppe.
   • 2.2 Ermittle die Entscheidungsmatrix.
   • 2.3 Gib die Sekunde der Phase, die sie entsprechend der Entscheidungsmatrix am meisten benötigt.
   • 2.4 Erhöhe die Grünzeit der in der Phase enthaltenen Signalgruppen um eins.

The candidate selection method based on a decision matrix explained in more detail above can also be defined by the following algorithm for the green time distribution:

1. 1. Initialize the signal group periods etc. according to the length of the phase transitions and the minimum phase durations corresponding to the supply.
2. 2. iterative per time unit to be distributed, in the example per second:
   • 2.1 Calculate the required green time and reserve per lane group.
   • 2.2 Determine the decision matrix.
   • 2.3 Give the second of the phase it most needed according to the decision matrix.
   • 2.4 Increase the green time of the signal groups contained in the phase by one.

• 1. Initialisierung: Alle Phasen der Phasenfolge sind mögliche Kandidaten zur Vergabe der Sekunde.
• 2. Bis nur noch ein Kandidat übrig ist oder alle Zeilen abgearbeitet sind:
  • 2.1 Nimm die nächste Zeile der Entscheidungsmatrix.
  • 2.2 Ermittle den kleinsten Entscheidungswert der Zeile.
  • 2.3 Ermittle aus der bisherigen Kandidatenmenge die Phasen, die den kleinsten Entscheidungswert besitzen; diese Phasen sind die neue Kandidatenmenge.

The decision making by means of the decision matrix (step 2.3 of the above algorithm) works like this:

• 1. Initialization: All phases of the phase sequence are possible candidates for awarding the second.
• 2. Until only one candidate is left or all lines have been processed:
  • 2.1 Take the next line of the decision matrix.
  • 2.2 Determine the smallest decision value of the line.
  • 2.3 Determine from the previous candidate set the phases that have the lowest decision value; these phases are the new candidate set.

A second embodiment is in FIG. 5 shown.

In the decision matrix in FIG. 5 adjacent to the block of the distribution of the green time reserves C _res of the lane groups of two blocks with the decision criteria "number of signal groups C _ped (s) in each phase" or "continuous instantaneous phase C _double" are given to the decision-making. Since after the resorting of the block C _res the phases P1 and P2 have not yet left due to the identical smallest value in the first row and in the third row, a selection of the suitable candidate must be found in a block for a decision criterion of low priority. In this case, this is the case only in the case of the instantaneous phase duration.

Accordingly, in this embodiment, finally, the phase P1 is allocated the second. It should be noted that the phases P3 to P5 have already been excluded in the first row of the decision matrix and all other rows of this matrix had no influence on the further selection procedure. This also applies if a value is smaller than in phase P1 or P2.

Another embodiment is in FIG. 6 shown.

FIG. 6 1 shows a decision matrix in which, in addition to the block for distributing the green time reserves C _{res of} the lane group groups, a priority block with the boundary condition of the maximum signal group duration C _max (i, n) is used for decision-making. In this example, due to the fact that the signal group SG5 has already reached the maximum value of 5 seconds, the phase 2 from the further decision making. The calculation of the decision value W for the signal group SG5 was carried out as 5 - 5 + 100 = 100. After no further signal group has reached the maximum value, the value for all was set as 99.

From line 2 it follows that the phase P1 receives here the additional green time of 1 second, because this with respect to the remaining phase P1, P3, P4 and P5 the smallest value in the line 2 of the decision matrix, the block for the green time reserve C _res having. Thus, this time the decision ends after the second line, because a candidate was found.

It is finally pointed out once again that the method described above in detail as well as in the illustrated traffic signal control system are merely exemplary embodiments which can be modified by the person skilled in many different ways without departing from the scope of the invention. In particular, the decision matrix can be replaced by a corresponding decision tree that enables reliable candidate selection based on the same systematic decision criteria as defined for the matrix and a corresponding priority list. Such a tree can be particularly advantageous if the decision-making can no longer be linearly described. Such a case could occur if a control unit controls a plurality of sub-nodes, which indeed run in a signal program and are thus switched simultaneously, for example, but which allow parallel phase sequences within the sub-nodes. The optimization of these parallel phase sequences could be optimized in parallel branches of the decision tree. Alternatively, such a case could also be mapped by means of a plurality of decision matrices, which, however, each have to be processed independently of one another in order to ensure that the solutions for the subnodes are consistent with one another are and can be implemented in a signal program. Furthermore, the use of the indefinite article "on" or "one" does not exclude that the characteristics in question may also be present multiple times. In addition, "units" can consist of one or more, even spatially distributed components.

Claims (15)

A method of controlling traffic signal systems for road traffic at one or more signalized nodes in a control area, wherein a distribution of green times on signal groups (SG) of the individual signalized nodes by means of a candidate selection method according to a number of predetermined decision criteria. A method according to claim 1, wherein the distribution of green times in predetermined periods, preferably in seconds, takes place. Method according to claim 1 or 2, wherein, when using several decision criteria, these priorities are assigned and during the distribution of the green times the decision criteria are processed according to their assigned priorities. Method according to one of the preceding claims, wherein the green times are distributed in a phase-oriented or blocking group-oriented manner and one or more decision tuples (ET), preferably 2-tuples, are determined for each decision criterion for each phase (P) or blocking group, with the aid of which the method the distribution of green times on the individual phases or blocking groups by selecting the favored phase or blocking group makes. Method according to claim 4, wherein, for each decision criterion, the decision tuple (ET) for each phase (P) or blocking group is input into a decision matrix which is executed in the candidate selection method in rows or columns. A method according to claim 4 or 5, wherein the decision tuples (ET) represent branches in a decision tree, which is then processed stepwise or branchwise. The method of claim 1, wherein one or more of the predetermined decision criteria is / are the minimum phase duration, minimum signal group duration, maximum phase duration, maximum signal group duration, lane group performance, free green distribution, number of enabled signal groups, or instantaneous phase duration. A method according to claim 7, wherein the performance of the lane groups is a decision criterion and this is weighted differently for each phase for distributing the green time. A method according to claim 7 or 8, wherein the performance of the lane groups is a decision criterion and this is determined by means of a load factor or a performance index of the lane groups (FG) associated signal groups (SG). Method according to one of the preceding claims, wherein the method first performs a green time distribution for determining a node cycle time of the respective signalized nodes in a control range and these node cycle times are used to select a control range cycle time of the entire control range. Method according to claim 10, wherein a trend of a minimum node round trip time and / or closed loop round trip time selection is taken into account when determining a node round trip time and / or closed loop round trip time. Method according to Claim 10 or 11, wherein a timer and / or hysteresis element is taken into account when determining a node cycle time and / or control cycle cycle time. Method according to one of Claims 10 to 12, wherein the signaled nodes of a control range which are to be considered for the selection of the control range round trip time are chosen to be planning. Traffic signal control system (1) for traffic control at one or more signalized nodes in a control area, comprising at least: a data transmission interface (10) for transferring traffic data from a traffic monitoring device (15), a traffic data processing unit (20) for detecting traffic signal control data, a data transfer interface (30) for the traffic signal control data to a traffic signal control unit (35),
wherein the traffic data processing unit (20) is arranged to distribute green times to signal groups (SG) of the individual signalized nodes using a candidate selection method according to a number of predetermined decision criteria. Computer program product, which is directly loadable into a processor of a computer device, with program code means to carry out all the steps of a method according to one of claims 1 to 13.

Images