JOURNAL OF COMPUTER SCIENCE AND ENGINEERING, VOLUME 10, ISSUE 2, DECEMBER 2011 6

A control strategy based on HPN and GRAFCET for real time traffic signal control Mohamed Kamel JBIRA Abstract—Traffic congestion becomes a serious problem in many big cities around the world. The most common interest to overcome this problem is to have a good management and control of signal traffic lights. For this intention, a control strategy which allows a dynamic and automatic traffic light control in real time is suggested. This strategy is based on using a class of hybrid Petri nets for the analyzing level and GRAFCET for the control level. The translation of analysis model to control model is insured through an appropriate conversion technique. Moreover, the suggested strategy quantifies the gain of using programmable logic controller and wireless sensors for a real time implementation. Index Terms— Control strategy, GRAFCET, Hybrid Petri nets, Traffic signal control.

——————————  ——————————

1 INTRODUCTION

W

ITH the rapid increase in the number of vehicles and the numbers of roads with limited capacity, traffic congestion becomes predominant problem causing transportation delay. The most common studies interested in managing urban traffic areas and road networks are the traffic signal control. The traffic signal control has been recognized as an important means of solving traffic congestion problems. A variety of models, methods, and strategies have been developed and applied for controlling urban traffic via signalized intersections. The currently available methods and strategies utilizing traffic-light controls may be classified into two categories [1],[2],[3]: Fixed-time strategies: These strategies consider a given time of a day and determine the optimal splits (green times) and optimal cycle time, based on the historical constant demands over the considered signalized urban area [4]. Traffic-response strategies: These strategies are that to construct a real-time control system with an optimum signal setting [5], [6]. Such strategies are based on realtime measurements using in most cases either inductive loops or pattern-recognition digital cameras to determine appropriate signal settings in real time. The most used strategies for urban traffic control are fixed-time strategies [7], nevertheless, nowadays, more interest is given to developing control strategies in which the control system is fully responsive. Many trafficresponse strategies are being developed and the wellknown and widely ones are split cycle offset optimization technique (SCOOT) [8], optimization policies for adaptive control (OPAC) [9], [10], and traffic-responsive urban control (TUC) [11]. These strategies are based on dynamic programming optimization procedure and allow continual incremental adjustments in real-time of cycle lengths, splits, and offsets. Most of these works are inter-

ested on solving vehicle delays at signalized intersections under a wide range of conditions. Furthermore, many others control strategies are developed for different and specific cases using new tools: state charts, genetic algorithms, expert systems, etc…, [12],[13],[14],[15]. In [29] “to be published” author proposed the development of a generic hybrid model describing both physical traffic flows and control of signalized intersections. This scheme is based on splitting green times proportionally to the predicted queue sizes in input links for each new cycle time. This work permitted suggesting in this paper a control strategy for dynamic and automatic traffic light control going through different phases from analysis level up to proposing a logic scheme for real time implementation. Based on detected flow of information and the real time measure of physical queue sizes of vehicles, this strategy allows a real time correction of green times phases. This paper is organized as follows. Section II describes the architecture of the suggested control strategy. In section III conversion procedures of Petri net model into GRAFCET model are recalled. In section IV a study case is discussed to prove the applicability of the proposed control strategy. Finally, we conclude in section V by some comments on the proposed modeling framework and a brief description of the on-going extensions of this work.

2 CONTROL STRATEGY In this section, the structure of the proposed control strategy is presented. The Fig. 1 shows a block diagram, which describes the architecture of this strategy. Two main parts are considered: Description and analysis part. Control system part

————————————————

M. K. JBIRA is with the College of Computer Science and Engineering, Taibah University, KSA. © 2012 JCSE www.journalcse.co.uk

7

Fig.1. Architecture of the control strategy

2.1 Description part: System analysis Urban traffic networks are considered to be a Discrete Event System (DES). Among the modeling methodologies applied for studying behavior of a DES, Petri Nets (PNs) provide a simple and clear means for analyzing and synthesis. Moreover, PNs are able to capture the precedence relations and interactions among the concurrent and asynchronous events typical of DES [16], [17]. More specifically, many works considered urban traffic networks not only as Discrete Event Systems but as hybrid systems and are interested on using Hybrid Petri Nets (HPNs) for modeling and analysis stage [18],[19],[20],[21],[22]. Some of those works used HPN to state and solve the problem of coordinating several traffic lights with the aim of improving the performance of some classes of special vehicles (public and emergency vehicles) [19], [21]. Some others mixed HPN with other tools to optimize the control policies for a traffic light systems [22]. In the presented work, a signalized intersection with its input and output flows is considered to be a hybrid system, to include both continuous-time and discreteevent components. A class of HPNs is then proposed for the analyze purpose. Hence, the vehicle flow behavior is represented by means of Continuous Petri Net (CPN) providing the “Flow Model”, and Timed Petri Net (TPN) represents the traffic light dynamics providing the “Traffic light model”. The reader can find more detailed presentation of HPNs and examples in [23]. 2.2 Control system part In the following, the main components of the control sys-

tem part are described. A. Logic controller The logic controller will be based on one local Programming Logic Controller (PLC) and has the job to regulate the traffic signals for all direction. The choose of PLCs to execute this type of real time control tasks is due to their flexibility, reliability, ease programming and also their acceptable cost. Furthermore, the logic controller has the job to switch traffic signals to a special timing plan for vehicles asking for privilege (VIP or emergency vehicles). The managing of a priority request can be allowed through the implemented model as follows: The implemented sensor on the direction from which a privileged vehicle asks for a phase modification send an interrupt signal to the logic controller; If the light for the direction the privileged vehicle comes from is expected to be green, then the green time is extended until the privileged vehicle crosses the intersection; If the red light is expected, the logic controller changes signals of all directions to amber and then red lights except the direction the privileged vehicle comes to green light. In this case privileged vehicles will only wait upstream vehicles to move making the road free. B. Programming tool: GRAFCET In general, to implement a control strategy, the control logic has to be translated into computer code. On the other hand, since their first use, PLCs are programmed using the Ladder Logic Diagram (LLD). Nevertheless, this programming tool presented limitation for complex application and is unable to capture the dynamics of the controlled system. Nowadays, the standard for PLC proBuy SmartDraw!- purchased copies print this document without a watermark . Visit www.smartdraw.com or call 1-800-768-3729.

8

gramming languages classified GRAFCET (English translation: Sequential Function Chart (SFC)) as the most important and more suitable programming tool for PLCs and useful in eliminating the disadvantages of LLD. GRAFCET, started as a French standard developed inside AFCET (Association Francaise de Cybernétique Economique et Technique), and was afterwards integrated in the IEC 1131-3 standard as a Programmable Controllers' language [24]. The GRAFCET allows a PLC program to be organized into a set of steps and transitions connected by directed links. To each step is associated one or many actions and with each transition a transition condition. For more detail on GRAFCET it’s recommend to consult [24]. The GRAFCET programming language evolved from a restricted variant of safe Petri nets [25], suitable for implementation in PLCs. Therefore, a Petri net for such description and analysis level can easily be translated into a GRAFCET for a real time implementation [26]. A recall of main translation procedures is presented in section III. C. Wireless sensors: Real time assessment In the past, the first problem faced when implementing a new real-time control system for a urban traffic system was how to insure vehicle detection dynamically. Available means at that time was only able to detect the vehicle in a fixed position. After that, looped detectors appeared and are considered for a long time as the best source of real-time traffic data, nevertheless, sometimes they contain many missing values or incorrect measures. Nowadays, Wireless Sensor Network (WSN) is one of the advanced technologies used in several real time implementations for different applications. For example, the work developed in [27] illustrates the powerful of using wireless sensors for real time of a transmission power control system. For real time traffic control, many vehicle detection methods using this technology are also successfully developed. The work presented in [28] justifies their powerful use. Moreover, wireless sensors are not expensive and they are battery powered sensing devices combining sensing, computation and communication capabilities. By using these faster devices in the proposed control strategy, we guarantee an easily measurement of parameters related to the vehicles motion in real-time with higher accuracy. Installed in the beginning and ends of all input links, wireless sensors send data directly to an interface connected to the logic controller for processing. D. Supervisor The supervisor has the job of managing the system working either in its regular conditions or during a special event. Furthermore, the supervisor switches the control of traffic signals to its classic timing plan if any part of the system did not work appropriately as if a sensor did not provide its measure to the logic controller. A warning signal is then launched by the control system until required repairing operations are conducted. E. Green times calculator Green light durations are determined for each next cycle. Therefore equation adopted for this job is evaluated with real values measured during current cycle C(t) and optimal splits are then determined. For this work, the pre-

diction model developed by the author in [29] “to be published” is considered. This model is based on the prediction of vehicles’ queue sizes at different input links Qi(t+1) and the calculation of optimal green durations is estimated proportionally to the real time queues sizes in these input links. This model is given by the following equation:

Q i (t 1)

G i (t 1)

n k

Q k (t 1) 1

.C (t )

This model should be implemented in the PLC as subroutines. The different delays associated to both transitions of TPN (traffic light model) and GRAFCET of control are then equal to determined durations: Gi(t+1).

3 CONVERSION OF PETRI NET MODEL INTO GRAFCET MODEL As far it is comfortable with the results obtained from analysis level through the Petri net model, a translation onto a GRAFCET model is required for a future real time implementation. This allows a PLC program to be organized into a set of steps and transitions connected by directed external input and output links. Petri nets, has strong similarities with GRAFCET, although the different graphical notations and execution semantics adopted. With GRAFCET we use two types of nodes: steps and transitions, which have a similar semantics to Petri net nodes: places and transitions. For the GRAFCET, a step has only two states which can be active or not active; this means, in Petri net terms, that the corresponding place can be marked with one token or unmarked. In Table I, we summarize the basic rules of the translation procedures from Petri nets to GRAFCET. For a complete analysis of translation, it is recommended to consult [26]. TABLE I BASIC TRANSLATION PROCEDURES FROM PETRI NETS TO GRAFCET Petri Net t1

GRAFCET (1)

p1

1

t2 t1

(2) Step to place t2

p1 Or-junction

(1)

1

(2)

9

1

p1

t1

t2 Or-disjunction

p1

(2)

(1) 2

1

p2

crossing of four vehicles in parallel. Indeed, such an assumption is not restrictive and it is supported by experimental observations, showing that lane changing occurrences are rare for this intersection. However, three traffic flows are allowed for each direction: left, through and right turns. On each corner of the intersection there is a traffic signal allowing three subsequent phases: green, amber and red.

t1

(1) 3

p3 Join 1

p1

(1)

t1

Fig. 2. The real traffic intersection.

p3

p2

2

3

Fork p1

p2

1

t1

p3

2

(1)

p4

3

4

Synchronization

4 STUDY CASE To prove applicability of the proposed control strategy, a study case is discussed in this section.

4.1 Traffic Intersection Description A real case of a signalized intersection which includes four directions, with two links (one input link and one output link) for each is considered. All links are formed of 4 lanes without any inflicted direction which allow the

Where: ILi/OLi: Input/Output Links for direction i (West, South, East or North) CA: Crossing Area.

4.2 HPN model The HPN model of the described intersection is shown in Fig. 3. This model is composed of two sub models: Physical model (Flow model): A continuous PN, modelling the flows of vehicles entering the intersection, the crossing area, and the flows of vehicles leaving the intersection. Control Model (Traffic light model): A discrete PN modelling the control part of traffic lights. 4.3 GRAFCET of the Control Level As explained previously, the GRAFCET is adopted as the programming tool for the logic controller. In this section a translation of the TPN model representing the control level into a GRAFCET model, which can be easily implemented for real time control, is completed. Starting from the designed HPN model and by applying rules summarized in section III, the "equivalent" GRAFCET model is presented in Fig. 4. The marking of the continuous model represents the number of vehicles. Whereas the marking of the discrete model represents the different states of control signals of the traffic lights.

10

rection i i : correspond to one direction: West, South, East, or North.

Supervising GRAFCET

GRAFCET of Control: Gc

Fig. 3. HPN of the signalized intersection. Fig. 4. GRAFCET of the control part

Places meaning: pi : Input link i is enabled. pGi : Waiting during green time for input link i. pAi : Waiting during amber time for input link i. pIi : Vehicles are in the queue at input link i pOi : Vehicles are leaving the intersection toward direction i pCA : Vehicles are crossing the intersection area. Transition meaning: ti : Enabling the input link i. tGAi : Green time ends and amber time starts for input link i. tIi1 : Vehicles entering in the queue at input link i tIi2 : Vehicles leaving the queue at input link i tOi1 : Vehicles leaving the cross area toward direction i tOi2 : Vehicles leaving the intersection toward di-

Condition associated to transitions: Pi : A vehicle asking for privilege from direction i. EP: End of privilege. [timer = RG i ]: Real green duration to be executed for input link i. [timer = RA ]: Real amber duration to be executed for input link i. Actions associated to steps: Gc{j}: Forcing the GRAFCET of control Gc to active only state of step j. GSi: Set green signal for the input link i. ASi: Set amber signal for the input link i. RGi: Real green duration associated to transition of the controller (timed Petri net and GRAFCET of control).

11

5 CONCLUSION A control strategy has been developed in this paper for dynamic and automatic traffic lights control in real time. The suggested strategy has been designed through different phase from analysis level up the proposing a logic scheme for real time implementation. The control of traffic lights may be insured using a local PLC and input measurements is then guarantied using wireless sensors, which assist a rapid signal settings in real time. The analysis of the system has been conducted with a hybrid Petri net and then translated to a control program using GRAFCET. Finally, a study case is discussed to prove applicability of the proposed control strategy. Work is in progress on considering the dynamicity of the cycle time according to global traffic flow dynamic to peak periods on the day. This, will allow decreasing of delayed times of vehicles in waiting around signalized intersections within a urban traffic network.

6 REFERENCES [1]

Diakaki, C., Papageorgiou, M., and Aboudolas, K.,"A multivariable regulator approach to traffic-responsive network-wide signal control", Control Engineering Practice, 10(2), 183–195, 2002. [2] Papageorgiou, M., Diakaki, C., Dinopoulou, V., Kotsialos, A., and Wang, Y., "Review of road traffic control strategies", Proceedings of the IEEE, 91(12), pp. 2043-2067, 2003. [3] Patel, M., and Ranganathan, N., "IDUTC: An intelligent decision making system for urban traffic-control applications", IEEE Transactions on Vehicular Technology, 50(3), pp. 816-829, 2001. [4] Robertson, D., “TRANSYT method for area traffic control”, Traffic Engineering Control, vol. 10, pp. 276-281, 1969. [5] Hunt, P. B., Robertson, D. L., and Bretherton, R. D., “The SCOOT on-line traffic signal optimization technique”, Traffic Engineering Control, vol. 23, pp. 190-192, 1982. [6] Farges, J. L., Henry, J. J., and Tufal, J., “The PRODYN real-time traffic algorithm”, Proceeding of 4th IFAC Symposium Transportation Systems, pp. 307-312, 1983. [7] Huang, Y. S., and Su, P. J., "Modelling and analysis of traffic light control systems", IET Control Theory and Applications, vol. 3, Issue 3, pp. 340–350, 2009. [8] Robertson, R., and Bretherton, R. D., “Optimization networks of traffic signals in real time – the SCOOT method”, IEEE Transactions on Vehicular Technology, vol.40, pp. 11-15, 1991. [9] Gartner, N. H., Tarnoff, P. J., and Andrews, C. M., “Evaluation of optimized policies for adaptive control strategy”, Transportation Research Record 1324, Washington, DC, 1991. [10] Gartner, N. H., "OPAC: A Demand-responsive strategy for traffic signal control ", U. S. Department Transportation, Transp. Res. Rec. 906, Washington, DC, 1983. [11] Dinopoulou, V.,Diakaki, C., and Papageorgiou, M. , "Simulation investigations of the traffic-responsive urban control strategy TUC", Proceedings of IEEE Intelligent Transportation Systems, pp. 458-463, Dearborn, MI, October, 2000. [12] Poundyal, N., Lee, H. C., Kwon, Y. J., Lee, B. S. ,”Delay-bound Admission Control for Real-time Traffic in Fourth Generation IMT-Advanced Networks based on 802.16m”, Advances in Electrical and Computer Engineering, Vol. 11, Number 1, pp. 31-38, 2011.

[13] Oh, B., Na, Y., Yang, J., Park, S., Nang, J., and Kim, J., “Genetic Algorithm-based Dynamic Vehicle Route Search using Car-toCar Communication”, Advances in Electrical and Computer Engineering, Vol.10, Number 4, pp. 81-86, 2010. [14] Lee, H., Nakao, A. and Kim, J., “Understanding of network operator-friendly P2P traffic control techniques in intra domain”, International Conference on Information Networking, pp. 97-102,2011. [15] Yaping, W. and Zheng, Z., “A Method of Reinforcement Learning Based Automatic Traffic Signal Control “,Third International Conference on Measuring Technology and Mechatronics Automation, pp.119-122, 2011. [16] Murata, T., “Petri Nets: Properties, Analysis and Applications”, Proceeding of the IEEE, vol.77, no.4, pp.541-580, 1989. [17] List, G. F., and Cetin, M., “Modelling Traffic Signal Control Using Petri Nets”, IEEE transactions on Intelligent Transportation Systems, vol.5, no.3, Sept. 2004. [18] Lee, J., and Song, S., “Modeling urban transportation systems with hybrid systems: An overview”, IEEE proceedings of Intelligent Transportation Systems, vol. 2, pp. 1269–1272, Oct. 2003. [19] Di Febbraro, A., and Sacco, N., “On modeling urban transportation networks via hybrid Petri nets”, Control Engineering Practice, vol. 12, no. 10, pp. 1225-1239, 2004. [20] Di Febbraro, A., Giglio, D., and Sacco, N., "Urban Traffic Control Structure Based on Hybrid Petri Nets", IEEE proceedings of Intelligent Transportation Systems, vol. 5, no. 4, Dec. 2004. [21] Tolba, C., Lefebvre D., Thomas, P., and El Moudni, A., "Continuous and timed Petri nets for the macroscopic and microscopic traffic flow modeling", Simulation Modelling Practice and Theory 13, pp.407-436, 2005. [22] Kim, Y.-W, Kato, T., Okuma, S., Narikiyo, T., "Traffic Network Control Based on Hybrid Dynamical System Modeling and Mixed Integer Nonlinear Programming With Convexity Analysis", IEEE Transactions on Systems, Man and Cybernetics – Part A: Systems and Humans, vol.38, no. 2, 2008. [23] Alla, H., and David, R., "On hybrid Petri nets", Journal of Discrete Event Dynamic Systems, vol. 8, no. 1–2, pp. 9–40, 2001. [24] ISO/IEC International standard IEC 1131-3, Programmable Logic Controllers, Part 3, ISO/IEC, 1993. [25] Lewis, R. W., “Programming Industrial Control Systems UsingIEC 1131-3, volume 50 of IEE Control Engineers Series “, The Institution of Electrical Engineers, London, UK, 199. [26] R. Devid and H. Alla, “Petri Nets & GRAFCET: Tolls for modeling discrete event systems”, Prentice Hall, ISBN 0-1 3-327537-X, 1992. [27] Ramakrishnan, M., Ranjan, V. P. ,”Adaptive Power Control with Overhearing Avoidance for Wireless Sensor Networks”, Advances in Electrical and Computer Engineering, vol.10, Number 3, pp. 84-91, 2010.. [28] Rai, A., and Patel, G.S., “Multiple Traffic Control Using Wireless Sensor and Density Measuring Camera”, Sensors & Transducers Journal, Vol. 94, Issue 7, pp. 126-132, 2008. [29] M. K. Jbira M. Ahmed, “Computer Simulation: A Hybrid Model for Traffic Signal Optimization Considering Real Time Estimation of Vehicles Queue Sizes”, Journal of Information Processing Systems, In Press, Corrected Proof, Available on March 2011.

12

Mohamed Kamel Jbira is an Assistant Professor of Computer Engineering and Control. He earned his PhD in Computer-Aided Manufacturing from ESSTT, Tunisia in 1997; his MSc in Automatic from ENSET, Tunisia in 1991 and BSc in Electrical Engineering from ENSET, Tunisia in 1988. Before joining Taibah University, Saudi Arabia as staff member at College of Computer Science and Engineering, he worked as Engineer at MTK Factory, Tunisia, Assistant Professor at: College of Engineering, Monastir University, Dhofar University, Sultonate of Oman and King Khaled Univesity, Saud Arabia. His research interests include Modelling and Simulation Applications, functional analysis of complex systems, urban traffic networks and Computer-Aided Manufacturing. He is an Associate or Life Member of the following Associations: Institute of Electrical and Electronics Engineers (IEEE), Saudi Computer Society (SCS) and Indian Society for Technical Education (ISTE).