Computers and Software

ISSN 1828-6003 Vol. 9 N. 4 April 2014

International Review on

Computers and Software (IRECOS)

T

Contents:

IN

A Modeling Tool for Dynamically Reconfigurable Systems by Sonia Dimassi, Abdessalem Ben Abdelali, Amine Mrabet, Mohamed Nidhal Krifa, Abdellatif Mtibaa

600 609

MSG SEVIRI Image Segmentation Using a Method Based on Spectral, Temporal and Textural Features by Mounir Sehad, Soltane Ameur, Jean Michel Brucker, Mourad Lazri

618

An Effective Selection of DCT and DWT Coefficients for an Adaptive Medical Image Compression Technique Using Multiple Kernel FCM by Ellappan V., R. Samson Ravindran

628

Optimal Object Detection and Tracking Using Improved Particle Swarm Optimization (IPSO) by P. Mukilan, A. Wahi

638

Prediction Algorithms for Mining Biological Databases by Lekha A., C. V. Srikrishna, Viji Vinod

650

A Grid-Based Algorithm for Mining Spatio-Temporal Sequential Patterns by Gurram Sunitha, A. Rama Mohan Reddy

659

Online Modules Placement Algorithm on Partially Reconfigurable Device for Area Optimization by Mehdi Jemai, Bouraoui Ouni, Abdellatif Mtibaa

667

Classification of Brain Tumor Using Neural Network by Bilal M. Zahran

673

An Improved Semi Supervised Nonnegative Matrix Factorization Based Tumor Clustering with Efficient Infomax ICA Based Gene Selection by S. Praba, A. K. Santra

679

Mobility Aware Load Balanced Scheduling Algorithm for Mobile Grid Environment by S. Vimala, T. Sasikala

686

A Novel Localization Approach for QoS Guaranteed Mobility-Based Communication in Wireless Sensor Networks by P. Jesu Jayarin, J. Visumathi, S. Madhurikkha

694

R

EP

R

Efficient Hardware Implementations for Tripling Oriented Elliptic Curve Crypto-System by Mohammad Alkhatib, Adel Al Salem

(continued on inside back cover)

Copyright © 2014 Praise Worthy Prize S.r.l. - All rights reserved

International Review on Computers and Software (IRECOS) Editor-in-Chief: Prof. Marios Angelides Brunel University School of Engineering and Design Electronic and Computer Engineering Department Uxbridge - UB8 3PH U.K.

Editorial Board: (Japan)

Pascal Lorenz

(France)

Francoise Balmas

(France)

Marlin H. Mickle

(U.S.A.)

Vijay Bhatkar

(India)

Ali Movaghar

(Iran)

Arndt Bode

(Germany)

Dimitris Nikolos

Rajkumar Buyya

(Australia)

Mohamed Ould-Khaoua

(U.K.)

Wojciech Cellary

(Poland)

Witold Pedrycz

(Canada)

Bernard Courtois

(France)

Dana Petcu

(Romania)

Andre Ponce de Carvalho

(Brazil)

Erich Schikuta

(Austria)

David Dagan Feng

(Australia)

Arun K. Somani

(U.S.A.)

Peng Gong

(U.S.A.)

Miroslav Švéda

(Czech)

Defa Hu

(China)

Daniel Thalmann

(Switzerland)

Michael N. Huhns

(U.S.A.)

Ismail Khalil

(Austria)

Catalina M. Lladó

(Spain)

R

IN

T

Mikio Aoyama

(Greece)

(Spain)

Brijesh Verma

(Australia)

Lipo Wang

(Singapore)

EP

Luis Javier García Villalba

The International Review on Computers and Software (IRECOS) is a publication of the Praise Worthy Prize S.r.l.. The Review is published monthly, appearing on the last day of every month. Published and Printed in Italy by Praise Worthy Prize S.r.l., Naples, April 30, 2014. Copyright © 2014 Praise Worthy Prize S.r.l. - All rights reserved.

R

This journal and the individual contributions contained in it are protected under copyright by Praise Worthy Prize S.r.l. and the following terms and conditions apply to their use: Single photocopies of single articles may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale and all forms of document delivery. Permission may be sought directly from Praise Worthy Prize S.r.l. at the e-mail address: [email protected] Permission of the Publisher is required to store or use electronically any material contained in this journal, including any article or part of an article. Except as outlined above, no part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. E-mail address permission request: [email protected] Responsibility for the contents rests upon the authors and not upon the Praise Worthy Prize S.r.l.. Statement and opinions expressed in the articles and communications are those of the individual contributors and not the statements and opinions of Praise Worthy Prize S.r.l.. Praise Worthy Prize S.r.l. assumes no responsibility or liability for any damage or injury to persons or property arising out of the use of any materials, instructions, methods or ideas contained herein. Praise Worthy Prize S.r.l. expressly disclaims any implied warranties of merchantability or fitness for a particular purpose. If expert assistance is required, the service of a competent professional person should be sought.

International Review on Computers and Software (I.RE.CO.S.), Vol. 9, N. 5 ISSN 1828-6003 May 2014

Using Interval Constrained Petri Nets and Timed Automata for Diagnosis of Dynamic Systems Dhouibi H.1, Belgacem L.1, Mhamdi L.1, Simeu-Abazi Z.2 Abstract – The purpose of the following article is a new approach to modeling, diagnosing and controlling of discrete-event systems. This approach is using a model which combines Interval Constrained Petri Nets (ICPN) and Timed Automata to describe the diagnosed system. The Petri net is used for modelling the system which needs controlling and the timed automata is being used for the controller. This article is a description of a case study, which is a cigarette production system where the tobacco density must be held in an interval. Copyright © 2014 Praise Worthy Prize S.r.l. - All rights reserved.

T

Keywords: Fault Diagnosis, Timed Automata, Petri Net, Discrete Event Systems

R Q N ai, bi, qi, Va V V

Val0

k

R

IS m  p

R

UCL, LCL

Density of tobacco Compactness of tobacco Level in tobacco reservoir Level of the supplying conveyor Levels Hrmin , Hrmax Levels Hcmin , Hcmax Level controls Centreline (average) of all the samples plotted) Upper and lower statistical control limits Set of real numbers Set of radial variables Set of positive integer Rational values Non-empty set of formulas A multiset Initial formulas associated to tokens A token The intervals associated to places A place marking

EP

d C Hr Hc n1, n2 n3, n4 C1, C2 CL

ICPN is a sub-class of High Level Petri Nets with Abstract Marking (AM-HLPN) [5]-[7]. ICPN model allows to model and guarantee a constraint on any parameter of dynamic systems. In our case it has been used to model the continuous part (liquid flow rate) of level regulation system. The aim is to guarantee the level in the tank between minimum and maximum values. What is more, timed automata is a tool for modelling and verification of real time systems [8], [9]. A timed automata is essentially a finite automata (FSM) extended with real-valued variables. Such automata may be considered as an abstract model of timed systems. This expressive modelling tool offers possibilities of model analysis like verification, controller synthesis and also faults detection and isolation to model dynamic systems whose activity times are included between minimum and maximum values. We use it for modelling the discrete parts of system command. Both tools are applied to a robustness control for regulation systems and for description of dynamical system. First of all, this article presents the process. The following section describes ICPN and gives some basic notions on the timed automata used in the modeling step. The ICPN presents a complement to the P-temporal Petri nets [10]. Therefore, the robust control laws of this model are proven to use production data information of manufacturing production systems. We use the timed automata model to describe the control law. When the global model [11] of the process is completely defined, we present an application to level regulation system in order to illustrate our approach. At the end, a conclusion is presented with some perspectives.

IN

Nomenclature

I.

Introduction

Fault diagnosis in dynamic systems is an issue that has received a lot of attention in the past few decades. Typically in such systems, behavioral deep models are state machine [1] or Petri nets [2]-[4]. This article focuses on fault diagnosis approach based on a model combining two tools: the Petri Net and timed automata. The article considers fault diagnosis of discrete-event systems and tries to combine Interval Constrained Petri Nets (ICPN) with timed automata in order to describe and diagnose the system.

II.

Manufacturing Process

The principle problem in the cigarette transformation and production systems is disregard for the weight of

Manuscript received and revised April 2014, accepted May 2014


735

Dhouibi H., Belgacem L., Mhamdi L., Simeu-Abazi Z.

C  [Cmin, Cmax]; in g/mm3 Hr  [n1, n2]; in mm Hc  [n3, n4]

III. The Modeling Approach

IN

T

The aim is to build a diagnostic system for a process of Fig. 2. We can build it by establishing faulty transitions and must include necessary information about the fault behavior dynamics (time aspect). The diagnoser is based on a global model which combines two tools: the ICPN and timed automata (TA) in order to evaluate the variations of the tobacco quality and Timed Automata to manage the flow type disturbance. For this reason, when the model is built, it is possible to describe the constraints on the quality parameters which are required for manufacturing of products in accordance with the specifications. This model allows setting the system functioning around a target state. Another policy is to control the workshop while following the evolution of the parameters in the course of time in order to compensate for the fluctuations.

EP

R

manufactured units (see Fig. 1). From a quality point of view, a too heavy cigarette has a difficult drawing while a light one gives consumers an impression of bad quality, because its ends are not well filled and can easily fray. From a cost point of view the excess tobacco in a cigarette is considered as a loss and can cause stopping due to stuffing in the circuit of the tube formation. For a given cigarette, the weight depends on the density of tobacco which again depends on three parameters: the compactness (C) of the tobacco, the levels (Hr) and (Hc), respectively, in tobacco reservoir and of the supplying conveyor. The evolution of these parameters is proper to the typology of the considered cigarette fabric. The compactness (C) of the tobacco is a parameter of the system entry which has to be considered. Its value depends essentially on the raw material. Its variation is random and its distribution can be modeled by a normal law (according to the production statistical data). The level of tobacco varies depending on the compactness and has to be in a given interval to guarantee the good functioning of the system. This level is a variable that depends on the tobacco flow; a parameter which must be supervised in order to maintain an optimal density. This flow is controlled in two steps: at the moment when the distributor is being filled by control C1 and at the moment when tobacco is being carried on the conveyor by control C2.

Fig. 2. Dynamic global model

R

III.1. Petri Nets for Regulation Control Petri nets, a mathematical modeling tool, makes graphical modeling, simulation and real time control modeling, more functional [12]. Several classes Petri networks have been developed, each trying to describe a "view" of production systems, their design and conduct [38]-[42]. Among those classes, we include models that integrate the dimension of time [13]-[17]. The extension covers modeling systems whose behavior depends on an explicit values time. The resulting model can treat problems related to the analysis and evaluation of performance through analytical methods. Consequently, Petri nets have been used to model various kinds of dynamic event-driven systems like computer networks [17] communication systems [18]; [19], manufacturing plants; [20] command and control systems [20], real-time computing systems

Fig. 1. Manufacturing system

The control C1 maintains the level of tobacco in the girdle (tobacco reservoir). The level Hr must be comprised between a minimum Hrmin under which the machine stops, locking tobacco, and a maximum level Hrmax over which there is a need for supply. The control C2 regulates the level of tobacco (Hc) on the conveyor. It must be comprised between a minimum level Hcmin under which the tobacco driving cylinder stops, locking tobacco, and a maximum level Hcmax over which the tobacco driving cylinder stops. Therefore our goal is to obtain a homogeneous density of the tobacco. This density depends on the compactness of the tobacco (C), the level (Hr) and the level (Hc), such as:


International Review on Computers and Software, Vol. 9, N. 5

736


[22], [23] logistic networks [24] and workflows [25]-[27] to mention only a few important examples. In this work we choose the Petri nets (Intervals Constrained Petri Nets) and the timed automata as modeling tools. In fact, these tools are known as being powerful tools in modeling of Dynamic Systems. Interval constrained Petri Nets (ICPN) are introduced to extend the application field of P-Time PN by proceeding to a functional abstraction of the parameter associated places. Furthermore, introducing a new formalism is an opportunity to review the initial definition. In this way, we present in an unequivocal manner the marking as a multi-set. equally, the transmission of a quantity conveyed by a token is represented explicitly.



Val0 defines to initial formulas associated to tokens.

A token in the place pi is taken into account in transition validations when it has reached a value comprised between ai and bi. When the value is greater than bi the mark is said to be “dead”. Logically, in the firing of an upstream transition, tokens are generated in output places and their associated variables are equal to:

Val  k   qi  k 

The signification of q and Val(k) are intentionally not defined in order to provide a general model. The following relation serves as an example with P-time PN:

Definition An ICPN is a t-uple R,m,IS ,D,Val,Val0 , X , X 0

Let V be a multiset defined on V . m : P  V

where t represents time. In ICPN the application X is not mathematically imposed. However there will be applications where, for example, q parameters represent weight variations of cigarettes. In this case, parameter values associated with pairs (place, token) are independent.  State definition State E is defined by a t-uple m,D,Val, X where:

R

p  P  m  p  , where m  p  is a place marking. We note M the application: M : P  N (set of positive

p  Card  m  p  



IS : P  R   ,   R  Q    , 

m , D , Val and X are the above defined applications D and m assign a variable qi  k  to each

token k in a place pi . A token k of the place pi can take part in the validation of output transitions if:



EP

defines the intervals associated to places. R is the set of real numbers:

qi  k    ai ,bi 

pi  ISi   ai ,bi  with 0  ai  bi 

where [ai, bi] is the static interval associated with the place pi . This token k “dies” when:

D is an application that associates to each pair (place, token) a rational variable q  q  bi  . This

qi  k   bi

R

variable corresponds to a modification of the associated value of a token in a place.

X is an application which provides a value for each variable of V . Actually, X defines the real value of each q . When X is not defined, there is a possibility to make the model evolve. Furthermore, some mathematical properties may be outlined. It is called mathematical abstraction.

D : m  p   P  Va

i, 1  i  n, n  card  P  . Let k  m  pi  .

k

k  qi | ai  qi  bi , where

be a token,

ai , bi

are

rational values fixed by IS :  X is an application that assigns to each variable a value. X : Va  Q ; va  u  Q ; X sets the qi . X 0 defines the initial values of variables.  

 Computing the next step There are two different ways of reaching a state from a given one. The first solution is to use the evolution of associated variables. The other one is the transition firings. The following two definitions correspond to these two evolution possibilities.

Val associates to each token a formula of values in Q . Val is an application of set of the tokens m  p  in V : m  p   V

(2)

IN

Let V be a non-empty set of formulas to use a variables of Va .



dq 1 dt

T

q,

where:  R is a marked PN;  m is an application associating token to places: Let Va be a set of rational variables.

integer) ,

(1)

 k  m  p    v  V , where

k is

a given token. Copyright © 2014 Praise Worthy Prize S.r.l. - All rights reserved


737


TABLE I COMPARISON OF PARAMETERS P-TIME PN/ICPN MODEL P-time PN ICPN model Parameter signification C Time cycle Node identifier Variation of weight per cycle Q Variation of time for a piece (compared to cycle reference) ∆Q Variation of Variation of the added weight effective time in place compared to a reference Lower bound indicates the ai Lower bound minimum weight added indicates the otherwise the quality of minimum time product is deteriorated needed to execute the operation Upper bound indicates the bi The upper bound maximum weight added fixes the maximum otherwise the quality of time to not exceed product is deteriorated m Product, resource, Product, resource, constraint constraint

Definition 1: State E'  m',D',Val', X '  is accessible from another state

E  m,D,Val, X 

according to

associated variable evolution if and only if: 1. m'=m 2.  j a token in pi: q’i(j)= qi(j) +qi(j) ai q’i(j)  bi where [ai, bi] is the static interval of the place pi. The possibility of reaching q’i(j) depends generally on the coupling with other q evolutions. This particular aspect is not presented here. Definition 2: State E'  m',D',Val', X '  is accessible state from another state E  m,D,Val, X  by the firing

T

transition ti if and only if: 1- ti is validated from E, 2- p  P, m'  p   m  p   Pr e  p,ti   Post  p,ti  .

An invariant is associated to each state. It corresponds to the conditions needed to remain in the state. The number of clocks depends on the parallelism in the system. The automata can stay in one state as long as the invariant condition is checked. Each transition of an automata is conditioned by an event or temporization called “guard” and its execution determines the discrete evolution of the variables according to its associated assignment. Let us consider the timed automata given in Fig. 3. This automata has two clocks x and y. The continuous evolution of time in this model is represented by x  1 and the labelled arcs in the graph represent the model of discrete evolution. The guard in each arc is a transition labelling function that assigns firing conditions with the transitions of the automata. The affectation is a function that associates with each transition of the automata one relation that allows actualizing the value of continuous state space variables after the firing of a transition. The invariant in the state S0 and S1 are respectively y ≤ 5 and x ≤ 8. The initial state of this system is represented by an input arc in the origin state (S0). In the dynamic model, active clocks are found in each state. A graphical interpretation of the timed automata is the automata graph (see Fig. 3).

arcs from p to ti, P ost  p,ti  corresponds to the weight of

Val  k '   Val  k   q  k 

R

the input arcs from ti to p. 3- Tokens that remain in the same place keep the same associated value between E and E'. The newly created tokens take zero values for the q counter associated to their new places. The value allocated to the token k’ by Val is:

IN

Pr e  p,ti  corresponds to the weight of the output

(3)

R

EP

where k is a token that is in an input place pj of ti and consumed to transition ti. The previous firing rule allows computing states and accessibility-relationships. The set of the firing sequences from an initial state specifies the PN behavior as well as sets of accessible markings or validated firing sequences in the case of Autonomous PN. Mathematically, P-time PN and ICPN have the same properties. However, the physical interpretation that must be given to the model is completely different. Table I summarizes signification of different parameters that take part in both P-time PN and the ICPN. III.2. The Timed Automata for Diagnosis

The timed automata tool [27], [28] is defined as a finite state machine with a set of continuous variables that are named clock. These variables evolve continuously in each location of the automata, according to an associated evolution function. As long as the system is in one state Li, the clock xi is continuously incremented. Its evolution is described by X  1 . The clocks are synchronized and change with the same step. Fig. 3. Example of Timed Automata



738


In our case verification (analysis) means searching accessible trace of timed automata (reverse path) [29]. This reverse path projects the evolution of the system, from a final faulty state to the initial state. The reverse path is also called diagnostic path. We suppose the initial state is known. Our task can be seen as retrace the automaton graph from the faulty states to the known origin state. The aim is to find from the set of reverse path the coherent ones. The principle of the analysis is shown in automaton graph with fault model (Fig. 4). From fault model one can see that fault F1 can occur from state 2, and the fault F2 from the state 3. The diagnostic model must be defined that if fault occurs in the system, fault must be located according to the time instant. If the fault occurs in the time 4tu, it is fault located as F1. In another case, the fault occurs in the time 7tu, thus the fault F2 is located.

above relation may be approximated by the following one doing a first order linearization [31]. This development gives the relations (4) which describes the behaviour of the process around a reference state:

d  1C   2 H r   3 H c

(4)

IN

T

with i (i: 1...3) are constant values where: 1=HrHc, 2=C Hc, 3=Hr, and C, Hr, Hc, d are target values. Fig. 5 describes model ICPN of the process presented by Fig. 1. In order to construct this global model with validity intervals, simulation of some real production data was performed. The target values are: d = 0.22, C = 5, Hr = 200 and Hc = 70.

Fig. 5. Validity intervals of the ICPN model

IV.

Modelling Process

IV.1. Modelling with ICPN

In this model places Pc, PHr, and PHc denote respectively the compactness, the tobacco level in reservoir, the tobacco level on conveyor belt and trimmer tobacco. Places P1, P2 and P3 represent, respectively, the constraints associated to Pc, PHr, and PHc. These places are considered as “test places” to maintain the specification of the above parameters, which obviously describe the ICPN tobacco flow.

R

Fig. 4. Principle of the backward time analysis

EP

It is possible to construct a RdP model of a shop in order to study the different laws of order while simulating the statistical distribution of the orders on the density [29]. However, it is necessary to assume that the synthesis of the model is subject to mistakes and approximations. Best case scenario, we can recover on the model the outputs of the shop for various values of the parameters. However, the data corresponding to the real outputs of the shops is available. Then it is possible to calculate the tolerances of the parameters directly on statistics of the shop. These last results will be logically more correct than those that integrate the imperfections ensuing the modelling phase. Note that these parameters (d, C, Hr and Hc) are related. Obviously, the variation of one of these parameters provides a variation of the density. When it is outside the validity range, the production has to be rejected or the machine blocks. Our objective is to make sure that the permitted tolerance concerning tobacco density will be respected by controlling Hr, Hc and E parameters. It must belong to a predefined interval. The aim of the controller is to maintain the density specification by changing the setting levels Hr and Hc, whereas they have to remain in a validity interval. We consider the variations of a parameter are always very small comparing to its setting value. Consequently, the

R

1. Verification of the robustness of the ICPN model Definition: the robustness of a system is defined by its capacity to maintain the specified characteristics facing variations (expected or unexpected). The robustness of a system can be divided in two classes:  A passive robustness when the objectives are kept without modification of the control.  An active robustness when the objectives are kept through a computation (in real-time) of a new control. This study deals with the passive robustness of the process with regard to the perturbations at its entry because of compactness variations of the tobacco. The operating objective is to maintain the specifications on the parameters: C and levels. For this, the Petri-Net model has been simulated using the intervals of Fig. 1, in same. Under these conditions we can conclude that our model is robust shown in Fig. 5 which represents the variation of tobacco density. 2. Control maintaining constant density The effective value of parameters can be calculated with polynomial algorithms [32].



739


conveyor belt. After the duration t2 the level n4 is reached and the machine goes to the nominal speed. As the level in the reservoir and on the conveyor are respectively greater than n1 and n3 the system is functioning normally. The control sequence is then the following: S0: is the initial position when the machine is initialized. S1: the shaft-off flap open the tobacco flows into reservoir. S2: After the time t1  t1min ,t1max  the cylinder begins

This can be done because the above algorithm is only based on the structural properties of P-time Petri Net. In this case, it has been shown that, under some particular assumptions, the property may be extended to ICPN.

Density

0

machine reaches its first speed. S4: After an additional time t3  t3 min ,t3 max  the

IN

T

machine reaches its nominal speed. Such as the values of times t1, t2 and t3 depend of compactness value. Fig. 7 represents the nominal model

100

200

300

Fig. 7. Nominal model

We suppose the variations of compactness are very small around a reference value. In this case, our control is cyclic. Then the diagnosis task will be solved in one cycle only. The next step in our work is the model checker. Its purpose to verify if all faulty states in the dynamic model are reachable or if it is necessary to add some other sensors to isolate the faults. The proposed approach deals with time analysis of timed automata. The principle of modeling task is to follow the control sequence. Then let us start building the state space model of the system. We use the levels in reservoir and on the conveyor (three states for each: LOW, MEDIUM and HIGH). This leads to nine states for the timed automation, which correspond to all the possibilities for the levels of tobacco:  State L1L2 (n1n3) denotes LOW level in both parts.  State M1L2 (n10, n3) denotes MEDIUM level in reservoir and LOW level on the conveyor.  State H1L2 (n2, n3) denotes HIGH level in reservoir and LOW level on the conveyor.  State L1M2 (n1, n20) denotes LOW level in reservoir and MEDIUM level on the conveyor.  State M1M2 (n10, n20) denotes MEDIUM level in both parts.  State H1M2 (n2, n20) denotes HIGH level in reservoir and MEDIUM level on the conveyor.  State L1H2 (n1, n4) denotes LOW level in reservoir and HIGH level on the conveyor.  State M1H2 (n10, n3) denotes MEDIUM level in

EP

LSD 23

LID 20

to rotate and the tobacco flows on the conveyor belt. S3: After an additional time t2  t2 min ,t2 max  the

R

3. Computing the robust control When the ICPN model of the process is completely defined, it is possible to analyse the structural properties. It has been proven that most of the structural properties of P-time PN can be extended to ICPN. Finally, using the production data information, a computing methodology has been applied in order to build the validity intervals of the ICPN model. This approach uses only a sub-part of the information, because we only want to find critical tests which are needed for designing experiments applied in the production data. An observation of the tobacco processing by different units during one month has resulted in picking out the variations of the output measures: compactness, trimmer and the tobacco density. Fig. 6 represents the variation of density. It has plotted control limits that present regulation boundaries, which are managing boundaries, and a centreline gained by calculating average arithmetic value of the measurement samples. In our case the measured values are within control limits and thus the process is under control. A centreline (CL) represents the mathematical average of all the samples plotted. UCL and LCL present respectively the upper and lower statistical control limits that define the constraints of common cause variations.

400

Time in sec

Fig. 6. Variation of density

R

We have modelled using the ICPN tool which presents a functional abstraction of the P-temporal Petri Nets, constraints subjected on flow and quality parameters while integrating the margins of robustness. The goal is to satisfy qualitative and quantitative needs of the market. IV.2. Modelling with Timed Automata

In this work we model the control part of the system (Fig. 1) using timed automata. Then we use the approach developed by [33] and the approach of [34]. These approaches are based on principle of state and events by integrating the time. We consider the production sequence as following: the system starts from a known initial state. Firstly, the shaft-off flap opens; the tobacco flows into reservoir. After the duration t1 the level n2 is reached, then the drive motor rotates at low speed and the cylinder begins to rotate and the tobacco flows on the



740


way of locating a fault, and to determining the time of its occurrence. Fig. 10 describes the variations of the tobacco on the conveyor and in reservoir. Fault detection: The results of real-time faultdetection (Fig. 10) show two fault scenarios: in the first one a fault occurred at t = 30 which corresponds to state H1L2. In the second one a fault occurred at t= 55 which corresponds to state L1H2. Fault Isolation: For the step of fault isolation, the time is considered when a fault occurred. The backward time analysis searches the possible reverse path to locate the fault according to the time of fault occurrence. In our case it is clear that the first default corresponds to Fc “Cylinder stuck stopped “ (Fig. 7) and the second defaults corresponds to FS “Schaft-off flap stuck open” .

IN

T

reservoir and HIGH level on the conveyor.  State H1H2 (n2n4) denotes HIGH level in both parts. In this work we are interested only in detecting the overflow of the reservoir (given by HR) and we consider two faults related to the reservoir:  FS (Schaft-off flap stuck open),  FC (Cylinder stuck stopped). When an overflow is detected, according to the time at which it happens we can determine which fault FS or FC occurred. Note that in this faulty model the place representing optimal states is M1M2.  This behavior is represented by the faulty model (Fig. 8) will be combined with ICPN model which guaranties a constraint on any parameter of dynamic systems and represents the continuous part (liquid flow rate).

Fig. 8. Model of the faulty system

VI.

Conclusion

This article deals with the diagnostic approach based on modelling of control laws and its flow. This model combines the timed automata and The Petri Nets. It should contain all considered evolution of the system. The proposed modelling with Petri Nets methodology is therefore validated by a large set of data, and it provides an interesting industrial efficiency for the considered case study. Diagnosis method is based on time analysis which uses the timed automata. Our approach is validated through an industrial application and for this validation we have chosen the simplest case. However, our approach can be applied to more complicated cases as well. For example, we interest in the event of fault diagnosis in the presence of common causes.

Result Application

EP

V.

R

Fig. 10. Variation of levels with fault

R

The designs of experiments are used to exploit the manual controlled data production in order to compute the valid behavior ranges [35]. Also the variation of levels Hr an Hc without faulty are shown in Fig. 9.

Acknowledgements The authors would like to thank the comments provided by the anonymous reviewers and editor, which help the authors improve this paper significantly. We would also acknowledge the help from Zineb SimeuAbazi, Institut National Polytechnique de Grenoble, INPG France.

Fig. 9. Variation of levels without fault

The results of the control are calculated by using a model based on studying real production data. In the next lines, the characteristics of this constructed model are used to calculate a control by using the faulty model. Our objective is to detect and identify the faults occurring in the process. That leads to determining the

References [1]


Reger J. and Schmidt, K. A finite field framework for modelling, analysis and control of finite state Automata. Mathematical and


741


[7]

[8] [9]

[10]

[11]

[12]

[13]

[14] [15]

[16]

[17] [18]

[19]

[20]

[21]

[22]

[23]

[24]

[28] [29]

[30]

[31]

[32] [33]

T

[6]

[27]

IN

[5]

[26]

nets and their application to schedulability analysis of real-time system specifications. IEEE Transactions on Software Engineering 21(1): 32-49. Landeghem, Rik and Carmen-Veronica Bobeanu. 2002. Formal modeling of supply chain: an incremental approach using Petri nets. 14th European Simulations Symposium and Exhibition Dresden, Germany. Mtibaa, S., Tagina, M., A combined Petri nets and model-driven engineering for requirements specification approach for ServiceBased Applications analysis, (2012) International Review on Computers and Software (IRECOS), 7 (5), pp. 2131-2138. Lin, Chuang, Liqin Tian and Yaya Wei. 2002. Performance equivalent analysis of workflow systems, Journal of Software 13(8): 1472-1480. H. Motameni, A.Movagar, M.Fadavi Amiri, IJE Transactions B: Applications, Vol. 20, No.1, April 2007. Trpakis S. Fault diagnosis for timed automata. In: Proc. 7th Int. Symp. Formal Technics in Real-time and Fault tolerant Systems (FTRTFT) 02), Oldenburg, Germany 2002. Wong K. C. and Wonham, W. M. On the Computation of observers in discrete-Event Systems. Discrete Event Systems 14 (1), January 2004, pp. 55-107. H. Dhouibi, M. Bochran, Z. Simeu Abasi, M. Hassani. Diagnosis approach using bond graph and timed automata. IJAREEIE, vol. 2, Issue 9, September 2013. Henri Klajnmic "Intervalles de confiances et intervalles de tolérance". EDF recherche et développement., 2002 Dhouibi H., S. Collart Dutilleul, E. Craye and L. Nabli , "Computing Intervals Constrainted Petri Nets: a tobacco manufacturing application", IMACS conference, Paris, 2005, pp. 440-446. Nabli L., Dhouibi H., Using Interval Constrained Petri Nets for Regulation of Quality: The Case of Weight in Tobacco Factory, International Journal of Intelligent Control and Systems, IJICS, VOL. 13, NO. 3, pp. 178-188, September 2008. Lunze J. and Schroder J. “Sensor and actuator fault diagnosis of systems with discrete input and output” Automatica, 2000. Zineb Simeu- Abazi et al. Fault diagnosis for discrete event systems: modeling and verification. Reliability Engineering et System Safety. Volume 95, Issue 4, April 2010, PP 369-378 Dhouibi H., Collart Dutilleul S., Nabli L., Craye E. Using Interval Constrained Petri Nets for reactive control design: a tobacco manufacturing application; The international journal for manufacturing science et production, Volume 9, NOs, 3-4, pp. 217-229, 2008. Zhao, Y., Bi, G., Modeling of complex rolling multi-agent system, (2012) International Review on Computers and Software (IRECOS), 7 (7), pp. 3532-3537. Fliss, I., Tagina, M., Combining coverability and invariants to diagnose multiple faults in discrete-event systems, (2012) International Review on Modelling and Simulations (IREMOS), 5 (4), pp. 1569-1579. Mazouzi, M., Kallel, O., Hasnaoui, S., Abid, M., Modelling and evaluating a crossbar Switched Fabric CAN Network using stochastic and colored Petri nets, (2012) International Review on Computers and Software (IRECOS), 7 (2), pp. 568-578. Jalilvand, A., Noroozian, R., Ashouri, A., Modeling and evaluating of protection schemes based on petri nets, (2011) International Review on Modelling and Simulations (IREMOS), 4 (1), pp. 180-187. Kuo, J.-L., Chao, K.-L., Grid-connected multilevel inverter with intelligent Petri nets controller for ocean current power generation system, (2010) International Review of Electrical Engineering (IREE), 5 (3), pp. 858-869.

[34]

[35]

R

[4]

[25]

[36]

[37]

EP

[3]

Computer Modelling of Dynamical Systems 10(3-4):253-285. M. Combacau, M. Courvoisier. A hierarchical and modular structure for FMS control and monitoring. First international Conference on AI, Simulation and Planning in high autonomy systems, Tucson, Arizona, March 1990. S. Elkhattabi, D. Corbeel, J.C. Gentina “Integration of dependability in FMS, 7th IFAC Symposium on Information Control Problems in Manufacturing Technologiy, Toronto, May 1992, pp. 249-255 Sahraoui, M. Couvoisier, R. Valette. Some considerations on monitoring in distributed real time control of flexible manufacturing systems. International Conference on Industrial Electronics Control and instrumentation. IECON 86, Milwaukee, sept. 29 - oct. 3, 1986, p. 805-810 Dhouibi H., Collart Dutilleul S., Nabli L., Craye E., “Using interval constrained Petri Nets for regulation of quality” LSS conference 2010 ,France. P. Yim, A. Lefort, and Hebrard "System Modelling with Hypernets" ETFA’96 IEEE Conferences, pp 37-47, Paris, 1996. Collart Dutilleul S., H. Dhouibi and E.Craye. , "Internal Robustness of Discret Event System with interval constraints in repetitive functioning mode", ACS'2003 conference, Miedzyzdroje Poland, pp. 353-361. Alur R. Dill D. “A theory of timed automata”, Theoretical Computer Science (TCS), 126 (2), 1994, pp. 183-235 Bengtsson J. and Wang Yi, “Timed automata: semantics, algorithms and tools” In lecture Notes on Concurrency and Petri Nets, W. Reisig and G. Rozenberg (eds), LNCS. W. Khansa, J.P. Denat, S. Collart-Dutilleul, "P-Time Petri Nets for Manufacturing Systems". Wodes.96, Edinburgh UK, August 19-21, pp. 94-102, International Workshop on Discret Event Systems, 1996. Mhamdi L., Dhouibi H., Simeu-Abasi Z., Liouane N. Using constrained Petri nets and timed automata for modelling of discrete event systems. IJE, vol.27, pp 113-122, January 2014 Asthana, R., Ahuja, N.J., Darbari, M., Model proving of urban traffic control using Neuro Petri Nets and Fuzzy Logic, (2011) International Review on Computers and Software (IRECOS), 6 (6), pp. 983-987. J. Sifakis. « Use of Petri Nets for Performance Evaluation, Measuring Modeling and evaluating Computer Systems”, E Bleiner and E. Gelenbe editors, Hollande, 1977, pp. 75-93. P. Merlin, A Study of the Recoverability of Computer system». Thèse de Doctorat, Université de Californie, 1974. Khansa W. ; Aygalinc P.; Denat J.-P., Structural analysis of ptime Petri nets, CESA'96 IMACS Multiconference : computational engineering in systems applications, Lille , France. Zhao, Y., Bi, G., Modeling for complex adaptive agent system based on time colored Petri nets, (2012) International Review on Computers and Software (IRECOS), 7 (5), pp. 2495-2501. M. Diaz, Vérification et mise en oeuvre des réseaux de Petri ». Paris, Hermès Sciences Publication, 2003. Ajmone Marsan, M., M. G. Balbo and G. Conte. 1986. Performance Models of Multiprocessor Systems, Massachusetts: The MIT Press. Merlin, P., and D. Farber. 1976. Recoverability of communication protocols - implication of a theoretical study. IEEE Transactions on Communications 1036-1043. Wang, J. 2006. Charging information collection modeling and analysis of GPRS networks. IEEE Transactions on Systems, Man and Cybernetics, Part C 36(6). Venkatesh K., M. C. Zhou, and R. Caudill. Comparing ladder logic diagrams and Petri nets for sequence controller design through a discrete manufacturing system. IEEE Trans. on Industrial Electronics 41(6): 611-619. 1994 Andreadakis, S.K., and A.H. Levis. 1988. Synthesis of distributed command and control for the outer air battle. Proceedings of the 1988 Symposium on C² Research. SAIC, McLean, VA. Mandrioli, D., A. Morzenti, M. Pezze, P. Pietro S. and S. Silva. 1996. A Petri net and logic approach to the specification and verification of real time systems. In: Formal Methods for Real Time Computing (C. Heitmeyer and D. Mandrioli eds), John Wiley & Sons Ltd. Tsai, J., S. Yang, and Y. Chang. 1995. Timing constraint Petri

[38]

[39]

[40]

R

[2]

[41]

[42]

Authors’ information


1

LARATSI- ENIM –University Monastir, Tunisia.

2

G-SCOP -University Joseph Fourier , France.


742


Hedi Dhouibi received his Engineer degree of maintenance and DEA at National School of Engineering - University of Center, Tunisia in 1997 and 1999 respectively. In 2005, he obtained his doctorate degree in Industrial automation: automatic and Industrial computing from University of the sciences and the technologies of Lille France. He is currently Assistant professor of Electrical Engineering at University of Kairawan Tunisia. His research interests include Modeling, Intelligente Control and Monitoring and command Manufactory systems.

IN

R

EP

R

Dr. Zineb Simeu-Abazi is Assistant Professor at the Polytech’Grenoble in the University Joseph Fourier where she teaches control processing, Automation and Industrial Engineering, dependability and Industrial Maintenance. She holds a Ph.D. in Computer Science and Automation at the Institut National Polytechnique de Grenoble, INPG France on 1987 and an « Habilitation à Diriger des Recherches” » HDR in 1998, from Grenoble University, France. She is particularly interested in the on line maintenance, diagnostic, recycling and performance evaluation fields. In relation to these topics she took scientific responsibility of French and International projects/groups on e-maintenance such as the CNRS MACOD working group (Modelling and Optimisation of Distributed vs. Collaborative Maintenance). She is a president of the scientific council of diag21 association.

T

Lobna Belgacem was born in Teboulba, Tunisia, in 1987. She graduated from the National School of Engineering of Monastir, University of Center. She obtained her Engineer degree of electrical and Master in 2012 and 2013, respectively. Actually, she is preparing her doctorate degree in automation. She is former member in LARATSI - (Labortoire d’Automatique, Traitement de Signal et Imagerie), University Monastir, Tunisia.



743

International Review on Computers and Software (IRECOS) (continued from outside front cover) 701

Trust Based Web Services with Secure Control Policies and Quality of Services Using PSO by E. S. Shamila, V. Ramachandran

710

PROCEOL: Probabilistic Relational of Concept Extraction in Ontology Learning by K. Karthikeyan, V. Karthikeyani

716

Artificial Fish Swarm Load Balancing and Job Migration Task with Overloading Detection in Cloud Computing Environments by A. Mercy Gnana Rani, A. Marimuthu, A. Kavitha

727

R

EP

R

IN

T

Energy Efficient Data Collection Framework for WSN with Layers and Uneven Clusters by L. Malathi, R. K. Gnanamurthy

Abstracting and Indexing Information: Cambridge Scientific Abstracts (CSA/CIG) Academic Search Complete (EBSCO Information Services) Elsevier Bibliographic Database - SCOPUS Index Copernicus (Journal Master List): Impact Factor 6.14 Autorizzazione del Tribunale di Napoli n. 59 del 30/06/2006

T IN R EP R 1828-6003(201404)9:4;1-J Copyright © 2014 Praise Worthy Prize S.r.l. - All rights reserved