Patrice Bonhomme

Towards the control of time-critical systems

Petri nets are a powerful formalism for the specification and analysis of concurrent systems, such as sequential systems, manufacturing systems etc. Adapted to time-critical systems (systems whose time issues become essential), different time Petri nets have been developed in the literature: Merlins model, p-time Petri-net, Time ER nets etc. The presented approach allows us to determine transitions firing instants in order to ensure functional timing correctness of the modeled system. It uses a p-time Petri-net as the modeling tool. It is based on an enumerative procedure and on the evaluation of transitions firing conditions successively at the first, the second and the q/sup th/ firing instant. Moreover, it provides a performance evaluation and control component.

State observer synthesis of real-time systems modeled by P-time Petri nets

This paper deals with a novel state observer synthesis method of real-time systems under partial observation. Indeed, the developed approach aims at estimating the marking of a P-time Petri net system based on the observation of the occurrence of particular transitions called observable transitions. As opposed to many existing approaches the application of the method does not require strong assumptions on the structure of the model considered. In addition, although the time factor is taken into consideration, the proposed technique is not hampered by the state space explosion problem as it relies on the underlying untimed structure of the P-time model studied - the building of the state class graph is not necessary.

Control and performances evaluation of time dependent systems using an enumerative approach

Petri nets are a powerful formalism for the specification and verification of concurrent systems, such as sequential systems and manufacturing systems. To deal with systems whose time issues become essential, different extensions of Petri nets with time have been proposed in the literature, each one being dependent on the application considered. In this paper, a new enumerative analysis technique for P-time Petri nets is proposed. It consists of simple operations on time intervals and does not require complex computations contrary to the other methods, dealing with time critical systems, which can be found in the literature. Moreover a control approach is also derived

Using firing instants approach and partial order to control time critical systems

Time-critical systems are characterized by operation times included between a minimum and a maximum value. A transgression of these specifications acts again the quality of service. In this paper, a dynamic robust control is proposed to cancel as quickly as possible the effects of a delay or an advance on a planned task. P-time Petri Net is used as modeling tool and the control is based on an extension of the firing instant approach permitting negative instants to update the tasks scheduling. The illustration of the method is made on a graphite ceramic cell.

Proactive/reactive approach for maintenance tasks in time critical systems

The maintenance of systems involved in production processes must be carried out with maximum efficiency and minimum downtime, especially on time-critical systems (operation times are included between a minimum and a maximum value). This paper presents a proactive/reactive control to fit preventive maintenance into a busy production schedule. As preventive maintenance is regarded as a non-preemptive task, p-time Petri net is used as modelling tool to generate temporal and logical constraints to be considered. To update the tasks scheduling, flexibility on time and on task order is obtained by permitting negative instants in the used firing instant approach. The illustration of the method is made on a graphite ceramic cell

A symbolic schedulability technique of real-time systems modeled by P-Time Petri nets

This paper addresses the schedulability analysis of real-time systems modeled by P-Time Petri nets. The proposed approach is based on the consideration of the reachability graph of the (untimed) underlying Petri net of the studied model. The schedulability analysis is then conducted in two steps. Once a feasible firing sequence (called occurrence sequence) is highlighted, thanks to a firing instant based approach the timing constraints are added, yielding an inequality system representing its timing behavior. Thus, deciding whether a firing sequence is schedulable or not takes the form of the solution of a linear programming problem. Furthermore, the modification of a timing constraint does not alter the form of the obtained system, helping adjusting accurately timing properties in a specifications refinement purpose.

Towards a new exhaustive simulation technique for P-time Petri nets

Petri nets are a powerful formalism for the specification and verification of concurrent systems, such as sequential systems and manufacturing systems where many processes can compete for limited and constrained ressources. To deal with systems whose time issues become essential, different extensions of Petri nets with time have been proposed in the literature, each one being dependent on the application considered. In this paper, a new analysis technique for P-time Petri nets is proposed. It consists of simple operations on time intervals to determine a superset of behaviors of the modelled system. Then, once a desired behavior is extracted, thanks to a linear programming problem, its feasibility is verified via the existence of a solution.

Verification technique for time Petri nets

Petri nets are a powerful formalism for the specification and verification of concurrent systems, such as sequential systems, manufacturing systems. To deal with systems whose time issues become fundamental, different time Petri nets extensions have been developed in the literature, each one being dependent on the application considered. A verification technique for time Petri nets (Merlins model) is proposed. It is based on the determination of an inequalities system generated by a possible evolution (in terms of a feasible firing sequence) of the model considered. This system can be used to check reachability problems as well as evaluating the performances of the model considered and determining the associated control for a definite functioning mode.

A TOKEN PLAYER ALGORITHM FOR THE CONTROL OF TIME CRITICAL SYSTEMS

Petri nets are a powerful formalism for the specification and verification of concurrent systems, such as sequential systems and manufacturing systems. To deal with systems whose time issues become essential, different extensions of Petri nets with time have been proposed in the literature, each one being dependent on the application considered. In this paper, a new control technique for time Petri nets is proposed. It is based on the firing instant notion and it consists in determining an inequalities system generated for a possible evolution (in terms of a feasible firing sequence for the untimed underlying Petri net) of the model considered. This system can be used to check reachability problems as well as evaluating the performances of the model considered and determining the associated control for a definite functioning mode. Furthermore, the proposed approach is not restricted to subclasses or safe Petri nets.

A control model for multi-products processing

Time-critical systems are characterized by operation times included between a minimum and a maximum value. An acceptable behaviour depends not only on the logical correctness of the task sequence but also on the time at which the operations are performed. So, a specific control is required. In this paper, the robustness property is used to build a time schedule of when the manufacturing system is to process several kinds of parts. This approach uses a p-time Petri net as its modeling tool and considers the multi-processing case as a disturbed mono-processing one. To make a more succinct and manageable description of the obtained control, p-time coloured Petri nets are defined. A typical illustration of an electroplating line is given.