Calin Ciufudean

Diagnosis of complex systems using ant colony decision Petri nets

Failure diagnosis in large and complex systems is a critical task. A discrete event system (DES) approach to the problem of failure diagnosis is presented in this paper. A classic solution to solve DESs diagnosis is a stochastic Petri nets. Unfortunately, the solution of a stochastic Petri net is severely restricted by the size of its underlying Markov chain. On the other hand, it has been shown that foraging behavior of ant colonies can give rise to the shortest path, which will reduce the state explosion of stochastic Petri net. Therefore, a new model of stochastic Petri net, based on foraging behavior of real ant colonies is introduced in this paper. This model can contribute to the diagnosis, the performance analysis and design of supervisory control systems.

Harmonie elimination and power factor improvement of three-phase rectifier using RNSIC variant

Three-phase rectifiers with low harmonics and power factor correction have been receiving wide interest in the last years. The proposed method is based on the rectifier with near sinusoidal currents which is called RNSIC and which functions as power factor compensation in a wide range of load currents. The decrease of input current harmonics, below the limits imposed by IEEE 519/1992 and IEC 61000-3-4 international standards, allows the decrease of THD, reduces the reactive power, hence the increase in real power factor. The performance of the proposed RNSIC in power factor correction (PFC) regime is verified by simulation, as well as by experimentation, under different loads.

An analysis of Rectifiier with Near Sinusoidal input current

The IEEE-519 and IEC 61000-3 standards limit the individual harmonics current as well as the total harmonic distortion (THD) for nonlinear power electronic loads. We present a comparative study of a classical three-phase rectififier diode (TPRD) and RNSIC rectififier (Rectififier with Near Sinusoidal Currents), with respect to harmonic current distortion. Simulation results are experimentally verifified.

Holonic Models for Traffic Control Systems

This paper proposes a new time-placed net model for traffic control systems, respectively railway control traffic systems. This model can be interpreted as a holonic one, and contains three modules: Transport Planning Module, Transport Control Module and Priority Control Module. For railway traffic systems we introduce a strategy in a timed-place Petri net model to solve collision and traffic jam problems.

Petri Net Based Diagnosis for Construction Design

1 Abstract— In this paper we propose a Petri net within a diagnosis system for construction design. We assimilate the construction process with an assembly process that composes parts and/or subassemblies into a unique product. We assume that the assembly supervisor (AS) system is distributed, and it solves several local AS attached to the nodes of the Petri net model of the assembly process. The research issues that we address in this work include the modeling of assembly process, determination of cost-effective assembly planes for efficient building, and real time adoption of a plan to a given product to be assembled. This work extends the known assembly Petri nets to a powerful framework enabling to derive the diagnosis of assembly process whose path may vary, and the objective function is maximized. The presented approach can be used to evaluate transient and steady-state performances of alternative design based on a construction example. Possible extensions of the work are also discussed.

Discrete Event Models for Flexible Manufacturing Cells

A manufacturing system includes a set of machines performing different operations, linked by a material handling system. A major consideration in designing a manufacturing system is its availability. When a machine or any other hardware component of the system fails, the system reconfiguration is often less than perfect. It is shown that, if these imperfections constitute even a very small percent of all possible system faults, the availability of the system may be considerably reduced. The system availability is computed as the sum of probabilities of the system operational states. A state is operational when its performance is better than a threshold value. In order to calculate the availability of a manufacturing system, its states (each corresponding to an acceptable system level) are determined. A system level is acceptable when its production capacity is satisfied. To analyze the system with failure/repair process, Markov models are often used. As a manufacturing system includes a large number of components with failure/repair processes, the system-level Markov model becomes computationally intractable. In this paper, a decomposition approach for the analysis of manufacturing systems is decomposed in manufacturing cells. A Markov chain is constructed and solved for each cell i to determine the probability of at least Ni operational machines at time t. Ni satisfies the production capacity requirement of machine cell i. The probability is determined so that the material handling carriers provide the service required between Ni operational machines in machine cell i, and Ni+1 operational machines in machine cell i+1. The number i=1,...,n at time t, where n is the number of machine cells in the decomposed system. Production lines are sets of machines arranged so as to produce a finished product or a component of a product. Machines are typically unreliable and experience random breakdowns, which lead to unscheduled downtime and production losses. Breakdown of a machine affects all other machines in the system, causing blockage of those upstream and starvation of those downstream. To minimize such perturbations, finite buffers separate the machines. The empty space of buffers protects against blockage and the full space against starvation. Thus, production lines may be modeled as sets of machines and buffers connected according to a certain topology. From a system theoretic perspective, production 2

Scheduling availability of discrete event systems

Fault detection is a crucial and challenging task in the automatic control of complex systems, e.g. in flexible manufacturing systems (FMSpsilas) as a representative class of discrete event systems (DESpsilas). In this paper a discrete event system approach to the problem of failure diagnosis is presented. We propose a systematic procedure for detection of failure events using diagnoses implemented with stochastic coloured Petri nets (SCPN). An analytical approach for the availability evaluation of cellular manufacturing systems (as basic components of FMSpsilas) is presented, where a FMS is considered operational as long as its production capacity requirements are satisfied. The approach is used to evaluate transient and steady-state performance of alternative designs based on an industrial example. Possible extensions of the work are also discussed.

Control Charts of Workflows

This paper focuses on the control of the performance characteristics of workflows modeled with stochastic Petri nets (SPNs). This goal is achieved by focusing on a new model for Artificial Social Systems (ASSs) behaviors, and by introducing equivalent transfer functions for SPNs. ASSs exist in practically every multi-agent system, and play a major role in the performance and effectiveness of the agents. This is the reason why we introduce a more suggestive model for ASSs. To model these systems, a class of Petri nets is adopted, and briefly introduced in the paper. This class allows representing the flow of physical resources and control information data of the ASSs components. In the analysis of SPN we use simulations in respect to timing parameters in a generalized semi-Markov process (GSMP). By using existing results on perturbation (e.g., delays in supply with raw materials, derangements of equipments, etc.) analysis and by extending them to new physical interpretations we address unbiased sensitivity estimators correlated with practical solutions in order to attenuate the perturbations.

Safety Discrete Event Models for Production Lines with Shared Resources

In this paper we study the problem of deadlock avoidance of production lines with shared resources. We use the Petri net model for production lines and a restrictive policy which prevents some enabled transition from firing for avoiding deadlock in the system. This is the safety control for non-stochastic discrete event systems. We generalise this notion of safety to the setting of stochastic discrete event systems, modelled as Markov chains. We propose a restriction policy for this generalised case.

Intelligent control of dependability workflows

This paper focuses on evaluation of the performance characteristics of workflows in constructions modeled with stochastic Petri nets (SPN). This goal is achieved by focusing on a new model for artificial social systems (ASS) behaviors. ASS exists in practically every multi-agent system, and play a major role in the performance and effectiveness of the agents. This is the reason why we introduce a more suggestive model for ASS. To model these systems, a class of Petri nets is adopted, and briefly introduced in the paper. This class allows representing the flow of physical resources and control information data of the ASSpsilas components. In the analysis of SPN we use simulations in respect to timing parameters in a generalized semi-Markov process (GSMP). By using existing results on perturbation (e.g., delays in supply with raw materials, derangements of equipments, etc.) analysis and by extending them to new physical interpretations we address unbiased sensitivity estimators correlated with practical solutions in order to attenuate the perturbations. An important advantage of this approach is that one simulation is needed for evaluating the stochastic Petri nets and the perturbation analysis and to take advantage of the state of the art.