Laurent Capocchi

Wound-Rotor Induction Generator Inter-Turn Short-Circuits Diagnosis Using a New Digital Neural Network

This paper deals with a new transformation and fusion of digital input patterns used to train and test feedforward neural network for a wound-rotor three-phase induction machine windings short-circuit diagnosis. The single type of short-circuits tested by the proposed approach is based on turn-to-turn fault which is known as the first stage of insulation degradation. Used input/output data have been binary coded in order to reduce the computation complexity. A new procedure, namely addition and mean of the set of same rank, has been implemented to eliminate the redundancy due to the periodic character of input signals. However, this approach has a great impact on the statistical properties on the processed data in terms of richness and of statistical distribution. The proposed neural network has been trained and tested with experimental signals coming from six current sensors implemented around a setup with a prime mover and a 5.5 kW wound-rotor three-phase induction generator. Both stator and rotor windings have been modified in order to sort out first and last turns in each phase. The experimental results highlight the superiority of using this new procedure in both training and testing modes.

DEVSimPy: A Collaborative Python Software for Modeling and Simulation of DEVS Systems

The Modeling and Simulation (M&S) of complex systems leans on the collaboration between different actors coming from specific domains. These actors have to communicate through an efficient way in order to improve the M&S process. We therefore propose in this article a collaborative M&S software framework called DEVSimPy. We point out the use of DEVSimPy through a concrete case study: hydraulic network management.

Wound-rotor induction generator short-circuit fault classification using a new neural network based on digital data

This paper deals with a new transformation and fusion of digital input patterns used to train and test feed-forward neural network for a wound rotor three-phase induction machine winding short-circuits classification. Used input/output data have been binary coded in order to reduce the computation complexity. A new procedure, namely addition and mean of the set of same rank, has been handled to fuse binary bits to eliminate the redundancy due to the periodic character of input signals. However, this approach has a great impact on the statistical properties on the processed data in terms of richness and of statistical distribution. The proposed neural network has been trained and tested with experimental signals coming from current sensors implemented around a set-up with a prime mover and a 5.5kW wound rotor three-phase induction generator. The experimental results highlight the superiority of using this new procedure in both training and testing modes.

Experimental inter-turn short circuit fault characterization of wound rotor induction machines

The aim of this paper is to compare different experimental techniques for detecting a stator inter-turn short-circuit in a Wound rotor induction machine working as generator. Three methods using different signals such as stator or rotor current and the leakage flux will be described and experimented. The rotor currents provide interesting signatures since the stator faults introduce new harmonics in the rotor windings. Hence, in this paper, it is proved that sensing one rotor current by using a current sensor is one of possible techniques for detecting this kind of electrical fault. The leakage flux is another alternative for this kind of problems since in this paper; it is proved that a simple external leakage flux sensor is efficient to detect, in the low frequency range, the intern-turn short circuit in three-phase induction machines even in presence of power supply harmonics. Stator currents space vector is also an interesting technique since it is proved that inter-turn short circuit faults can be considered as a stator unbalance which can be detected using negative sequence of stator currents. The comparison of these three methods is performed using experiments on a 5.5kW–220V/380V–50Hz–8 poles wound-rotor induction machine working in both healthy and faulty modes at different load conditions.

Fuzzy inference models for Discrete EVent systems

For several years, we worked to improve a discrete events modeling formalism: called DEVS. Having defined a method to take into account the inaccuracies iDEVS, in this paper, we present the second part of our research work. Generally, our approach is to associate the DEVS formalism with an object class, which allows using it to new fields of study, and in our case fuzzy systems. This paper describes a new modeling methodology. It allows to modeling and to use fuzzy inference systems (FIS) with DEVS formalism in order to perform the control or the learning on systems described incompletely or with linguistic data. The advantages of this method are numerous: to extend the DEVS formalism to other application fields; to propose new DEVS models for fuzzy inference; to provide users with simple and intuitive modeling methods. Throughout this paper we describe the tools and methods which were developed to make possible the combination of these two approaches.

Inter-turn short circuit fault detection of wound rotor induction machines using Bispectral analysis

The aim of this paper is to develop a technique for detecting an inter-turn short-circuit in a wound rotor induction machine working as generator by using the bispectral analysis. Bispectral analysis is able to provide more information than power spectrum analysis. In the present investigation, the rotor and stator currents of the machine operating in both healthy and faulty modes are thoroughly analyzed. In the first stage, the bispectral analysis is applied to signals generated by a wound-rotor induction machine model developed by using MATLAB/Simulink© environment. This model allows the simulation of the inter-turn short circuit in the stator of the machine in any system with control circuits and/or connections to the grid by means of power electronics converters. Then, the same analysis has been performed using experiments on a 5.5kW-–220V/380V – 50Hz – 8 poles wound-rotor induction machine working in both healthy and faulty modes at different load conditions. Finally, a comparison between simulation and experimental results is given. Very promising results are obtained and presented. The results and the analysis indicate that bispectrum can be successfully applied to machine asymmetric faults, and stator winding fault analysis.

BFS-DEVS: A General DEVS-Based Formalism For Behavioral Fault Simulation

Discrete event modeling allows designing an easy-to-handle and reusable representation of a system but, in its classical form, only permits one simulation at a time for a system. Concurrent and Comparative Simulation (CCS) with Multi-List Propagation (MLP) appears to be an adapted solution, by providing a way to perform several simulations in a single execution run. Concurrent Fault Simulation (CFS) has been one of the first applications of the CCS. The main obstacles to a wide use of this technique are the high complexity of the concurrent simulation algorithms, along with the difficulty to integrate them in a simulation kernel. We focus in this paper on the CFS with MLP of systems described in the new BFS-DEVS formalism, which is an evolution of the original DEVS Simulator that integrates the CCS algorithm. The appli- cation is performed in the behavioral digital domain of systems described in the VHDL language.

System entity structure extension to integrate abstraction hierarchies and time granularity into DEVS modeling and simulation

The modeling and simulation (M&S) of complex systems often requires models described at different levels of detail characterized by differences in abstraction hierarchies and/or time granularity. The discrete-event system specification (DEVS) is a framework based on mathematical systems theory that offers a computational basis for application of M&S to systems engineering and that has become widely adopted for its support of discrete-event, continuous, and hybrid applications. A fundamental representation of DEVS hierarchical modular model structures is the system entity structure (SES), which represents a design space via the elements of a system and their relationships in a hierarchical and axiomatic manner. As has been described in a number of publications, the SES supports development, pruning, and generation of DEVS simulation models. The goal of this paper is to propose an extension of SES in order to integrate both the concepts of abstraction hierarchies and time granularity into DEVS. This paper explains in detail: (i) the concepts of abstraction hierarchies and time granularity; (ii) the extension of SES in order to take into account these concepts; (iii) DEVS M&S of complex systems according to different levels of detail (abstraction hierarchies and time granularity); (iv) the use of a Python DEVS simulator (DEVSimPy) to implement the management of abstraction hierarchies and time granularity. A real case study is given to illustrate the proposed approach, and follow-on research needed to implement the concepts is discussed.

Comparative study of two modelling implementation methods of a wound-rotor induction machine: Simulation, fault diagnosis and validation

The aim of this paper is to compare two modeling methods for a wound-rotor induction machine in order to simulate it in both healthy and faulty modes. The circuit-oriented approach which represents the machine model as a rotating transformer will be compared to the discrete event method. First, the wound-rotor induction machine model will be described using the classical equations in the abc reference frame. In fact, the circuit oriented method is based on a representation with only resistances, inductances and controlled voltage sources. It has been implemented by using the MATLAB/Simulink© software. With the discrete event method, the classical abc reference frame model is based on ordinary differential equations which are translated in block-scheme to be interpreted as a coupling of basic models based on a discrete event system specification. Finally, the performances of the two methods have been verified by comparison between simulation and experiments on a 5.5kW-220V/380V-50Hz-8 poles wound-rotor induction machine working in both healthy and faulty modes at different load conditions.

Simulation of AC Electrical Machines Behaviour Using Discrete Event System Simulator

This paper deals with the adaptation of AC electrical machine models for discrete event system simulator. The formalism chosen is DEVS (Discrete Event system Specification) which has been adapted recently for hybrid system simulation. The software PowerDEVS is close to MATLAB/Simulink(C) but without any toolbox adapted to power systems. As always, the model which is basically based on system of nonlinear differential equations has to be integrated to compute the output variables corresponding to any deterministic or stochastic inputs. In this way, the integrator atomic model has been optimized to perform many simulations with realistic CPU time. Results obtained with PowerDEVS and optimized graphical model have been validated by the same simulations using MATLAB/Simulink(C) as found in the classical literature.