Frédéric Wurtz

A methodology and a tool for the computer aided design with constraints of electrical devices

A methodology for the computer aided constrained design of electrical devices is presented and validated through the design of a slotless permanent magnet structure. It is based on the use of the analytical design equations of the device. Symbolic calculation is widely used to generate an analysis program and a sensitivity computation program. Those programs are linked with an optimisation algorithm that can take constraints into account. The methodology is tested with an experimental software named PASCOSMA.

An optimizer using the software component paradigm for the optimization of engineering systems

Purpose – This paper is a synthesis paper which seeks to discuss an optimisation framework using software components, which is a new emerging paradigm in computer science.Design/methodology/approach – The goal of this paper is to show the efficiency of the software component approach for the implementation of optimisation frameworks for engineering systems in general, and electromagnetic systems in particular.Findings – This paper highlights the component standard, a generator based on analytical expressions of the system, and an optimization service. References and examples show application in the area of electromagnetic components and systems.Practical implications – This paper presents CADES, a framework dedicated to system design, based on optimization needs. The framework is defined with a standard implementing the software component paradigm and a pattern to use it. Indeed, this pattern details how to create and use a component (the model of the device to design).Originality/value – This paper shows...

Automatic generation of sizing static models based on reluctance networks for the optimization of electromagnetic devices

During the presizing of electromagnetic devices, the designer needs fast calculation tools allowing him to make optimizations. Reluctance networks are thus very often used as approximate models: they are easy to interpret and provide very short computing times. However, to set them to equations is a tiresome task and often leads to errors. The paper proposes to automate this task. Starting from a simple and intuitive graphic interface, the model and its partial derivatives are automatically generated. Thus, optimization sizing and sensibility computation can be carried out

Automatic generation of sizing models for the optimization of electromagnetic devices using reluctance networks

Models of electromagnetic devices using reluctance networks are very interesting, especially for sizing and optimization. Unfortunately, too often, designers do not use them because they are time consuming to implement, mainly for complicated aspects like the need to write and solve the network equations. We have developed a tool that automates all those symbolic and numeric tasks. We also generate software components that allow to use those models very efficiently in an optimization process by providing the formal right sensitivity. We present the methodology and software, as well as their application on examples. This paper proposes also an analysis of the global design and modeling process with a reluctance network that defines the relative place of the designer versus methodologies and tools.

Neural Network-Based Model Design for Short-Term Load Forecast in Distribution Systems

Accurate forecasts of electrical substations are mandatory for the efficiency of the Advanced Distribution Automation functions in distribution systems. The paper describes the design of a class of machine-learning models, namely neural networks, for the load forecasts of medium-voltage/low-voltage substations. We focus on the methodology of neural network model design in order to obtain a model that has the best achievable predictive ability given the available data. Variable selection and model selection are applied to electrical load forecasts to ensure an optimal generalization capacity of the neural network model. Real measurements collected in French distribution systems are used to validate our study. The results show that the neural network-based models outperform the time series models and that the design methodology guarantees the best generalization ability of the neural network model for the load forecasting purpose based on the same data.

A Component-Based Framework for the Composition of Simulation Software Modeling Electrical Systems

This article deals with a component-based approach to design electrical systems modeled by several simulation software. The component paradigm can be applied to simulation software, thus taking advantage of composition facilities. To do so, a visual composition tool has been developed and is presented in this article. It produces recursively a global simulation component from several pieces of simulation software. It also deals with uncertainty by taking into account different kinds of components as well as different kinds of coupling methods. A specific coupling method for components dedicated to optimization is detailed. The sizing process of a home system filter is then presented to illustrate the use of the visual composer tool and this specific coupling method.

A methodology for the sizing and the optimization of an electromagnetic release

In this paper we present a multi modeling problem, and the design process that we manage to associate with it. In the first part, the different models are described. The reasons for such modeling are also exposed. Then the applied design process is detailed and justified.

Automatic differentiation for electromagnetic models used in optimization

Purpose – The purpose of this paper is to illustrate automatic differentiation (AD) as a new technology for the device sizing in electromagnetism by using gradient constrained optimization. Component architecture for the design of engineering systems (CADES) framework, previously described, is presented here with extended features.Design/methodology/approach – The paper is subject to further usage for optimization of AD (also named algorithmic differentiation) which is a powerful technique that computes derivatives of functions described as computer programs in a programming language like C/C++, FORTRAN.Findings – Indeed, analytical modeling is well suited regarding optimization procedure, but the modeling of complex devices needs sometimes numerical formulations. This paper then reviews the concepts implemented in CADES which aim to manage the interactions of analytical and numerical modeling inside of gradient‐based optimization procedure. Finally, the paper shows that AD has no limit for the input prog...

E-cooperative design among mechanical and electrical engineers: implications for communication between professional cultures

This paper looks at the collaborative design activity involved in a design experiment of an electromechanical plunger. Much of the coordination was achieved through Internet-based communication. As mechanical and electrical researchers involved in the design project, we discuss the information exchanges highlighted by our different professional cultures and relate how these exchanges lead us to propose some methodology to improve the efficiency of virtual meetings. Moreover, we show the need for new communication tools, ones dedicated to specific tasks that are not currently supported, especially shared concept formalization among technical experts.

Application of a new optimization approach to the design of electrical wheels

The goal of this paper is the optimization of the design of electrical wheels. For this, a new approach that automatically generates the optimization software is used. It allows the design to be made with an elaborated analytical model of the structure chosen for the wheel. Thus it is possible to obtain a very efficient optimization program. This approach is illustrated using the example of an electrical wheel for a bicycle.