Marc De Wulf

Measuring setup for the investigation of the influence of mechanical stresses on magnetic properties of electrical steel

An experimental setup is discussed allowing measurement of the magnetic properties at any angle between the magnetic flux and applied tensile stress. The magnetic measurements are performed locally by means of needles and H-coils. Some preliminary results for a unidirectional flux under different values of tensile stresses are presented.

Magnetic properties and loss separation in iron powder soft magnetic composite materials

New developments in powder metallurgical composites make soft magnetic composite (SMC) material interesting for application in electrical machines, when combined with new machine design rules and new production techniques. In order to establish these design rules, one must pay attention to electromagnetic loss characteristics of SMC material. In this work, five different series of iron based SMCs are produced and studied: (1) Pure iron powder with resin; (2) sintered iron based powders; (3) pure iron powder with additions of Zn-st and carbon; (4) iron based powder alloys (Fe,Nb,Si); (5) commercially available iron powder “Somaloy.” The specimens were shaped as rectangular rods and characterized on a miniature single sheet tester which was calibrated to Epstein. The measured energy losses are analyzed following the loss separation theory of Bertotti, in which the total energy loss is decomposed into hysteresis loss, classical Foucault loss, and an excess loss component.

Modeling of quasistatic magnetic hysteresis with feed-forward neural networks

A modeling technique for rate-independent (quasistatic) scalar magnetic hysteresis is presented, using neural networks. Based on the theory of dynamic systems and the wiping-out and congruency properties of the classical scalar Preisach hysteresis model, the choice of a feed-forward neural network model is motivated. The neural network input parameters at each time step are the corresponding magnetic field strength and memory state, thereby assuring accurate prediction of the change of magnetic induction. For rate-independent hysteresis, the current memory state can be determined by the last extreme magnetic field strength and induction values, kept in memory. The choice of a network training set is motivated and the performance of the network is illustrated for a test set not used during training. Very accurate prediction of both major and minor hysteresis loops is observed, proving that the neural network technique is suitable for hysteresis modeling.

Accurate field strength measurement in rotational single sheet testers

The accuracy of the field strength measurement in rotational single sheet testers (RSST) has been investigated by means of both finite element calculations and measurements on the built set-up. Shielding laminations should be used above and underneath the actual sample in order to obtain a two-dimensional and homogeneous magnetization and to measure the field strength accurately.

Dynamic hysteresis modelling using feed-forward neural networks

A computational model for dynamic hysteresis in laminated SiFe alloys is proposed, based on feed-forward neural networks. The model employs the loss-separation property of ferromagnetic materials and combines a rate-independent hysteresis model with a correction technique for dynamic effects at each time point. The model yields accurate prediction of BH loops for arbitrary waveforms and frequencies, as they occur in electrical motors.

Quasistatic measurements for hysteresis modeling

In order to validate and test the correctness of hysteresis models used in the simulation of soft magnetic materials (electrical steel), one has to dispose of experimental data describing the hysteresis behavior of the material e.g., measurements under slow time varying excitation conditions in order to reduce the dynamic effects to a strict minimum. Based on measured loops (1–5 Hz) on nonoriented 3% SiFe alloys, and starting from the statistical loss theory of Bertotti, where the total applied field is subdivided into a hysteresis, a classical and an excess field, one can obtain the dc loop of the material by eliminating the remaining dynamic effects in the lamination. A comparison between three types of measurements is made, namely, hysteresis loops obtained by extrapolating from dynamic measurements using the statistical loss theory, second loops measured under slow time varying induction (0,1 Hz)—the influence of the used frequency (0,1–5 Hz) and the used excitation wave form (constant dH/dt…constant ...

Combined Preisach–Mayergoyz-neural-network vector hysteresis model for electrical steel sheets

Three versions of a vector hysteresis model for electrical steel sheets are presented, based on the function approximation capabilities of feed-forward neural networks and the memory mechanism of vector hysteresis proposed by Mayergoyz. The first model handles arbitrary vector magnetization patterns, but requires a very extended data set for the training of the neural network. The second model is suitable for convex induction loci and allows a reduction of the required training set. The third model handles the features of the considered magnetization pattern in an alternative way and relaxes the convexity requirement. The choice of the specific model, its parameters, and the network training set depends on the types of magnetization patterns concerned. Arbitrary high accuracy can be reached by extending the complexity of the model and/or the size of the training set. Experimental results for the third model are presented and show the good accuracy of the approach. Standard neural network algorithms are used.

Field homogeneity in a two-phase rotational single sheet tester with square samples

Finite element method (FEM) calculations have been used to determine the geometry and field homogeneity in a two-phase rotational field single sheet tester with square samples. Mounting shielding laminations above and underneath the actual sample improves the field homogeneity. Chamfering the yoke also enhances field distribution, but this effect turns out to be less important than shielding the sample.

Magnetic shielding properties of sheet metal products taking into account hysteresis effects

Analytical expressions are presented to find the shielding effectiveness and the losses of a shield consisting of ferromagnetic, isotropic, nonlinear, and hysteretic material, characterized by the Preisach distribution function in the Rayleigh region. The nonlinear shield is divided into a sufficient number of piecewise linear sublayers with a permeability that is constant (space independent) and complex (to model hysteresis). Simulations of an infinitely long cylindrical shield in transverse sinusoidal flux show that the shielding of perfectly linear material is better than the one of nonlinear metal sheets. More hysteresis and nonlinearity deteriorate the shielding factor, as eddy current losses decrease.

Isotropic vector hysteresis modeling with feed-forward neural networks

A new vector hysteresis model is presented, based on the function approximation capabilities of feed-forward neural networks. Two-dimensional circular and elliptical magnetization of laminated SiFe steel sheets can be successfully handled by the model. A feed-forward neural network with four inputs, derived at each time step from the time-dependent magnetic induction vector, yields an accurate prediction of the magnetic field strength vector. Measurement results for a steel sheet sample are used to train and test the neural network. The model accuracy is good and can be easily adapted to the requirements of the application by extending or reducing the network training set and thus the required amount of measurement data. Besides, the presented technique is fast, requires no large data set, and applies standard neural network algorithms. Future extension of the model to other magnetization patterns is possible.