Search results

The lightning phenomenon is one of the main threats in photovoltaic (PV) applications. Suitable protection systems avoid major damages from direct strikes but also nearby strikes may induce overvoltage transients in the module itself and in the power conditioning circuitry, which can permanently damage the system. The effects on the PV system sensibly depend on the converter topology and on the adopted power switch. In the present study, a comparative analysis of the transient response due to a nearby lightning strike (LS) is carried out for three PV systems, each equipped with a different converter, namely, boost, buck and buck–boost, based on either silicon carbide metal oxide semiconductor field effect transistors (SiC MOSFET) or insulated gate bipolar transistors controlled power switch devices, allowing in this way an analysis at different switching frequencies. The purpose of this paper is to present the results of the numerical analysis to help the design of suited protection systems.

Using a recently introduced three-dimensional semi-analytical method to simulate the electromagnetic transients caused in PV modules by nearby LSs, we investigate numerically the effect of a LS on the electronic circuits connecting the module to the alternate current (AC) power systems. This study adopts numerical simulations because experimental analyses are not easy to perform and does not grant a sufficient coverage of all statistically relevant aspects. The approach was validated in a previous paper against available experimental data.

It is found that the load voltage is not severely interested by the strike effects, thanks to the low pass filters present at the converter output, whereas a relatively high overvoltage develops between the negative pin of the inner circuitry and the “ground” voltage reference. The overcurrent present in the active switches is hardly comparable because of the different topologies and working frequencies; however, the highest overcurrent is observed in the buck converter topology, with SiC MOSFET technology, although it shows the fastest decay.

This research proposes, to the best of the authors’ knowledge, a comprehensive comparison of the indirect lighting strike effects on the converter connected to PV panels. A proper design of the lightning and surge protection system should take into account such aspects to reduce the risk of induced overvoltage and overcurrent transients.

Inverse problems in electromagnetism, namely, the recovery of sources (currents or charges) or system data from measured effects, are usually ill-posed or, in the numerical formulation, ill-conditioned and require suitable regularization to provide meaningful results. To test new regularization methods, there is the need of benchmark problems, which numerical properties and solutions should be well known. Hence, this study aims to define a benchmark problem, suitable to test new regularization approaches and solves with different methods.

To assess reliability and performance of different solving strategies for inverse source problems, a benchmark problem of current synthesis is defined and solved by means of several regularization methods in a comparative way; subsequently, an approach in terms of an artificial neural network (ANN) is considered as a viable alternative to classical regularization schemes. The solution of the underlying forward problem is based on a finite element analysis.

The paper provides a very detailed analysis of the proposed inverse problem in terms of numerical properties of the lead field matrix. The solutions found by different regularization approaches and an ANN method are provided, showing the performance of the applied methods and the numerical issues of the benchmark problem.

The value of the paper is to provide the numerical characteristics and issues of the proposed benchmark problem in a comprehensive way, by means of a wide variety of regularization methods and an ANN approach.

This study aims to investigate the possible use of a deep neural network (DNN) as an inverse solver.

Different models based on DNNs are designed and proposed for the resolution of inverse electromagnetic problems either as fast solvers for the direct problem or as straightforward inverse problem solvers, with reference to the TEAM 25 benchmark problem for the sake of exemplification.

Using DNNs as straightforward inverse problem solvers has relevant advantages in terms of promptness but requires a careful treatment of the underlying problem ill-posedness.

This work is one of the first attempts to exploit DNNs for inverse problem resolution in low-frequency electromagnetism. Results on the TEAM 25 test problem show the potential effectiveness of the approach but also highlight the need for a careful choice of the training data set.

The problem of direct measurement of magnetic fields is in several cases not trivial when the region of interest is not directly accessible, or when conventional probes could be of disturbance. There are several “indirect” schemes, based on different physical effects, proposed as alternative to direct probing. We discuss here a novel technique for the identification of 1‐D and 2‐D magnetic fields, based on the analysis of trajectories of charged projectiles crossing the region under analysis. The magnetic field map is obtained through an inverse formulation of the motion problem of charged particles in the field. The inverse problem is solved by minimization procedures that identify the coefficients of suitable representations for the field. In this way we are able to identify typical magnetic field configurations with great accuracy even in presence of noisy data. We give an overview on the method for the 1‐D and 2‐D cases, with reference to the field representation and the used optimization algorithms; moreover, the robustness of the technique in presence of noisy data is assessed via a numerical analysis.

The identification of magnetic field profiles is crucial in many applications where a direct measurement is difficult. We discuss here a technique, based on the injection of charged particles in the region under examination, which promises to be an innovative and effective tool in the analysis of 1‐D field profiles in high current plasma discharges. After the decription of the inverse problem related to the field construction, we consider a suitable discrete identification scheme, and analyze some properties of the latter. The field map in the interest region is reconstructed via a minimization procedure, which identifies the coefficient of a well‐suited expansion for the field. In particular, we discuss the precision and robustness of the identification procedure, with respect to the chosen optimization scheme, the amount of data, the order in the field expansion, and the influence of noise on the data.

The paper seeks to present a novel pulsed eddy currents (PEC) non‐destructive technique to investigate deep cracks in metallic structures.

The method is based on the time‐domain analysis of the “defect response” and joins the PEC approach with the exploitation of the peculiar auto‐correlation properties of the Galois sequences. The procedure, relying on the deconvolution of Galois sequences, greatly improves the signal to noise ratio (SNR) and thus the operating depth. De‐convolving even short Galois sequences allows one to investigate at depth ranges larger than those allowed by the conventional pulsed excitation techniques.

The technique has been tested on a benchmark and compared with numerical simulations. The experimental results showed that the SNR and the detection depth range have been significantly improved.

Some limitations of the measuring set up were evidenced requiring a new measuring apparatus if explorations at larger depths are of interest: the 0.1 per cent impedance differences among the four coils in the bridge, although limited by an accurate construction, resulted in a limitation of the measuring system in DSP procedure adopted to null the background signal: different probe configurations must be pursued in order to allow further improvements in the deep defect detection.

For the first time the peculiar Galois sequences excitation was applied in a PEC system. By using these inputs the signal energy was significantly enhanced, allowing one to reconstruct by the deconvolution process the crack signatures.

The purpose of this paper is to analyze the impact of different current representation models in the high field magnets characterization. Inverse source methodology used for current reconstruction is discussed. The regularizing effect of successive field map computation in different regions is also assessed.

Under suitable hypotheses, the resulting inverse source problem is linear, and different current representation bases are used to assemble different matrices. Properties of matrices are then assesses using SVD. The following field computation problem is also formulated using a projection matrix, and the properties of combined matrix operators are analyzed and compared to the inversion matrix.

The characteristics of the inverse matrix depend on the choice of the current representation basis, but in any case the application of the further projection matrix has a relevant regularizing effect.

The method is intrinsically tied to the linearity assumption, and the regularizing effect of the projection operator is stronger for further field regions.

The accuracy in the current reconstruction procedure can be reduced if data will be used only to compute field in distant regions.

The paper casts the problem of field computation in distant regions from magnetic measurements in the language of direct and inverse operators, allowing to assess its properties and fine tune the procedure parameters to achieve satisfactory results with minimum effort.

The purpose of the present paper is to show a comparative analysis of classical and modern heuristics such as genetic algorithms, simulated annealing, particle swarm optimization and bacterial chemotaxis, when they are applied to electrical engineering problems.

Hybrid algorithms (HAs) obtained by a synergy between the previous listed heuristics, with the eventual addiction of the Tabu Search, have also been compared with the single heuristic performances.

Empirically, a different sensitivity for initial values has been observed by changing type of heuristics. The comparative analysis has then been performed for two kind of problems depending on the dimension of the solution space to be inspected. All the proposed comparative analyses are referred to two corresponding different cases: Preisach hysteresis model identification (high dimension solution space) and load‐flow optimization in power systems (low dimension solution space).

The originality of the paper is to verify the performances of classical, modern and hybrid heuristics for electrical engineering applications by varying the heuristic typology and by varying the typology of the optimization problem. An original procedure to design a HA is also presented.

Inductances of complex coils, in the presence of linear materials only, can be computed by discretizing coils into simpler elements, whose magnetic behavior is analytically expressible, and suitably combining elementary contributions. Reliable results require high numbers of elements. In such cases, advantages can be taken from Graphic Processor Unit (GPU) capabilities of dealing efficiently with high numbers of repeated simple computational tasks. The purpose of this paper is to set up a fast and prompt numerical procedure to cope with the above described task.

The coils are first decomposed into current segments, taking into account accuracy, relative position and shape of coils to determine the number of segments. An analytical formula is then used to compute elementary contributions using GPUs to speed up the process, and finally superposition is used to recover the result.

The main advantages of the proposed approach are first demonstrated using simple examples, with analytical solutions, to validate the method accuracy and promptness, then more complex cases are taken to demonstrate its generality.

The method is intrinsically limited by the linearity assumption, excluding the presence of magnetic materials. The adopted formulas require in addition that coils must lie in free space.

The proposed method can help in the design of complex coils or coils systems, where the performance depends on total magnetic energy or magnetic forces among coils.

The paper presents an original implementation in GPU-based computational environment of a procedure to compute inductances, based on the superposition of a high number of current segments. The procedure includes an original method to self-adaptively define number and position of current segments used in the coils discretization.