• Energy Research

Abstract Wind energy evaluation is an important goal in the conversion of energy systems to more environmentally friendly solutions. In this paper, we present a novel approach to wind speed spatial estimation on the isle of Sicily (Italy): an incremental self-organizing neural network (Generalized Mapping Regressor – GMR) is coupled with exploratory data analysis techniques in order to obtain a map of the spatial distribution of the average wind speed over the entire region. First, the topographic surface of the island was modelled using two different neural techniques and by exploiting the information extracted from a digital elevation model of the region. Then, GMR was used for automatic modelling of the terrain roughness. Afterwards, a statistical analysis of the wind data allowed for the estimation of the parameters of the Weibull wind probability distribution function. In the last sections of the paper, the expected values of the Weibull distributions were regionalized using the GMR neural network.

Abstract This paper presents a method for modeling a PEMFC by using electrical circuits. In particular, it focuses on temperature and voltage distribution of fuel cell. The current distribution is calculated by using the Newton-Raphson method in order to estimate the physical parameters (connection resistances) of the model. Several test on a single PEMFC cell have been carried out during this study. In order to validate the model, temperature and voltage sensors have been installed in different segments of a single cell. A distinguishing advantage of the developed model is its ability to detect and localize the faults within the PEMFC cell, as well as simulate different faults in all of the three directions of the PEMFC cell.

Stator inter-turn faults (SITFs) are electrical abnormalities in the windings of a motor or generator, resulting from short circuits between adjacent coil turns, potentially leading to reduced performance or even catastrophic failures. This paper aims to detect SITFs and classify their level of severity using a combination of prominence-based features and recently developed neural networks that rely on self-attention mechanisms. The approach involves transforming 3-phase currents using the extended Park Vector approach (EVPA), extracting features based on prominence from the frequency spectrum, and studying their geometry to gain important insights about the data. After this feature-engineering and data exploration step, neural-based classifiers have been trained and tested. Through a comparative study with other neural-based approaches, the Transformer Encoder achieves the highest classification accuracy of 97.25% when tested using the experimental data, outperforming other trained networks. The authors also present the importance of self-attention maps for exploring the interpretability of the Transformer Encoder, revealing the significant contribution of the prominence-based features in classification.

Abstract One of the first steps for the exploitation of any energy source is necessarily represented by its estimation and mapping at the aim of identifying the most suitable areas in terms of energy potential. In the field of renewable energies this is often a very difficult task, because the energy source is in this case characterized by relevant variations over space and time. This implies that any temporal, but also spatial, estimation model has to be able to incorporate this spatial and temporal variability. The paper deals with the spatial estimation of the wind fields in Sicily (Italy) by following a data-driven approach. Starting from the results of a preliminary study, a novel technique resulting from the integration of neural and geostatistical techniques was developed in order to obtain the wind speed maps for the region at 10 and 50 meters above the ground level. The mean values of the theoretical Weibull distribution function describing the wind regime at each of the available measurement sites were used to train a multi-layer perceptron (MLP) whose goal is to compute the most of the wind spatial trends. Other pieces of information about the territory (altitude, land coverage) were also used as inputs of the network and organized into a geographic information system (GIS) environment. The remaining de-trended linear means have been computed by using a universal kriging (UK) estimator. The results of these steps were then summed up and it was thus possible to obtain a map of the estimated wind fields.

This paper deals with the scroll compressor, which is a machine used for compressing air or refrigerant. By using a novel reference frame, it proposes an original way of describing the geometry of the scroll wraps (represented as circle involutes) in which the symmetries are exploited in order to establish a thermodynamic model of the scroll compressor. This approach allows the chamber volumes to be analytically described without any special assumption and takes into account the discharge as a non-symmetrical process. The proposed geometric model is aimed to be coupled with the thermodynamic model by using the standardized VHDL-AMS language and should be then considered as preliminary to the scroll overall simulation and design of a functional virtual prototype. Simulations and experiments have shown good agreement.

This contribution wants to show some recent applications of neural networks in the field of power electronics, with particular emphasis on the sensorless control of AC drives and examines improvements that can be attained when using linear neural networks. At first it presents some theoretical aspects of linear neural networks, particularly the TLS EXIN neuron. Then it describes some original applications in electrical drives and power quality as follows: 1) least square and neural identification of electrical machines 2) neural sensorless control of AC drives 3)neural enhanced single-phase DG system with APF capability Simulation and experimental results are provided to validate the theories.

Abstract The exploitation of the renewable energy sources plays a key role for achieving the CO 2 emissions reduction targets established by the Kyoto Protocol, as well as for facing the shortage of world fossil fuels reserves. In countries like Italy, with an high potential in terms of wind power generation, an efficient energy planning based on renewables is a very complex task. It encompasses many aspects: the resource availability assessment, the compliance with environmental and legislative constraints and last, but not least, the technical aspects linked to the safe integration to the grid of the intermittent power generated by the wind farms. This paper is the first part of a study addressing the first of the aforementioned issues. The wind measurements recorded in several stations of Sicily (Italy) were used for the spatial modelling of the wind fields over the region. A statistical analysis of the wind data has allowed the estimation of the parameters of the wind probability distribution function, that is a Weibull, as predicted by theory. In the last sections of the paper the results of some traditional deterministic and geostatistical interpolation techniques are shown. In the companion paper the maps of the estimated wind fields have been obtained by using the results of the statistical investigation accomplished here and coupling neural and geostatistical techniques. For a comprehensive evaluation of the forecasting accuracy of this neural kriging approach, those maps have to be compared with the maps showed in this paper.

This paper presents a new model reference adaptive system (MRAS) speed observer for high-performance field-ori- ented control induction motor drives which employs the flux error for estimating the rotor speed, but overcomes the pure integration problems by using a novel adaptive integration method based on neural adaptive filtering. A linear neuron (the ADALINE) is employed for the estimation of both the rotor speed and the rotor flux-linkage with a recursive total least-squares (TLS) algorithm (the TLS EXIN neuron) for online training. This neural model is also used as a predictor, that is with no feedback loops between the output of the neural network and its input. The proposed scheme has been implemented in a test setup and compared with an MRAS ordinary least-squares speed estimation with low-pass filter integration, with the well-known Schauder's scheme and with the latest Holtz's scheme. The experimental results show that in the high- and medium-speed ranges with and without load, the four algorithms give practically the same results, while in low-speed ranges (that is, below 10 rad/s ) the TLS-based algorithm outperforms the other three algorithms. Successful experiments have also been made to verify the robustness of the algorithm to load perturbations and to test its performance at zero-speed operation.