To more accurately obtain the feature information embedded in the acoustic pattern of transformers, a transformer fault diagnosis method is proposed based on multilevel acoustic information of 14 state types. In this method, a parallel dual-channel fault diagnosis model, CNN-BiLSTM-Transformer, is established. First, the modified Mel inversion coefficients and Mel spectrograms are extracted from the original acoustic pattern data. The modified Mel inversion coefficients and Mel spectrograms are then input into the parallel dual-channel model. In the first channel, a convolutional neural network model is used to extract the feature information of maps. In the second channel, a bidirectional long- and short-term memory network and a Transformer encoder are used to partially extract the temporal features in the MFCCs. Finally, the temporal features extracted from the two channels are fused through multimodal fusion for training. The experimental results show that the proposed diagnostic method can achieve an average accuracy of 99.5% in multiple fault diagnosis. Compared with current mainstream acoustic single-channel diagnostic models, the diagnostic rate of this model is improved by an average of 4.8%, exhibiting higher accuracy and robustness.

This research presents the modelling of non-Fourier heat conduction using the Single and Dual-Phase-Lag (SPL/DPL) in the frequency domain. Thermal structures can be modelled using Foster, Cauer networks or their compact equivalents, using the analogy with electrical networks. The key aim of such modelling is to simplify the heat transfer processes of geometrically complex thermal structures composed of layers with different materials. In this paper, we propose the analytical solution of the heat transfer equation for a single-layer thermal object in the frequency domain. It is achieved by the transformation of Kirchhoff-Fourier equations into the multi-cell Cauer network that can be easily solved using the node potentials method. In the case of non-Fourier SPL and DPL models, the thermal conductivity is no longer a real value, but it varies with frequency. As a result, the spatially distributed thermal model can be presented as an electrical network consisting of thermal resistances, capacitances and inductances. It makes it possible to take into account diffusive, wave propagation and mixed conductive heat transfer. Such modelling allows confirming the non-Fourier heat transfer in porous-like materials and biostructures with a better understanding of the physical nature of heat transfer, which is still discussed in the scientific literature.

This paper proposes a modified geometric mean optimizer (MGMO) based on nonlinear functions of the control parameter, the first being a logarithmic function (MGMOI) and the second being an exponential function (MGMOII), while the original geometric mean optimizer (OGMO) has a linear control parameter. To demonstrate the ef-fectiveness and efficiency of an MGMOI and MGMOII, they were implemented to benchmark functions, the results observed a best balance in exploration and exploitation and faster convergence to the best solution compared to the OGMO. On the other hand, the OGMO, MGMOI and MGMOII algorithms have been implemented on the standard IEEE 33 bus and the practical IRAQI 71 bus to obtain the optimal position and capacity of photovoltaic distributed generator (PV-DG), while the beta probability distribution func-tion (BPDF) is implemented to model the uncertainty of solar irradiation. Moreover, the performances of radial distribution networks (RDNs) are improved after incorporating a PV-DG into the RDN; the power losses are minimized (65.159%, 65.500%, and 65.505% for the OGMO, MGMOI, and MGMOII, respectively, in the IEEE 33 bus, 66.739%, 66.858%, and 67.490% for the OGMO, MGMOI, and MGMOII, respectively, in the IRAQI 71 bus), while the voltage profile and stability are maximized (15.205%, 15.205%, and 15.230% for the OGMO, MGMOI, and MGMOII, respectively, in the IEEE 33 bus, 12.813%, 12.997%, and 12.997% for the OGMO, MGMOI, and MGMOII, respectively, in the IRAQI 71 bus). Finally, the proposed modified algorithms proved their superiority and dominance over the original algorithm in converging to the optimal solution for find-ing the optimal position and sizing of a PV-DG (after 28, 25, and 19 iterations for the OGMO, MGMOI, and MGMOII, respectively, in the IEEE 33 bus, after 27, 23, and 20 it-erations for the OGMO, MGMOI, and MGMOII, respectively, in the IRAQI 71 bus), and the performance of the MGMOII is better than the performance of the MGMOI because it requires fewer iterations to reach the optimal solution.

In order to ensure the safety of the train, the reliability of the catenary system in icy weather must be analyzed. The Kriging model was used to predict and analyze the reli-ability changes of the catenary system under three different icing conditions: rime, mixed, rime and rime. The change in icing thickness of the contact line of a high-speed railway was selected as a representative of the icing change of the catenary, and the actual icing cross-section was converted into an ice-covered cross-section under the ideal state that was easy to calculate, and the relationship between icing quality and thickness was discussed. The catenary components that are more affected by ice and snow weather are selected, the reliability of the catenary system is quantified, and the fault tree is used to analyze and construct the relevant model parameters. Compared with other catenary system reliability analysis methods, the prediction function of the Kriging model can meet the requirements of the analysis and study of catenary reliability. Comparing the simulation results of the models under the three icing conditions, it is found that the change rate of catenary reliabil-ity in the hoarfrost, mixed rime and glaze ice states increases sequentially, and the catenary reliability changes the least and causes the least harm in the hoarfrost state. The catenary reliability changes the most in the glace ice state, and the harm caused is also the most serious. The research results can further provide a reference for the selection of catenary maintenance and de-icing methods in icy weather.

This research investigates the application of nearest level control (NLC) in a three-phase transistor-clamped H-bridge (TCHB) multilevel inverter (MLI) to enhance power quality and reduce harmonic distortion. The TCHB topology offers high perfor-mance with fewer components compared to traditional MLI configurations, making it par-ticularly suitable for renewable energy applications. By operating at fundamental switching frequency, the NLC technique effectively minimizes switching losses while reducing total harmonic distortion (THD). The study includes MATLAB/Simulink-based simulation modeling and experimental validation using dSPACE, tested under varying load conditions. For a 13-level TCHB inverter with equal DC supplies and a modulation index of M = 1, the voltage THD was 5.22% (resistive load) and 5.17% (inductive-resistive load) in simula-tions, which was further reduced to 4.5% and 4.2% in experiments. The NLC technique demonstrated superior harmonic performance and efficiency compared to conventional methods, particularly at higher voltage levels. This study highlights the practical advantages of the TCHB inverter, including reduced component count, simplified control implementa-tion, and enhanced output waveform quality. These findings confirm the potential of the NLC-based TCHB inverter for high-performance motor drives and grid-tied renewable en-ergy systems, positioning it as a promising solution for modern power electronics applica-tions.

The paper presents the results of the author's research on effectively determining the initial conditions for the time-stepping model of a high-speed inverter-driven induction machine. The classical time-harmonic and multi-harmonic models based on the multidimensional effective magnetic permeability were used and compared as a preconditioner for the time-stepping model to speed up the steady-state solution. The carried-out simulation experiment proved that using both approaches radically accelerates computations. Furthermore, it has been shown that the multi-harmonic model is much more effective for problems with strong harmonic effects.

The paper is focused on improvements of the conventional speed controller based on Active Disturbance Rejection Control (ADRC) applied for a two-mass electric drive system. The described ADRC structure is based on load-side speed measurement. The paper compares conventional structure dynamics with overall system behavior in the case plant parameters are changed. The proposed ADRC algorithm extension performs soft controller parameters adjustment to improve the dynamics and plant response. The presented approach accomplishes adaptation capabilities with the use of a Radial Basis Function Neural Network (RBFNN). The article presents the dynamic response of the plant controlled by the conventional ADRC algorithm and the designed neural adaptation extension. The results are based on the performed experimental tests.

To optimize the parameter setting of the support vector machine and improve the classification performance and computational efficiency of power transformer fault diagnosis, this study proposes an improved grey wolf optimization algorithm. By optimizing the global search and local optimization capabilities of the greywolf algorithm and combining them with stacked denoising autoencoders, a new power transformer fault warning model is constructed. Firstly, the grey wolf optimization algorithm is optimized through four strategies: elite reverse learning, nonlinear control parameters, Lévy flight, and particle swarm optimization, which improve its global search and local optimization capabilities. Secondly, the stacked denoising autoencoder is utilized to extract high-level features of fault data, and the improved GWO algorithm and SVM are combined to complete fault classification. The results indicated that the proposed diagnostic model achieved a diagnostic accuracy of 0.979, a recall rate of 0.986, and an F1 value of 0.983 in benchmark performance testing. In practical applications, the average fault diagnosis accuracy of this model could reach up to 99.21%, and the average diagnosis time was only 0.08 s. The developed power transformer fault warning model can provide an efficient and reliable technical solution for fault diagnosis in the power system.

This paper proposes a novel improved hybrid permanent magnet Vernier machine (IHPMVM), which is characterized by less-rare-earth (LRE) and high torque-density. The proposed machine features a hybrid magnet arrangement, which adopts both rare earth (RE) and LRE magnets in one magnetic pole simultaneously. The proposed improved design can reduce the consumptions of RE materials by employing low-cost LRE magnets in place of RE magnets. Besides, the hybrid magnet arrangement design has a good magnetic flux-concentrated effect, resulting in high torque density. Particularly, dummy slots are introduced to achieve a flux modulation effect. This unique design effectively reduces the inevitable leakage flux, thereby further improving the utilization of PMs and torque density. Firstly, the machine configuration and its improved design are introduced and investigated. Then, a multi-objective optimization is carried out to obtain the optimal design of the proposed machine considering comprehensive performance. Furthermore, the preliminary electromagnetic characteristics of the proposed machine are compared and analyzed using finite element (FE) methods, which verifies the effectiveness of the optimization. Finally, the demagnetization risk of the LRE magnets is evaluated. This paper is expected to provide a technical reference for designing LRE machines.

To address the overvoltage problem caused by the reverse flow of current when a high proportion of distributed photovoltaic (PV) is connected to the distribution network, this paper proposes a grid-connected voltage regulation control strategy based on the cluster division of the Distributed Model Predictive Control (DMPC) algorithm. Firstly, the overvoltage responsibility of each node is calculated using the Shapley value method. This is combined with k-means clustering to achieve effective cluster division, enabling dynamic adjustment of the active and reactive power of photovoltaic power generation units to stabilize regional voltage. Secondly, a group grid-connected voltage control strategy is introduced. This strategy controls the active and reactive power outputs by integrating real-time power output and voltage information from PV generating units in the region with the DMPC algorithm, ensuring overall voltage stability of the grid-connected system. Finally, actual overvoltage data from a 10 kV distribution line in the Dingxi power grid, Gansu Province, is used to verify that under the proposed control strategy, PV grid-connected overvoltage nodes are maintained within 1.06 p.u. The control effect is improved by a margin of 0.05 compared to traditional control methods. This demonstrates the effectiveness of the grouped grid-connected voltage regulation control strategy, achieving smoother voltage regulation performance in distributed PV grid-connected systems.