Elektronika ir Elektrotechnika

A New Fault Recognition Method Based on Empirical Mode Decomposition and Texture Attributes

2025-01-01T17:08:33+02:00

Small faults developed in coal seams are one of the major causes of coal mine accidents. Accurately predicting small faults in coal fields is an urgent requirement for efficient and safe production in coal mines. This article proposes a new small fault identification method that combines the empirical mode decomposition method and the seismic texture attribute extraction method to address the problem of large errors caused by noise in the results of small fault prediction. Firstly, the basic principles of the empirical mode method and the texture attribute method were studied, and then the fault recognition ability of this method was tested and analysed based on a small fault seismic forward modelling. Meanwhile, empirical mode decomposition is performed on actual seismic data to identify small faults by using texture attributes and by adding noise to the seismic record; this article compared the seismic record of texture properties in the presence and absence of noise. The results indicate that the texture attribute method can predict small faults well, but this method is easily disturbed by noise. The empirical mode decomposition method used in this paper can remove noise interference and highlight characteristics of the texture attribute. Therefore, the small fault prediction method that combines empirical mode decomposition with texture attributes can effectively identify small faults and play an important geological guarantee role in ensuring safe and efficient production in coal mines.

Title

2025-03-17T12:03:50+02:00

Energy Demand and CO2 Emission Forecast Model for Turkey with Deep Learning and Machine Learning Algorithms

2025-01-24T10:01:07+02:00

This study has conducted a forecast analysis of the energy demand and carbon dioxide (CO₂) emissions of Turkey, a developing country. Considering Turkey’s rapidly increasing energy demand, various economic and social parameters have been used for the years 1990-2024. Both machine learning and deep learning methods have been applied, and artificial neural network (ANN), convolutional neural network (CNN), recurrent neural network (RNN), long short-term memory (LSTM), and linear regression (LR) algorithms have been used for two models. The performance of these models has been assessed using various error metrics. The ANN has demonstrated the highest accuracy in modelling energy demand, achieving a coefficient of determination of 98.89 %, while the RNN has shown the best performance in modelling CO₂ emissions, with a coefficient of determination of 96.80 %. The findings have shown that the growth rates in energy demand and CO₂ emissions are high in the early years but slowed in the following years. However, it has been determined that the general trend continued to increase. The study emphasises the need for Turkey to diversify its energy sources and increase the use of renewable energy to meet its increasing energy demand. It also has concluded that accelerating efforts to achieve net zero emission targets are critical to long-term energy security and environmental sustainability.

Incremental Gate State Output Decomposition Model for Highway Traffic Forecasting Using Toll Collection Data

2025-01-07T16:09:39+02:00

Traffic flow on long-distance highways, especially at sections with multi-interchanges and ramps, exhibits nonlinear trends affected by long-term and short-term spatiotemporal dependencies, resulting limited fitting capabilities for the major applied spatiotemporal forecasting models in use. This paper tackles this challenge by integrating an incremental gate state output decomposition (IGOD) mechanism into the recurrent neural network (RNN) model framework, accounting for the interdependencies of spatiotemporal traffic data. The proposed method improves the ability of the RNN model to estimate traffic data series by segmenting consecutive time intervals and accumulating incremental changes across these time intervals, allowing for more precise traffic predictions. This study also explores how threshold amplitudes affect prediction effectiveness. We applied it to real traffic data from segment k602+630 to k625+420 on the Changjiu Highway. The results demonstrate that the proposed model consistently exhibits robustness, with variations in threshold magnitude having little impact on its prediction accuracy.

Performance Analysis of Two 8-Bit Floating-Point-based Piecewise Uniform Quantizers for a Laplacian Data Source

2025-01-07T15:57:32+02:00

In this paper, we employ the analogy between the representation of the floating-point (FP) format and the representation level distribution of the piecewise uniform quantizer (PWUQ) to assess the performance of FP-based solutions more thoroughly. We present theoretical derivations to assess the performance of the FP format and the PWUQ determined by this format for input data from the Laplacian source. We also provide a performance comparison of two selected 8-bit FP-based PWUQs. Beyond the typical evaluation of the applied FP format, through the accuracy degradation caused by the application of both FP8 solutions in neural network compression, we also use objective quantization measures. This approach offers insights into the robustness of these 8-bit FP-based solutions with respect to changes in input variance, which can be important when the input variance changes. The results demonstrate that the allocation of bits to encode the exponent and mantissa in the FP8 format is important, as it can significantly impact overall performance.

Editorial Board

2025-03-17T12:05:04+02:00

COLFSR - A Hybrid Random Number Generator Based on Chaos Optimisation and Linear Feedback Shift Register

2024-07-30T13:05:03+03:00

Many researchers are trying to make our lives easier with developments in the Internet of Things, industry 4.0, and artificial intelligence. However, when the security of the data, which is at the centre of all these developments, is not ensured, the processes that try to make the lives of human beings more comfortable turn into nightmares. The problem that is tried to be addressed in this study is to share the details of an approach that can be used as an encryption key in hardware encrypted data storage units that can be used to address security concerns that may arise during the transmission, processing, and storage of sensitive data. The proposed method has contributed to the hybrid random number generators, both by optimising the deterministic generators and the chaotic selection algorithm. The results of the successful analysis of the proposed architecture have confirmed that it will have potential in many practical applications in the future. It is thought that with projections for future studies, it will contribute to the field of global encryption software.

Ground Fissure Identification in Mining Areas from UAV Images Based on DN-CAMSCBNet

2025-01-07T16:42:53+02:00

The development and use of mine resources have had many adverse impacts on the environment of mining areas. Among them, ground fissures are the most serious. They not only threaten the ecological protection of mining areas but also hinder the sustainable exploitation of energy. To mitigate the damage to the ecological environment caused by mining areas and ensure sustainable long-term resource exploitation, it is of particular importance to identify ground fissures in mining areas efficiently. Therefore, this paper proposes a ground fissure identification model for UAV images in mining areas named DN-CAMSCBNet. This method integrates the channel attention mechanism and the dropout mechanism on the basis of the traditional U-Net. Meanwhile, it introduces the multiscale convolution block and Nesterov-accelerated adaptive moment estimation. These are used to enhance its ability to capture complex image features, expand the receptive field of the original model, reduce the number of parameters, and reduce computational complexity. To verify the segmentation performance of the model, it is compared with U-Net, D-CAMNet, and D-MSCBNet models. The experimental results show that the accuracy and precision of the DN-CAMSCBNet model can reach 99.47 % and 92.25 %, respectively, and the F1 score is 0.7699. All these are superior to comparison models and can provide strong support for the identification of ground fissures in mining areas.

Exploring the Impact of Imaging Resolution and Sharpness on Dermatological Diagnostics Using eSFR Measurements

2025-01-27T18:20:10+02:00

High-resolution, small-form-factor image sensors enable the integration of mobile device cameras, which are increasingly being used for photographic documentation in many fields, including medicine. With the interface and handheld dermatoscopy, the smartphone camera forms an alternative tool to professional dermatoscopic systems for performing teledermatology and teledermoscopy. For the accurate diagnosis of skin diseases, image quality is essential, with sharpness and resolution being essential criteria. This paper focusses on measuring the sharpness and resolution of cameras used for image acquisition in dermatology using the spatial frequency response (SFR) method, which is based on standardised test charts featuring characteristic slanted contrast edges, known as edge SFR (eSFR) charts. The images were captured with mirrorless and DSLR cameras, smartphones, and a professional dermatoscopy video camera under typical dermatological conditions with digital cameras, mobile phones, and professional video dermatoscopes. Captured images were analysed, and the modulation transfer function (MTF) was defined to evaluate the performance of different camera optical systems applied for dermatological imaging. The results provide insight into the strengths and limitations of the various imaging devices and highlight their effectiveness in meeting the requirements of dermatological practice.