Scipedia
Not a member yet
33380 research outputs found
Sort by
Adjoint-based Stress Identification via Material Tensor Reconstruction in Membrane Structures using Displacement or Strain Measurements
No full text
Enhancing Wind Power Forecasting Using Hybrid Multi-Head Attention and 1-Dimensional Convolutional Neural Networks
No full textThe accurate forecasting of wind power plays a veritable part in integrating renewable energy from wind turbines into power grids. Wind power, being a highly volatile mode of energy generation owing to temporal variations and complex weather patterns, renders reliable predictions essential for energy management and grid stability. In order to tackle this, we propose a hybrid Multi-Head Attention and 1D-Convolutional Neural Network (MHA-CNN) architecture that combines attention mechanisms and convolutional layers to capture both long-term dependencies and localized features in time-series data from a Supervisory Control and Data Acquisition (SCADA) system. The model effectively improves forecasting performance by attaining an R2score of 99.42 for hour-ahead and 96.52 for day-ahead predictions on a 50,540-sample, 10-min SCADA dataset using 5-fold chronological cross-validation, outperforming traditional methods without any manual feature engineering. The proposed method is also evaluated across multiple scenarios to assess the robustness of the proposed approach.OPEN ACCESS Received: 01/10/2025 Accepted: 10/11/2025 Published: 23/01/202
Bearing Fault Diagnosis Based on AVMD and HPO-DBN
No full textTo overcome difficulties such as non-stationary vibrations, highdimensional feature redundancy, and mode selection issues that may arise during signal decomposition in bearing fault diagnosis. We propose an adaptive method called Adaptive Variational Mode Decomposition (AVMD) for extracting time-frequency domain characteristics from the bearing vibration displacement signals to the maximum extent possible. Next, the ReliefF algorithm is employed to select desired features, and an autoencoder is used to reduce the selected features dimensionally. Furthermore, because the Hunter-Prey Optimisation (HPO) algorithm can balance multiple objectives during the search process by utilising the concepts of hunter and prey to generate a better solution set, incorporating this algorithm into the Deep Belief Network (DBN) establishes an HPO-DBN fault diagnosis model. Subsequently, we validate the proposed method using both public datasets and field compressor data. Moreover, we compare the results with those obtained from the Support Vector Machine (SVM). The findings indicate that this approach enhances the bearing fault identification rate, thus supporting predictive maintenance of bearings.OPEN ACCESS Received: 13/08/2025 Accepted: 16/10/2025 Published: 23/01/202
Bayesian and Non-Bayesian Inference for Discrete Model Based on Censored Samples with Optimal Test Plan
No full textIn reliability and lifetime studies, it is often impractical to observe the failure times of all units in a sample, particularly when the process is timeconsuming or expensive. Type II censoring addresses this by terminating the experiment after a predetermined number of failures from a total sample of size. The Discrete Alpha Power Extended Inverted Weibull distribution is particularly suitable for modeling such censored discrete lifetime data. Its flexible shape parameters allow it to capture a wide range of failure behaviors, including over-dispersion, which is common in censored datasets. In this context, the likelihood function and estimation procedures (maximum likelihood and Bayesian) explicitly account for the censoring, ensuring unbiased parameter estimates and reliable predictive inferences. Consequently, the Discrete Alpha Power Extended Inverted Weibull distribution provides a practical and statistically robust framework for analyzing discrete lifetimes under type II censoring.OPEN ACCESS Received: 28/07/2025 Accepted: 19/09/2025 Published: 23/01/202
Analysis of Burr-XII Lifespan Using Adaptive Progressive Type-II Hybrid Binomial Censoring with Physical Modeling of Polyester and Carbon Fibers
No full textThis study introduces advanced statistical methods, allowing for more efficient and accurate reliability testing of fibers such as polyester and carbon. Polyester ficbers are suitable for textiles and industrial use due to their wrinkle resistance and affordability, while carbon fibers offer superior strength, thermal stability, and corrosion resistance. To guarantee greater efficiency of inference methodologies and reduce overall testing time, the adaptive Type-II progressive hybrid censoring via binomial removals has gained popularity in reliability analysis and life-testing problems. The proposed scheme allows survival units to be removed at random stages according to a binomial law, thereby reducing experimental time while preserving statistical efficiency. When lifetimes are gathered using the suggested censoring technique, point and interval estimates of the unknown parameters of the Burr-XII model are obtained using both classical and Bayesian approaches. We obtain various Bayesian estimates using the squared loss function. Some numerical methods are employed to obtain the suggested estimators due to their complexity. The various Bayes estimates and related credible intervals are created using Markov chain Monte Carlo techniques. To assess estimator performance, extensive simulation studies are conducted, comparing bias, mean squared error, coverage probabilities, and interval lengths under varying censoring and removal settings. The simulation results confirm that the Bayesian framework, particularly with informative priors, provides more accurate and stable estimates than asymptotic likelihood-based methods. We examine two physics data sets representing polyester and carbon fibers to demonstrate the relevance of the suggested approaches in a real-world setting. These applications highlight the practical value of the proposed approach for material design, maintenance planning, and broader reliability engineering problems.OPEN ACCESS Received: 13/06/2025 Accepted: 12/09/2025 Published: 23/01/202
Adaptive Power Splitting Strategies for Smart Microgrids with Enhancing Energy Efficiency and Resilience through Dynamic Load Management
No full textThe integration of a renewable energy distributed generation into microgrids poses a significant constraint in the way power is managed, further so due to the natural variability in renewable generation and the variability in the load demands. To address these issues, this paper introduces a novel approach to the Spider Swarm Optimization (SSO) algorithm, the Dynamic LoadAdaptive Power Splitting (DLAPS) strategy, to enable real-time adaptive power sharing and enhance system resilience. Unlike the classical methods of power allocation that are static, according to which the power is divided between sources of renewable energy and storage systems, and between these sources and critical loads, the DLAPS-SSO applies the idea of a machine learning based predictive model to predict the power and dynamically optimize power allocation between the sources of renewable energy and storage systems and the sources and the critical loads. The model provides a multi-objective optimization framework that aims to minimize power losses and grid frequency variations, and to maximize the system’s resilience to disturbances, including disconnection from the grid, component malfunctions, and the availability of renewable energy sources. The comparison of simulation results with those of the Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) methods shows that the energy efficiency of the DLAPS-SSO increases by 15%–20%, and the amount of power losses across various load profiles decreases by 30%–35%. Moreover, the proposed solution offers 60% faster recovery time in case of grid disconnection, maintains 65.9% of the critical load in case of component failure, and provides 40%–50% less resilience than state-of-the-art techniques. The analysis of seasons and real data shows that there is stability of the behavior with the increase of efficiency (18%–22% during winter, and 23%–25% during summer), and the ability of the suggested approach to be robust when changing plant configuration/operation. Integration of optimization of dynamic load management and adaptive power splitting will spur microgrid control strategies and offer a viable strategy to stabilize the grid, reduce operation costs, and enable sustainable changes in energy transformations. The results demonstrate the essential role of bio-inspired optimization and reactivity in the next generation of smart grids
The Effects of Energy Efficient Behavior Reminders on Electricity Conservation in a New York Public High School
No full textExcessive electricity consumption negatively affects the economic growth of developing countries, and is a major cause of carbon emissions throughout the globe. This experiment investigated how the use of energy efficient behavior reminder affects electricity conservation in a New York City Public High School. In this experiment, during the first 10 days of the experiment, custodians were not reminded to turn off lights. From Day 11-20, reminders were sent to custodians every Monday and Fridays to turn off the light. During Day 21-30, reminders were sent everyday to remind custodians to turn off the lights. The results showed daily reminders had a significant decrease in electricity consumption compared to when no reminders were sent, but there was no significant difference between the percentage of lights being on, making it difficult to confirm if the custodians followed the reminders. The reminder of turning off lights was not a major source of electricity consumption and other factors might have consumed a larger portion of electricity.. More schools should be studied in order to validate the results. Future studies/research can test the effects of energy saving behavior reminders on electricity conservation from turning off electronic devices, such as smart boards or computers, at the end of the day
Computational Investigation of Fractal-Fractional Nonlinear Viscoelastic Fluids Using Local Radial Basis Function Method
No full textFractal-fractional derivatives generalize both traditional and fractional differentiation approaches by integrating memory effects with fractal properties. This mathematical framework is especially valuable for describing complex systems in which conventional continuum mechanics becomes inadequate, particularly in scenarios involving porous or discontinuous structures. This research investigates the behavior of a non-linearWalter’s-B fluid subjected to time-varying thermal and concentration conditions. Beyond the extended derivative formulation, the analysis incorporates phenomena including first-order chemical reactions, radiative heat transfer, Joule heating, Soret effect, and viscous dissipation. Thesystem is also subjected to a transverse magnetic field with magnitude B0.The fluidmodel is initially formulated through traditional constitutive equations and subsequently generalized using a fractal-fractional operator. Solutions to this extendedmodel are computed employing ameshfree numerical approach utilizing localized radial basis functions (LRBF), which eliminates the requirement for structured grids and improves precision when addressing intricate geometries.The computational outcomes, displayed through graphical representations, illustrate how the fractional and fractal parameters influence the rheological characteristics of the Walter’s-B fluid. These findings establish that adjusting these parameters enables retrieval of classical, fractional, and fractal formulations as particular instances within this comprehensive mathematical structure