Accurate Results Research Articles

A deep clustering framework for load pattern segmentation

In recent years, the widespread use of smart meters in power networks has generated a wealth of data from electricity customers. However, much of this real-world smart meter dataset lacks labeled data, posing a significant challenge that can be addressed by clustering. Traditional clustering methods often struggle in high-dimensional spaces, leading to less accurate results and increased computational demands. In response to these challenges, this research introduces a framework that utilizes a deep-learning-based clustering approach to address the issue of accurate customer clustering based on usage patterns in unlabeled data. Using an autoencoder, our approach integrates dimensionality reduction and clustering into an end-to-end unsupervised learning framework. Our algorithm significantly improves load profiling by tackling challenges related to nonlinear decision boundaries at the autoencoder bottleneck. Unlike traditional approaches, we propose separating the optimization of reconstruction and cluster loss, bridging the gap between clustering quality and reconstruction efficiency. We rigorously analyze the performance of our approach by comparing classical and state-of-the-art algorithms using two real-world smart meter data. We provide a comprehensive comparative analysis of our method against five common dimension reduction techniques used in high-dimensional clustering. The experimental analysis concludes that the proposed algorithm significantly enhances load profiling more than others, as confirmed through detailed load curve analysis and clustering validity indexes. This comprehensive assessment highlights the effectiveness and versatility of our proposed methodology when compared to others. Moreover, this research advances load profiling in smart grid analytics, providing practical insights for utilities and stakeholders looking to optimize power network operations.

Water Quality Prediction using Statistical, Ensemble and Hybrid models

<p>Water is an essential elixir for several living organisms to function and survive. But it gets contaminated through several sources such as industrial wastes, oil spills, marine dumping, etc. With a growing population, availability of good quality water is of grave importance. This has become the motivation to probe into analysis of water quality from the outcomes of Statistical and Ensemble methods and to find the best working models from both methods. Research has been done to predict water quality analysis using standalone statistical and ensemble models. So, this research focuses on obtaining the best Statistical and Ensemble model separately among the models tried. The statistical models implemented for comparison are Principal Component Analysis (PCA), Hierarchical Clustering Analysis (HCA), Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA). The Ensemble models used are Bagging, Boosting and Stacking. The models are then combined to build a Hybrid model to observe the comparisons between the three. The performance metrics used are Confusion Matrix, Accuracy, Precision, Recall, F1-score and ROC curve. While comparing the models, it is observed that Hybrid model produces the most accurate results, hence proving that the combination of Statistical and Ensemble model is efficient.</p>

AMFGP: An active learning reliability analysis method based on multi-fidelity Gaussian process surrogate model

Multi-fidelity modeling is widely available in theoretical research and engineering practice. Although high-fidelity models often necessitate substantial computational resources, they yield more accurate and reliable results. Low-fidelity models are less computationally demanding, while their results may be inaccurate or unreliable. For the reliability analysis based on complex limit state functions, a method based on active learning multi-fidelity Gaussian process model, called AMFGP, is proposed by combining surrogate model with adaptive strategy, ensuring a balance between prediction accuracy and computational cost in terms of both surrogate modeling and active learning: A dependent Gaussian process surrogate model using complete statistical characteristics is developed under the multi-fidelity framework, and the surrogate performances of different single-fidelity and multi-fidelity models with different learning functions are investigated; based on the proposed model, an adaptive strategy considering the dependence between predictions, the model correlation, and the sample density is designed, and the adaptive performance of different learning functions in different models is explored. The proposed method is validated for effectiveness and adaptability in three mathematical examples with different dimensions and demonstrated for efficiency and practicality in an engineering application to aero engine gear.

Deep learning based ultrasonic reconstruction of rough surface morphology

This paper introduces a methodology to recover the morphology of a complex rough surface from ultrasonic pulse echo measurements with an array of equidistant sensors using the one dimensional convolution neural network (1DCNN). The neural network is trained by the datasets simulated from high-fidelity finite element simulations for surfaces with a range of roughness parameters and is tested on both numerical and real experimental data. To assess the performance of our proposed method, the rough surface reconstruction results from the deep learning approach are compared with those obtained from conventional ultrasonic array imaging methods. Unlike array imaging-based methods that require a large number of sensors (e.g., 128, 64 or 32), the deep learning-based method uses pulse echo signals and can achieve accurate results with much fewer sensors. The developed deep learning approach has the potential to enable low-cost, accurate, and real-time reconstruction of complex surface profiles.

Weak signal extraction enabled by deep neural network denoising of diffraction data.

The removal or cancellation of noise has wide-spread applications in imaging and acoustics. In applications in everyday life, such as image restoration, denoising may even include generative aspects, which are unfaithful to the ground truth. For scientific use, however, denoising must reproduce the ground truth accurately. Denoising scientific data is further challenged by unknown noise profiles. In fact, such data will often include noise from multiple distinct sources, which substantially reduces the applicability of simulation-based approaches. Here we show how scientific data can be denoised by using a deep convolutional neural network such that weak signals appear with quantitative accuracy. In particular, we study X-ray diffraction and resonant X-ray scattering data recorded on crystalline materials. We demonstrate that weak signals stemming from charge ordering, insignificant in the noisy data, become visible and accurate in the denoised data. This success is enabled by supervised training of a deep neural network with pairs of measured low- and high-noise data. We additionally show that using artificial noise does not yield such quantitatively accurate results. Our approach thus illustrates a practical strategy for noise filtering that can be applied to challenging acquisition problems.

Accurate Geometries of Large Molecules by Integration of the Pisa Composite Scheme and the Templating Synthon Approach.

An effective yet reliable computational workflow is proposed, which permits the computation of accurate geometrical structures for large flexible molecules at an affordable cost thanks to the integration of machine learning tools and DFT models together with reduced scaling computations of vibrational averaging effects. After validation of the different components of the overall strategy, a panel of molecules of biological interest have been analyzed. The results confirm that very accurate geometrical parameters can be obtained at reasonable cost for molecules including up to about 50 atoms, which are the largest ones for which comparison with high-resolution rotational spectra is possible. Since the whole computational workflow can be followed employing standard electronic structure codes, accurate results for large-sized molecules can be obtained at DFT cost also by nonspecialists.

Mittag-Leffler wavelets and their applications for solving fractional optimal control problems

Herein, we design a new scheme for finding approximate solutions to fractional optimal control problems (OCPs) with and without delay. In this strategy, we introduce Mittag-Leffler wavelet functions and develop a new Riemann–Liouville fractional integral operator for these functions utilizing the hypergeometric function. The properties of the operational matrix have reflected well in the process of the numerical method and affect the accuracy of the proposed method directly. Employing the Riemann–Liouville fractional integral operator, delay operational matrix, and Galerkin method, the considered problems lead to systems of algebraic equations. An error analysis is proposed. Finally, some illustrative numerical tests are given to show the precision and validity of the suggested technique. The proposed method is very efficient for solving the OCPs with delay and without delay, and gives very accurate results.

Particle Receiver Models for Systems Analysis

Particle receivers are gaining importance in the field of Concentrating Solar Power (CSP) due to the high temperature that particles can achieve without degradation. Several researchers are studying the potential of this technology by means of system analyses, which need simple and light models of the system. This study presents two simple models for the particle receiver. The simplest model is a correlation obtained by fitting the results calculated with a more complex receiver model simulated in CFD. The other model is a 1D model, which is benchmarked against the same CFD results. Although both models achieve high coefficient of determination, R2, when compared to CFD results, the 1D model seems to provide more accurate results (especially during sunsets and sunrises). Both models are integrated into a tecno-economic model developed in previous work. The LCOE obtained with the 1D model is between 7% and 10% greater than the one obtained with the correlation.

Accuracy tests of the envelope theory

The envelope theory is an easy-to-use approximation method to obtain eigensolutions for some quantum many-body systems, in particular in the domain of hadronic physics. Even if the solutions are reliable and an improvement procedure exists, the method can lack accuracy for some systems. In a previous work, two hypotheses were proposed to explain the low precision: the presence of a divergence in the potential or the lack of a variational character for peculiar interactions. In the present work, different systems are studied to test these hypotheses. These tests show that the presence of a divergence does indeed cause less accurate results, while the lack of a variational character reduces the impact of the improvement procedure.

Simplified analytical model for prediction of collapse resistance of restrained steel beam-column substructure exposed to fire

This paper provides a simplified analytical model for the restrained steel beam-column substructure under a fire-induced progressive collapse scenario, which is applied to predict the collapse resistance of substructures under fire. The simplified model is formulated for axially and rotationally restrained two-span steel beam-column substructures under column-loss scenarios and involves rigid and semi-rigid connections. The response characteristics of restrained substructures exposed to fire under column-loss scenarios in each response stage are elaborately analyzed. A “negative” catenary stage, in consideration of the unique negative axial force effects induced by the temperature thermal expansion, was first introduced in the five-stage simplified model for restrained substructures exposed to fire against progressive collapse. Subsequently, the explicit calculation formulas for determining the vertical loads and deflection of both rigid and semi-rigid joint substructures exposed to fire were derived in a quantified way, based on rigid-plastic mechanisms and restraint coefficient methods. In addition, the corresponding numerical simulation analysis on a restrained two-span steel beam-column substructure with different end constraints and fire conditions was carried out to verify the applicability and reliability of established theoretical formulas. Comparisons of vertical load-deflection relationships between the calculation formulas and corresponding numerical simulation shows good accuracy. For more accurate prediction results, the error analysis was further conducted, and several theoretical formulas were modified, which were verified to be accurate and reliable after correction. With derived resistance functions, the resistance of steel beam-column substructures in fire-induced progressive collapse scenarios can be predicted.

Standard technical specifications for methacholine chloride (Methacholine) bronchial challenge test (2023)

Standard technical specifications for methacholine chloride (Methacholine) bronchial challenge test (2023)

Diagnosis and Prognosis of Lung Cancer & Lung Nodule using Machine Learning Techniques

Lung cancer is a serious and challenging cancer to diagnose. It frequently results in death in both men and women; thus, prompt, precise nodule analysis is crucial to the course of treatment. Early cancer detection has been accomplished through a variety of techniques. This research compares machine learning techniques for lung cancer nodule detection. To find anomalies, we used machine learning techniques such as principal component analysis, K-nearest neighbors, support vector machines, Naïve Bayes, decision trees, and artificial neural networks. We examined every technique with and without preprocessing. According to the experimental results, decision trees produce the most accurate results with 93,24% effectiveness without image processing while artificial neural networks produce the finest results with 82,43% effectiveness after image processing.

Insights into the Electronic, Optical, and Anti-Corrosion Properties of Two-Dimensional ZnO: First-Principles Study

The electronic, optical, and anticorrosion properties of planer ZnO crystal and quantum dots are explored using density functional theory calculations. The calculations for the finite ZnO quantum dots were performed in Gaussian 16 using the B3LYP/6-31g level of theory. The periodic calculations were carried out using VASP with the plane wave basis set and the PBE functional. The subsequent band structure calculations were performed using the hybrid B3LYP functional that shows accurate results and is also consistent with the finite calculations. The considered ZnO nanodots have planer hexagonal shapes with zigzag and armchair terminations. The binding energy calculations show that both structures are stable with negligible deformation at the edges. The ZnO nanodots are semiconductors with a moderate energy gap that decreases when increasing the size, making them potential materials for anticorrosion applications. The values of the electronic energy gaps of ZnO nanodots are confirmed by their UV-Vis spectra, with a wide optical energy gap for the small structures. Additionally, the calculated positive fraction of transferred electrons implies that electron transfer occurs from the inhibitor (ZnO) to the metal surface to passivate their vacant d-orbitals, and eventually prevent corrosion. The best anti-corrosion performance was observed in the periodic ZnO crystal with a suitable energy gap, electronegativity, and fraction of electron transfer. The effects of size and periodicity on the electronic and anticorrosion properties are also here investigated. The findings show that the anticorrosion properties were significantly enhanced by increasing the size of the quantum dot. Periodic ZnO crystals with an appropriate energy gap, electronegativity, and fraction of electron transfer exhibited the optimum anticorrosion performance. Thus, the preferable energy gap in addition to the most promising anticorrosion parameters imply that the monolayer ZnO is a potential candidate for coating and corrosion inhibitors.

Statistical Applications in Pharmaceutical Product Development: A Comprehensive Review

The pharmaceutical industry often depends on statistics to improve product quality, optimize the compositions, streamline operations, and conform to regulatory requirements. In Bio Equivalency investigations, Statistics plays an integral role in decision-making while demonstrating the similarity between the test and reference products. In this review, we explore how statistics play a crucial role in pharmaceutical product development. We start with examining experiments, where statistics help scientists find the best combinations of ingredients and manufacturing conditions. We explored different types of experiments, like factorial designs, Plackett-Burman designs, Box-Behnken designs, and Taguchi designs. While optimizing the formulations and process parameters, these methodologies assist researchers in making meaningful conclusions supported by statistical inference. The next sections include Response Surface Methodology (RSM) and Mixture Designs, which help in the development of drug formulations and the comprehension of complex responses. The development of pharmaceuticals is made simpler by these approaches. Then, we examine optimal designs like A-optimal, I-optimal, and D-optimal. A-optimal designs save resources, I-optimal designs focus on precise estimates, and D-optimal designs are great for screening and getting accurate results. These designs help scientists make decisions based on data. The starting point for demonstrating dissolution similarity between generic and reference products is the statistical approach. This paper examines the use of model-dependent techniques, such as zero-order, first-order, Higuchi, and Weibull models, and model-independent techniques such as F1, F2, Bootstrap, MSD, and PCA techniques, it analyzes important parameters and bounds that are essential for determining the dissolution similarity. In the field of biostatistics, we discuss how statistics ensures that generic drugs are safe and work just as efficiently as brand-name drugs. Both regulatory approval and patient safety rely on this. We briefly discuss statistical process control (SPC) in closing. SPC checks that drugs/drug products are consistently of quality using graphs and figures. It facilitates the production of consistently effective pharmaceuticals by organizations. Advanced statistical techniques, including data mining, machine learning, and artificial intelligence, are revolutionizing the pharmaceutical industry. These tools enhance drug discovery, optimize manufacturing processes, and pave the way for personalized medicine. As real-time monitoring and predictive analytics become integral, the future holds exciting possibilities for improving patient outcomes. This review shows how statistics are like a guiding tool in the pharmaceutical industry. Whether it's in the lab or in manufacturing, statistics play a big role in making sure patients get the best medicines possible.

Explaining decisions of a light-weight deep neural network for real-time coronary artery disease classification in magnetic resonance imaging

In certain healthcare settings, such as emergency or critical care units, where quick and accurate real-time analysis and decision-making are required, the healthcare system can leverage the power of artificial intelligence (AI) models to support decision-making and prevent complications. This paper investigates the optimization of healthcare AI models based on time complexity, hyper-parameter tuning, and XAI for a classification task. The paper highlights the significance of a lightweight convolutional neural network (CNN) for analysing and classifying Magnetic Resonance Imaging (MRI) in real-time and is compared with CNN-RandomForest (CNN-RF). The role of hyper-parameter is also examined in finding optimal configurations that enhance the model’s performance while efficiently utilizing the limited computational resources. Finally, the benefits of incorporating the XAI technique (e.g. GradCAM and Layer-wise Relevance Propagation) in providing transparency and interpretable explanations of AI model predictions, fostering trust, and error/bias detection are explored. Our inference time on a MacBook laptop for 323 test images of size 100x100 is only 2.6 sec, which is merely 8 milliseconds per image while providing comparable classification accuracy with the ensemble model of CNN-RF classifiers. Using the proposed model, clinicians/cardiologists can achieve accurate and reliable results while ensuring patients’ safety and answering questions imposed by the General Data Protection Regulation (GDPR). The proposed investigative study will advance the understanding and acceptance of AI systems in connected healthcare settings.

Точка росы как основа оперативного показателя лесопожарной опасности

The problem of preventing forest fires and timely mobilization of extinguishing agents is particularly acute in the Irkutsk Region, where forests cover considerable areas. The use of classical indices for assessing forest fire danger does not always provide accurate results in the operational planning of fire extinguishing measures. The aim of this study is to develop a simple and effective indicator for predicting the occurrence of forest fires. The forestries of the Irkutsk Region, which were most severely affected by forest fires, were chosen as the object of this study. In the course of the study, it has been necessary to solve a number of problems, such as assessing the adequacy of the existing methods, developing a more accurate and simple indicator for local conditions, establishing the algorithm for calculating and the scale for assessing the fire danger class. As the initial data, the publicly available data from the meteorological observations archives and the data from the Remote Monitoring Information System of the Federal Forestry Agency (ISDM-Rosleskhoz) have been used. The methods of cluster analysis and correlation and regression analysis have been applied. The dependence of the number of forest fires on weather conditions in sparsely populated areas has been revealed. The frequency of the cycles of the number and area of forest fires is 3–4 years. Various options for calculating the accumulated dewpoint deficit for predicting the number and area of forest fires have been considered. The closest correlation has been found for the indicator calculated over 10 days. Foreign experience in assessing fire danger according to weather conditions having been analyzed, an assumption has been made about the potential partial use of the Australian FFDI index. In the calculations, a satisfactory result has not been achieved. Therefore, it has been proposed to use the accumulated dew-point deficit over 10 days as an indicator of forest fire danger. The number of forest fires is described through the quadratic dependence on this indicator. A scale for assessing fire danger, adapted to the scale based on the Nesterov index, has been developed. For practical use, the calculation algorithm in MS Office Excel has been described, which makes it possible to apply the proposed indicator in the forestries of the Irkutsk Region. The indicator based on the accumulated dew-point deficit over 10 days will make it possible to check the state of forest fire danger according to weather conditions in accordance with the developed scale using the described algorithm and the macro for calculation in MS Excel.

Разработка интеллектуальной системы с нечеткой стабилизацией управления в условиях неопределенности входных воздействий

Abstract. The problems of complex control systems functioning under conditions of input influences uncertainty are discussed. Existing approaches to solving these problems, based on the use of fuzzy control methods, adaptive control and intelligent algorithms, are described. The synthesis of an intelligent system with fuzzy stabilization control under conditions of input influences uncertainty is carried out to increase the efficiency of the object’s functioning. The use of fuzzy control makes it possible to take into account the input influences uncertainty and stabilize control based on a production model of knowledge representation, which makes the system more flexible and resistant to change. Optimal defuzzification method dynamic selection for the purpose of fuzzy stabilization control ensures the effective functioning of each specific system. An example is considered that implements an intelligent system with fuzzy stabilization of electric current control for an electroplating process with uncertainty about its duration, part area, temperature and acidity of the electrolyte. To confirm the effectiveness of the developed intelligent system, a computational experiment is carried out on the example of controlling the electroplating process of applying nickel coating in a Watts electrolyte using fuzzy inference by the Mamdani algorithm with the triangular norm and the Zadeh conorm. The results show that using an intelligent system with fuzzy stabilization control leads to a more accurate result (in terms of coating thickness) compared to using the most common defuzzification methods (centroid; bisector; smallest, middle and largest of maximum) on your own.

Financial Derivatives and Its Effects on Financial Performance of Deposit Money Banks in Nigeria

This study examined the relationship between financial derivatives and financial performance of deposit money banks (DMBs) in Nigeria from the period of 2013 to 2022 (10years). Financial derivatives [proxied with Financial Liabilities Derivatives (FLD), Foreign Exchange Derivatives (FED), Trading Income on Derivatives (TID), Loan and Advance to Customers (LADC) and Bank Size (BS)] (independent variables) and financial performance [proxied with Return on Assets (ROA)]. The Ex-Post Facto research design was used. Ex-Post Facto research design aids in answering the who, what, when, where, and how questions linked with a certain study problem. The ex-post facto research design is used to acquire information on the current state of a phenomenon and to define ‘what exists' in terms of variables or conditions in a setting that is specifically relevant to the issue under investigation. Data on financial derivatives and financial performance were obtained from the annual reports and accounts of ten (10) DMBs listed in Nigeria Exchange Group that has international presence. The data set was described using descriptive statistics, followed by the correlation analysis was used to ascertain the co-movement of the independent variables in relation to the dependent variable and several diagnostics tests. Since the data are panel series that the unit root test was conducted to ascertain if the data are stationary in order to have accurate regression result followed by single equation co-integration test while the Multiple Regression analysis were employed with the aid of E-VIEW version 9.0 for the purpose of testing the research hypotheses raised in chapter one. It evident that measures of financial derivatives used has mixed effects on ROE of DMBs in Nigeria. However, majority of the independent variables such TID, LADC and BS has significant effects on ROE of DMBs while FLD and FED established an insignificant effects on ROE of DMBs in Nigeria. Hence, the study

SOC estimation of lead–carbon battery based on GA-MIUKF algorithm

The paper proposes a SOC (State of Charge) estimation method for lead–carbon batteries based on the GA-MIUKF algorithm. The GA-MIUKF algorithm combines GA (Genetic Algorithm) for global search and optimization with the MI-UKF (Multi-innovation Unscented Kalman Filter) algorithm for estimating the SOC of lead–carbon batteries. By establishing an equivalent circuit model for the battery, the GA is employed to globally search and optimize the battery model parameters and the noise variance parameters in the MI-UKF algorithm. Comparative analyses with the UKF (Unscented Kalman Filter) algorithms and MI-UKF algorithms reveal that the SOC estimation method based on the GA-MIUKF algorithm yields more accurate results for lead–carbon battery SOC estimation, with an average estimation error of 2.0%. This highlights the efficacy of the proposed approach in enhancing SOC estimation precision.

Accuracy characteristics of the testing bench for dynamic transmission error

A high-precision testing bench is required for accurately assessing the dynamic Transmission error (TE) of precision reducers. However, the relationship between the testing bench’s accuracy and deviations in testing conditions has not been determined, leading to an inaccurate evaluation of its metrological performance. In this study, we examined the accuracy characteristics of the testing bench. We analyzed the structure of the bench and identified the main error sources that affect its accuracy, including mechanical system deviation, sensor measurement deviation, and testing condition deviation. In addition, we studied the mechanism of these error sources and proposed a method to convert drive speed deviation and load torque deviation into angle deviation by using a three-dimensional characterization model of lost motion. Based on the expression of TE, we proposed an evaluation method for the accuracy of the testing bench. Furthermore, we presented the methods for calculating the error sources and proposed a method for determining the TE of adjacent units based on the specific coupling structure. Finally, we conducted practical tests to validate the effectiveness of the proposed method in evaluating the accuracy characteristics of the testing bench. We found that the relative test error between quasi-static TE and dynamic TE is small, and more accurate test results are obtained when applying an appropriate load.