Novel Meta-Heuristic Algorithm for Feature Selection, Unconstrained Functions and Engineering Problems

IEEE Access, Journal Year: 2022, Volume and Issue: 10, P. 40536 - 40555

Published: Jan. 1, 2022

This paper proposes a Sine Cosine hybrid optimization algorithm with Modified Whale Optimization Algorithm (SCMWOA). The goal is to leverage the strengths of WOA and SCA solve problems continuous binary decision variables. SCMWOA first tested on nineteen datasets from UCI Machine Learning Repository different numbers attributes, instances, classes for feature selection. It then employed several benchmark functions classical engineering case studies. applied solving constrained problems. two examples are welded beam design tension/compression spring design. results emphasize that outperforms comparative algorithms provides better accuracy compared other algorithms. statistical analysis tests, including one-way variance (ANOVA) Wilcoxon's rank-sum, confirm performs better.

Language: Английский

---

Feature Selection and Classification of Transformer Faults Based on Novel Meta-Heuristic Algorithm

Mathematics, Journal Year: 2022, Volume and Issue: 10(17), P. 3144 - 3144

Published: Sept. 1, 2022

Detecting transformer faults is critical to avoid the undesirable loss of transformers from service and ensure utility continuity. Transformer diagnosis can be determined based on dissolved gas analysis (DGA). The DGA traditional techniques, such as Duval triangle, Key gas, Rogers’ ratio, Dornenburg, IEC code 60599, suffer poor diagnosis. Therefore, recent research has been developed diagnose fault diagnostic accuracy using combined methods with artificial intelligence optimization methods. This paper used a novel meta-heuristic technique, Gravitational Search Dipper Throated Optimization Algorithms (GSDTO), enhance faults’ accuracy, which was considered novelty in this work reduce misinterpretation faults. robustness constructed GSDTO-based model addressed by statistical study Wilcoxon’s rank-sum ANOVA tests. results revealed that enhanced up 98.26% for all test cases.

Language: Английский

---

CDLSTM: A Novel Model for Climate Change Forecasting

Computers, materials & continua/Computers, materials & continua (Print), Journal Year: 2021, Volume and Issue: 71(2), P. 2363 - 2381

Published: Dec. 7, 2021

Water received in rainfall is a crucial natural resource for agriculture, the hydrological cycle, and municipal purposes. The changing pattern an essential aspect of assessing impact climate change on water resources planning management. Climate affected entire world, specifically India’s fragile Himalayan mountain region, which has high significance due to being climatic indicator. coming from rivers 1.4 billion people living downstream. Earlier studies either modeled temperature or area; however, combined influence both long-term analysis was not performed utilizing Deep Learning (DL). present investigation attempted analyze time series correlation (1796–2013) changes (1901–2015) over states India. Long Short-Term Memory (CDLSTM) model developed optimized forecast all states’ values. Facebook’s Prophet (FB-Prophet) implemented assess performance CDLSTM model. models assessed based various metrics shown significantly higher accuracies low error rates.

Language: Английский

---

Meta-Heuristic Optimization of LSTM-Based Deep Network for Boosting the Prediction of Monkeypox Cases

Mathematics, Journal Year: 2022, Volume and Issue: 10(20), P. 3845 - 3845

Published: Oct. 17, 2022

Recent technologies such as artificial intelligence, machine learning, and big data are essential for supporting healthcare monitoring systems, particularly Monkeypox confirmed cases. Infected uninfected cases around the world have contributed to a growing dataset, which is publicly available can be used by intelligence learning predict of at an early stage. Motivated this, we propose in this paper new approach accurate prediction based on optimized Long Short-Term Memory (LSTM) deep network. To fine-tune hyper-parameters LSTM-based network, employed Al-Biruni Earth Radius (BER) optimization algorithm; thus, proposed denoted BER-LSTM. Experimental results show effectiveness when assessed using various evaluation criteria, Mean Bias Error, recorded (0.06) prove superiority approach, six different models included conducted experiments. In addition, four algorithms considered comparison purposes. The approach. On other hand, several statistical tests applied analyze stability significance These include one-way Analysis Variance (ANOVA), Wilcoxon, regression tests. these emphasize robustness, significance, efficiency

Language: Английский

---

Solving Optimization Problems of Metamaterial and Double T-Shape Antennas Using Advanced Meta-Heuristics Algorithms

IEEE Access, Journal Year: 2022, Volume and Issue: 10, P. 74449 - 74471

Published: Jan. 1, 2022

This study offers an adaptive dynamic sine cosine fitness grey wolf optimizer (ADSCFGWO) for optimizing the parameters of two types antennas. The antennas are metamaterial and double T-shape monopoles. ADSCFGWO algorithm is based on technique recently developed powerful optimization techniques: a modified (GWO) value (SCA). suggested approach utilizes capabilities both algorithms to balance better exploration exploitation responsibilities process while achieving rapid convergence. First, new feature selection proposed choose most significant features from dataset using ADSCFGWO-based ensemble model optimal performance. also optimizes bidirectional recurrent neural network (BRNN) estimate monopole antenna characteristics. Several experiments were undertaken demonstrate superiority by comparing their results those existing algorithms, selectors, regression models. In addition, statistical analysis offered evaluate algorithm's effectiveness stability. achieved findings efficacy method over numerous competing algorithms.

Language: Английский

---

Pothole and Plain Road Classification Using Adaptive Mutation Dipper Throated Optimization and Transfer Learning for Self Driving Cars

IEEE Access, Journal Year: 2022, Volume and Issue: 10, P. 84188 - 84211

Published: Jan. 1, 2022

Self-driving car plays a crucial role in implementing traffic intelligence. Road smoothness front of self-driving cars has significant impact on the car's driving safety and comfort. Having potholes road may lead to several problems, including damage occurrence collisions. Therefore, should be able change their behavior based real-time detection potholes. Various methods are followed address this problem, reporting authorities, employing vibration-based sensors, 3D laser imaging. However, limitations, such as expensive setup costs danger discovery, affected these methods. it is necessary automate process identification with sufficient precision speed. A novel method adaptive mutation dipper throated optimization (AMDTO) for feature selection random forest (RF) classifier presented paper. In addition, we propose new dataset balancing, referred optimized hashing SMOTE, boost performance model. Data different weather conditions circumstances were collected augmented before training proposed The effectiveness shown experiments classifying accurately. Eleven methods, WOA, GWO, PSO, three machine learning classifiers included conducted measure superiority method. method, AMDTO+RF, achieved pothole classification accuracy (99.795%), which outperforms by other approaches, WOA+RF 97.5%, GWO+RF 98.6%, PSO+RF 98.1%, transfer AlexNet 86.8%, VGG-19 87.3%, GoogLeNet 90.4%, ResNet-50 93.8%. an in-depth statistical analysis performed recorded results study significance stability

Language: Английский

---

Metaheuristic Optimization Through Deep Learning Classification of燙OVID-19 in Chest X-Ray Images

Computers, materials & continua/Computers, materials & continua (Print), Journal Year: 2022, Volume and Issue: 73(2), P. 4193 - 4210

Published: Jan. 1, 2022

As corona virus disease (COVID-19) is still an ongoing global outbreak, countries around the world continue to take precautions and measures control spread of pandemic. Because excessive number infected patients resulting deficiency testing kits in hospitals, a rapid, reliable, automatic detection COVID-19 extreme need curb infections. By analyzing chest X-ray images, novel metaheuristic approach proposed based on hybrid dipper throated particle swarm optimizers. The lung region was segmented from original images augmented using various transformation operations. Furthermore, were fed into VGG19 deep network for feature extraction. On other hand, selection method select most significant features that can boost classification results. Finally, selected input optimized neural detection. optimizer. experimental results showed achieved 99.88% accuracy, outperforming existing models. In addition, statistical analysis performed study performance stability confirm effectiveness superiority approach.

Language: Английский

---

Dipper Throated Optimization Algorithm for Unconstrained Function and Feature Selection

Computers, materials & continua/Computers, materials & continua (Print), Journal Year: 2022, Volume and Issue: 72(1), P. 1465 - 1481

Published: Jan. 1, 2022

Dipper throated optimization (DTO) algorithm is a novel with very efficient metaheuristic inspired by the dipper bird. DTO has its unique hunting technique performing rapid bowing movements. To show efficiency of proposed algorithm, tested and compared to algorithms Particle Swarm Optimization (PSO), Whale Algorithm (WOA), Grey Wolf Optimizer (GWO), Genetic (GA) based on seven unimodal benchmark functions. Then, ANOVA Wilcoxon rank-sum tests are performed confirm effectiveness other techniques. Additionally, demonstrate algorithm's suitability for solving complex real-world issues, used solve feature selection problem. The strategy using DTOs as evaluated commonly data sets from University California at Irvine (UCI) repository. findings indicate that outperforms all in addressing demonstrating capabilities situations.

Language: Английский

---

Improved Prediction of Metamaterial Antenna Bandwidth Using Adaptive Optimization of LSTM

Computers, materials & continua/Computers, materials & continua (Print), Journal Year: 2022, Volume and Issue: 73(1), P. 865 - 881

Published: Jan. 1, 2022

The design of an antenna requires a careful selection its parameters to retain the desired performance. However, this task is time-consuming when traditional approaches are employed, which represents significant challenge. On other hand, machine learning presents effective solution challenge through set regression models that can robustly assist designers find out best achieve intended In paper, we propose novel approach for accurately predicting bandwidth metamaterial antenna. proposed based on employing recently emerged guided whale optimization algorithm using adaptive particle swarm optimize long-short-term memory (LSTM) deep network. This optimized network used retrieve given features. addition, superiority examined in terms comparison with multilayer perceptron (ML), K-nearest neighbors (K-NN), and basic LSTM several evaluation criteria such as root mean square error (RMSE), absolute (MAE), bias (MBE). Experimental results show could RMSE (0.003018), MAE (0.001871), MBE (0.000205). These values better than those competing models.

Language: Английский

---

Hourly predictions of direct normal irradiation using an innovative hybrid LSTM model for concentrating solar power projects in hyper-arid regions

Energy Reports, Journal Year: 2022, Volume and Issue: 8, P. 15548 - 15562

Published: Nov. 1, 2022

Although solar energy harnessing capacity varies considerably based on the employed technology and meteorological conditions, accurate direct normal irradiation (DNI) prediction remains crucial for better planning management of concentrating power systems. This work develops hybrid Long Short-Term Memory (LSTM) models assessing hourly DNI using datasets that include relative humidity, air temperature, global irradiation. The study proposes a unique model, combining balance-dynamic sine–cosine (BDSCA) algorithm with an LSTM predictor. Combining optimizers predictors, such are rarely developed to estimate DNI, especially in smaller intervals. Therefore, various commonly adopted algorithms relevant studies have been considered references evaluating new algorithm. results show errors proposed do not exceed 2.07%, minimum correlation coefficient 0.99. In addition, dimensionality inputs was reduced from four variables two most cost-effective prediction. these suggested reliable estimating arid desert areas Algeria other locations similar climatic features.

Language: Английский

---