Transactions on Machine Intelligence2821-16935420221201Flood Routing Using the Muskingum-Cunge Method and Genetic Algorithm22223021463310.47176/TMI.2022.222ENSh.SadiDepartment of Civil Engineering, Faculty of Engineering, Shahid Madani University of Azerbaijan, Tabriz, Iran.Journal Article20220714Flooding remains one of the most catastrophic natural hazards, often causing widespread human casualties and extensive economic damage. Nonetheless, the severity of its consequences can be significantly mitigated through precise modeling, thorough analysis, and the implementation of effective flood management strategies. A deep understanding of flood behavior and trends is crucial for improving forecasting accuracy and enabling timely preventive actions in flood-prone areas. When integrated with early warning systems, flood control infrastructures, and coordinated emergency responses, reliable flood forecasting can dramatically reduce the risk to human life and infrastructure. This study adopts a documentary and library-based research methodology to gather and analyze relevant data, with the objective of enhancing the accuracy of flood modeling techniques. Specifically, the study evaluates the effectiveness of the Maskingham-Cunge method in conjunction with genetic algorithms for modeling flood behavior. The integration of these approaches allows for the dynamic adjustment of parameters, replacing static inputs with variable ones to better reflect real-world conditions. Additionally, incorporating one-dimensional kinematic wave theory to compute wave speed improves the precision of output hydrograph estimation. The findings demonstrate that this combined approach significantly enhances the predictive performance of flood models. As a result, it offers a robust tool for informed decision-making in flood management, contributing to more efficient disaster preparedness and risk reduction efforts in vulnerable regions.https://www.tmachineintelligence.ir/article_214633_57c4c062268a7d3da934702b1471c0dd.pdfTransactions on Machine Intelligence2821-16935420221201Adaptive Model Predictive Control(AMPC) for Current Control of Three-Phase Three-Level Inverters23124722201010.47176/TMI.2022.231ENN.Sohrabi VeskeDepartment of Control, Faculty of Electrical Engineering, University of Science and Technology, Tehran, Iran.M.R.NowroziDepartment of Power, Faculty of Electrical Engineering, University of Science and Technology, Tehran, Iran.M.SadeghiNadaja Self-Sufficiency Research and Jihad Organization, Tehran, Iran.Journal Article20220723Model Predictive Control (MPC) is widely used in power electronics due to its ability to handle multi-variable constraints and optimize control actions in real time. However, its performance heavily relies on an accurate system model, making it sensitive to parameter variations and model mismatches. To address this limitation, this paper proposes a novel Model-Predictive Adaptive Control (MPAC) method for current control in three-phase, three-level voltage source inverters (VSI). The proposed MPAC approach integrates the Recursive Least Squares (RLS) algorithm for online system parameter estimation, eliminating the need for a predefined model and allowing real-time adaptation to changes in system dynamics. The proposed MPAC method is implemented and tested in a MATLAB/Simulink environment, where its performance is analyzed under various operating conditions. Simulation results demonstrate that MPAC achieves fast and precise current tracking, robust disturbance rejection, and significantly reduced harmonic distortion compared to conventional MPC methods. Furthermore, MPAC exhibits superior robustness against system parameter uncertainties, ensuring stable operation even in the presence of load variations and model inaccuracies. By improving adaptability and robustness, the proposed MPAC approach has significant potential for application in a wide range of power electronic systems, including motor drives, renewable energy conversion systems, and grid-connected converters. The findings of this study highlight the advantages of integrating adaptive estimation techniques into predictive control strategies, paving the way for more efficient and resilient power electronic control systems.https://www.tmachineintelligence.ir/article_222010_2a351a7b2214755bdef6488b07e95676.pdfTransactions on Machine Intelligence2821-16935420221201Image Steganography of MRI Using 2D Wavelet Transform24825522201410.47176/TMI.2022.248ENM.SedaghatianDepartment of Biomedical Engineering, Faculty of Electrical and Computer Engineering, Babol Noshirvani University of Technology, Babol, IranM.R.Karami-MollaeiAssociate Professor, Department of Biomedical Engineering, Babol Noshirvani University of Technology, Babol, IranJournal Article20220529With the increasing use of the internet and the growing volume of exchanged data, it has become possible to store and send medical information about patients. On the other hand, the use of these methods poses significant risks to the privacy of patient information. Therefore, the use of information hiding techniques, in such a way that the information is not easily detectable or alterable, has become essential. One such method is steganography. In this paper, magnetic resonance imaging (MRI) images are used as hosts for steganographic operations. In the proposed method, all wavelet coefficients of the secret image are embedded in the coefficients of the host image using the suggested α relation. In the experiments conducted, the performance of the wavelet-based steganography in embedding the secret image coefficients into the approximation, horizontal, vertical, and diagonal detail coefficients of the host image at four wavelet decomposition levels are compared. The comparison is performed using the Mean Squared Error (MSE) and Signal to Noise Ratio (SNR) metrics. The results show that steganography at the fourth level of wavelet decomposition, with embedding in the diagonal details, provides the best outcome in terms of both visual and statistical quality.https://www.tmachineintelligence.ir/article_222014_c8d1b34afed7d56eb832c91f60b908a0.pdfTransactions on Machine Intelligence2821-16935420221201Detection of Benign and Malignant Breast Cancer Using Data Shuffling Ensemble Method25626422201910.47176/TMI.2022.256ENM.R.Rahmani DahaghiM.Sc. Student, Department of Electrical Engineering, Payam Noor Higher Education Institute, Golpayegan, IranS. M.HashemiFaculty Member, Department of Electrical Engineering, Payam Noor Higher Education Institute, Golpayegan, IranZ.AssarzadehFaculty Member, Payam Noor Higher Education Institute, Golpayegan, IranJournal Article20220505Breast cancer remains one of the most serious health concerns affecting women worldwide. Early detection of malignant tumors significantly improves survival rates and enhances patients' quality of life. As a result, the development of accurate and efficient diagnostic systems for breast cancer is of critical importance. Artificial neural networks (ANNs) have been widely applied in medical diagnosis, particularly in data classification tasks. The backpropagation algorithm is a commonly used technique for training neural networks; however, it suffers from certain limitations, such as slow convergence, susceptibility to local minima, and sensitivity to initial weight selection. To address these challenges, this study employs a data-shuffling ensemble method to improve classification accuracy. The effectiveness of both approaches is assessed in distinguishing between benign and malignant breast tumors using a well-established dataset. Experimental results indicate that the neural network utilizing the data-shuffling ensemble method achieves an accuracy of 99.3%, outperforming the backpropagation algorithm. These findings highlight the potential of ensemble learning techniques in enhancing diagnostic accuracy and reliability in medical applications. The study contributes to the ongoing advancements in artificial intelligence-based diagnostic tools, emphasizing the importance of robust classification techniques in improving breast cancer detection and patient outcomes.https://www.tmachineintelligence.ir/article_222019_f310141c80df38871369f363e2bb5b2b.pdfTransactions on Machine Intelligence2821-16935420221201Optimization of Capacitor Placement and Sizing in Radial Distribution Systems Using Enhanced Harmony Search Algorithm, PSO, and TLBO with Power Loss Index (PLI)26527622202210.47176/TMI.2022.265ENS.AzadDepartment of Electrical Engineering, Shahid Beheshti University, Tehran, IranM. M.AmiriDepartment of Electrical Engineering, Shahid Beheshti University, Tehran, IranM.YarmohammadiDepartment of Electrical Engineering, Iran University of Science and Technology, Tehran, IranM.Setayesh Nazarssistant Professor, Department of Electrical Engineering, Shahid Beheshti University, Tehran, IranJournal Article20220720Given the significant contribution of the distribution network to the total system losses, it is essential to implement fundamental measures to reduce losses in the distribution network. The placement and optimal sizing of parallel capacitors to reduce power losses and improve voltage profiles is a common issue in power system design and control, which has been extensively studied. In this paper, the Enhanced Harmony Search Algorithm (IHA) is proposed to determine the optimal capacitor placement and sizing, utilizing the Power Loss Index (PLI) in radial distribution systems. The procedure begins by using the PLI to select the buses for optimal capacitor installation, followed by the development of the IHA to determine the optimal location and size of capacitors using PLI. The proposed method has been applied to the IEEE 33-bus and 69-bus radial distribution systems. The results of this algorithm are compared with those of the Particle Swarm Optimization (PSO) algorithm and the Teaching-Learning-Based Optimization (TLBO) algorithm to demonstrate its superiority. Additionally, the IHA is tested under different loading conditions, and the impact of this method on the results has been proven.https://www.tmachineintelligence.ir/article_222022_3af4de2bc726571a4b6f2d9f09e048e1.pdfTransactions on Machine Intelligence2821-16935420221201Improving Accuracy in Breast Cancer Diagnosis Using Data Mining Techniques27728522202310.47176/TMI.2022.277ENF.FatahiDepartment of Computer Engineering, Faculty of Technical and Engineering, Kermanshah Branch, Islamic Azad University, Kermanshah, Iran.Journal Article20220804This study introduces a novel, integrated approach for breast cancer diagnosis, addressing one of the most critical challenges in medical sciences: the lack of timely and precise detection. Breast cancer remains a leading cause of mortality worldwide, and early diagnosis plays a pivotal role in improving survival rates. Currently, diagnostic practices heavily rely on physicians' expertise, supported by complex and time-consuming laboratory tests, which are prone to human error and often lead to delays in treatment. To overcome these limitations, this research proposes a comprehensive methodology that combines principal component analysis (PCA) for dimensionality reduction, decision trees for feature selection, and artificial neural networks (ANNs) for classification and prediction. By integrating these techniques, the proposed system optimizes the use of database features, offering an adaptable, efficient, and accurate solution for breast cancer detection. The results demonstrate that this method achieves superior diagnostic accuracy compared to conventional techniques and existing artificial intelligence-based methods referenced in related studies. Furthermore, the system significantly reduces diagnostic costs and time without compromising performance. This research highlights the potential of combining machine learning and data mining techniques to enhance diagnostic precision, providing researchers and clinicians with an effective tool for improving early detection, treatment planning, and patient outcomes.https://www.tmachineintelligence.ir/article_222023_d48146e02f7ac54c44f0a2cf747156ee.pdf