Papers by Keyword: Clustering

Abstract: Predicting the deformation behavior of rolled and extruded light metal alloys is a challenging task. Due to the high cost of experimental analysis, finite element simulations are often required. A variety of material models at different scales are available for practical use. In this work, the viscoplastic self-consistent (VPSC) approach is employed to consider microstructural effects. These can be incorporated by using measured crystal sizes and orientations - called texture - of the alloy under consideration. For each integration point in the FE mesh, a corresponding texture is assigned and individually deformed in LS-Dyna^®, where VPSC is implemented as a user-defined material model - referred to as FE-VPSC. This study focuses on preprocessing of texture data as well as their compression for accurate and faster FE simulations. For verifying the simulations, a comparison with digital image correlation (DIC) of experimental puncture tests was conducted.

Abstract: In virtual process chains for discontinuous fiber-reinforced polymers, clustering of fiber orientation tensors reduces the number of macroscopic material cards required for downstream structural and warpage simulations. However, it remains unclear whether including the additional information provided by the fourth-order fiber orientation tensor improves clustering quality. This study investigates the influence of second-order vs. fourth-order informed clustering on clustering outcomes and the resulting orientation-averaged mechanical properties. Using parameterizations based on harmonic decomposition, rotation-invariant clustering is performed in both the second-order and fourth-order parameter spaces. Results from injection molding simulation data indicate that the level of tensorial information has limited effect when the fourth-order tensor is computed via a closure approximation, as the deviatoric parameters are nonlinearly dependent on the second-order parameters. In contrast, the choice of clustering algorithm -- KMeans vs. Birch -- has a more pronounced influence on cluster shapes and allocations. Furthermore, we demonstrate that clustering affects orientation-averaged stiffness properties, with deviations most pronounced near cluster boundaries and rarely occurring tensor shapes.

Abstract: Data clustering is a critical data mining technique for grouping similar objects and differentiating dissimilar ones. While advancements in machine learning, statistical, and metaheuristic methods have addressed some challenges, issues like accuracy, efficiency, and scalability persist. Building on the History-Based Artificial Bee Colony (HD-ABC) algorithm, this paper introduces the Enhanced History-Driven Artificial Bee Colony (EHD-ABC) algorithm. Refining the historical memory mechanism and optimization process, the proposed algorithm achieves improved clustering accuracy, reduced computational complexity, and enhanced efficiency. Experimental results on artificial and real-world datasets demonstrate EHD-ABC's superiority over existing methods in clustering quality and error reduction, such as HD-ABC and K-means.

Abstract: This research examines the use of machine learning to group a collection of data related to the tax compliance of manufacturing firms in the Greater Jakarta area of Indonesia. The data set was obtained through a survey and was able to collect data from 209 respondents who represented the finance department of the companies. The k-means algorithm is applied to develop machine learning. The clustering aims at dividing the data set on the basis of similarity into three clusters. The result showed that the machine learning model was able to cluster the data into three groups. An evaluation was presented by comparing the clustering result with a classification result based on the average survey score that has been studied previously. The evaluation shows a small correlation between the clusters and the average survey score. Compliance of tax payers is a complex system and cannot be merely indicated based in the survey score. The clustering technique demonstrated its usefulness in uncovering intriguing patterns, distributions, and the fundamental structure of the data.

Abstract: The ultra-rapid mobility of cellular device-to-device (D2D) users degrades network connections and enhances renewal interconnection issues, eventually lowering the overall throughput performance. Moreover, a limited cellular infrastructure in rural areas hinders smooth network communication, leading to massive network outages and disconnections. Therefore, to maintain high data reception, the users require a safe moving zone, i.e., an area where the performance degradation is still tolerable. However, determining the safe moving zone of D2D users is categorized as a nonlinear problem that magnifies solutions into combinatorial possibilities. Based on that challenging issue, this study proposes a binary message-passing method to cluster safe moving zones by finding the minimum pathloss. The messagepassing technique splits the centralized D2D issues into local cases involving only adjacent D2D users. By enabling a distributive manner, the proposed algorithm determines the moving zone with low computational complexity. The evaluation results show that the proposed algorithm provides a higher homogeneity and completeness value than conventional clustering techniques, indicating a clustering model’s high fitness. The simulation result demonstrates that the proposed algorithm only requires 29 iterations with a high homogeneity of 0.727 and completeness of 0.752, Moreover, the proposed algorithm outperforms D2D conventional techniques regarding pathloss degradation. For 25 numbers of the dataset, the proposed algorithm only experiences 59.52% data loss, which provides more efficient performance than K-means and color graphs that correspond to 60.94% and 61.97%.

Abstract: This work presents results of a study of the influence assessment based on the recorded acoustic emission (AE) parameters of thermo-oxidative aging conditions on the destruction process of a polymer composite material (PCM). The objects of the study were specimens cut from a fiberglass reinforced plastic (FGRP) plate. The plate was made by vacuum infusion technique using Derakane 411-350 resin and 9 layers of St-62004 glass fabric. Specimens aging were done by holding in a muffle furnace for 96 hours at temperatures of 60, 100, 120, and 200 ° C. Mechanical test was the three-point static bending method. For the AE recording was used a hardware-software complex developed at KnASU. The AE signal Fourier spectra were two-stage clustered with the self-organizing Kohonen map according to the technique previously developed and tested by the authors. The types of the PCM structure damage were characterized by the obtained clusters centroids. The fracture process kinetics is described depending on the conditions of thermo-oxidative aging and based on the accumulation of clusters during mechanical tests. The negative influence of high temperatures on the polymer matrix degradation, leading to a decrease in the ability of the matrix to effectively distribute internal stresses over the PCM volume due to the adhesion corruption with the reinforcing material, has been established.

Abstract: In recent years, tremendous advances in communication technologies coupled with the advent of the Internet of Things (IoT) have led to the emergence of the Big Data phenomenon. Big Data is one of the big IT challenges of the current decade. The amount of data produced is constantly increasing and makes it more and more difficult to process. Managing these masses of data requires the use of new data management systems with efficient access methods. Considered as one of the main sources of Big Data, wireless sensors used in networks offer a credible solution to the problem of Big Data management, especially its collection. Several solutions for Big Data collection based on large-scale wireless sensor networks (LS-WSN) are proposed, taking into account the nature of the applications. The hierarchical architecture is the one used for the deployment of these applications. In such an architecture, relay sensors play an important role in finding the balance of the network and maximizing its lifetime. In most LS-WSN applications, once deployed, the LS-WSN does not provide a mechanism to evaluate and improve the positions of the initially deployed relay sensors. This paper proposes, based on the growth model of physarum polycephalum and its ability to prune unnecessary links and retain only those deemed useful for food routing, a mechanism for evaluating and optimizing relay sensors in LS-WSNs. Simulation results indicate that the proposed approach significantly improves the network lifetime compared to the initial deployment and that can be a useful approach for LS-WSNs dedicated to Big Data collection. The effectiveness of the proposed technique is demonstrated by experimental results in terms of connectivity and network lifetime.

Abstract: Nowadays Wireless Sensor Networks (WSNs) are a very promising device to monitor environmental events, which are used in many application areas such as agriculture, environmental monitoring, air-water pollution, greenhouse, health monitoring, structural monitoring etc. Also, WSN needs effective and efficient way of collecting data from the area over which it is deployed, since the network consists of battery-operated nodes with limited energy. Improving the lifetime of WSN for continuous monitoring applications are very essential because of simple deployment, low installation cost, lack of cabling, and high mobility. The better energy conservation and reliable data transmission would be achieved in WSN by adopting efficient routing protocols for better and reliable communication. Also, security in data transmission in WSN is still a challenging task. Hence, a better way of encryption has to be computed to promote a reliable transmission between the nodes. In this paper discusses about different routing protocols which support energy efficiency and security techniques in WSN.

Abstract: Applications of Flexible AC Transmission Systems (FACTS) devices for enhancement of Available Transfer Capability (ATC) is gaining attention due to economic and technical limits of the conventional methods involving physical network expansions. FACTS allocation which is sine-qua-non to its performance is a major problem and it is being addressed in recent time with heuristic algorithms. Brain Storm Optimization Algorithms (BSOA) is a new heuristic and predicting optimization algorithms which revolutionizes human brainstorming process. BSOA is engaged for the optimum setting of FACTS devices for enhancement of ATC of a deregulated electrical power system network in this study. ATC enhancement, bus voltage deviation minimization and real power loss regulation are formulated into multi-objective problems for FACTS allocation purposes. Thyristor Controlled Series Capacitor (TCSC) is considered for simulation and analyses because of its fitness for active power control among other usefulness. ATC values are obtained for both normal and N-1-line outage contingency cases and these values are enhanced for different bilateral and multilateral power transactions. IEEE 30 Bus system is used for demonstration of the effectiveness of this approach in a Matlab software environment. Obtained enhanced ATC values for different transactions during normal evaluation cases are then compared with enhanced ATC values obtained with Particle Swarm Optimization (PSO) set TCSC technique under same trading. BSO behaved much like PSO throughout the achievements of other set objectives but performed better in ATC enhancement with 27.12 MW and 5.24 MW increase above enhanced ATC values achieved by the latter. The comparative of set objectives values relative to that obtained with PSO methods depict suitability and advantages of BSOA technique.

Abstract: In Nigeria, a crucial responsibility of the executive arms of the government is to submit annual budgetary allocations to the national assembly for approval. Due to the diversity and complexity of the budget, the national assembly is mandated to carry out its constitutional duty of scrutinizing the budget to discover irregularity or anomaly, make recommendations, or substantial modification upon what it received. This is very challenging, particularly in Nigeria where there are many different ethnicities and regional, to ensure inclusiveness, the national assembly must carry out its constitutional duty diligently and carefully without fear or favor that often has unintended consequences. This might not be very easy to accomplish within a short period. Thus, this research aims at detecting an anomaly in the budget that will ease the legislative duty thereby facilitating the process of appropriation. The concept of Clustering for Machine learning technique was used for the detection of an anomaly, where the detected ones are noted and indicated for critical examination.