International Journal of Enterprise Network Management (12 papers in press)
Investigation on effect of geometric, material and load parameters on strength of composites with cutouts
by Anoop Kumar S, Rajesh R
Abstract: Composite applications require presence of multiple holes for mechanical fasteners or cutouts in laminates. Holes and cutouts help in lightening composite laminates used as aerospace structures. Unlike isotropic materials, composite materials experience change in stress values due to different parameters such as geometric, material and loading parameters. The present study is devoted to primarily determine whether geometric, material or loading parameters have dominant influence on strength of composite laminates. Geometric parameters such as cutout shape, size, orientation, proximity (in case of multiple holes) as well as edge interaction are taken into consideration. Among material parameters, fiber material, fiber orientation and stacking sequence are varied. Numerical study using ABAQUS CAE software is employed for the analyses. Experiment on Open Hole Tension specimen as per ASTM standards is also conducted to validate the numerical model. Results reveal that geometric parameters have much significant influence on stress concentration factor and thereby the strength of composite laminates, when compared to material parameters. Among the geometric parameters, edge interaction is the most critical factor affecting the stress concentration. An elliptical cutout is seen to have comparatively more adverse effect on strength of laminate, when compared with other cutout shapes. Further, effect of load parameters in-plane tension, compression and shear, is also studied. However, no effect was evidenced in stress concentration factor due to load parameters.
Keywords: Stress Concentration Factor; Composite Laminate; Edge Interaction; Open Hole Tension; Geometrical parameters; Material parameters; Load parameters.
Experimental Study on the Influence of Fiber Surface Treatments and Coconut shell Powder addition on the Compressive strength, Hardness and Tribological properties of Sisal fiber-Natural Rubber Composites.
by Gopakumar R Nair, Rajesh R
Abstract: In the contemporary age of global imbalance and disasters, toxic wastes and their waste disposal is a major concern for the humankind. Development and utilization of natural materials as a viable replacement for synthetic materials in product design and development is the only long-term solution for this problem. This work aims to design and develop an elastomer composite material using 100% natural materials only- natural rubber composite reinforced with sisal fibers with improved tribological and compressive properties. Since rubber products are widely applied in industries and automotive parts, the developed material can find its place in elastomer components. A total of six composites are made with sisal natural fibers subjected to various surface modifications as reinforcement and a 10%w/w coconut shell powder as fillers in a natural rubber matrix. Sisal fibers used are raw fiber, Alkalized fibers, rubber pre-impregnated raw fibers and Rubber Pre-impregnated Alkalized fiber. The specimens are tested for wear resistance, compressive strength and hardness. It is found that sisal is a good reinforcement for enhancing the above properties of natural rubber. Maximum wear resistance exhibited by the Alkalized pre-impregnated Sisal-Rubber composite, followed by raw pre-impregnated-Coconut shell powder-Rubber Composite. The hardness of the composites Raw Sisal-Rubber and Raw Pre-impregnated improved by 228% than pure rubber (25 Shore A). Compressive strengths also improved to a notable level. The Composite raw Sisal-Coconut Shell powder-Rubber has the best compressive strength.
Keywords: Natural Rubber; Sisal fibers; Elastomer Composite; Hardness; Tribology; Compressive Strength; Bushes.
Combined Economic and Emission Dispatch using Whale Optimization Algorithm (WOA)
by Faseela C K, H. Vennila
Abstract: This paper highlight the use of latest Whale Optimization meta heuristic algorithm for solving economic dispatch problem efficiently. This is used to solve the combined economic and emission dispatch problems for standard 3 generators system and 30 bus IEEE system. The Whale Optimization algorithm was found to provide optimum results with easy convergence in comparison with other algorithms like PSO algorithm. Fuel cost and emission costs are combined to derive better result for economic dispatch.
For checking the effectiveness of the algorithm, the results obtained using the same are compared with the results of Particle Swarm Optimization (PSO) and analyzed the same against minimum generation cost and easy convergence. The results are found to be excellent for the systems considered.
Keywords: Particle Swan Optimization (PSO); Whale Optimization Algorithm (WOA); Economic and Emission Dispatch (EED); optimum; solution; fuel cost; emission cost; optimization methodology.
Comparative Study of Machine Learning Techniques for Breast Cancer Identification/Diagnosis
by Ganapathy G, Sivakumaran N, M. Punniyamoorthy, Surendheran R, Srijan Thokala
Abstract: This study is sought to determine closely and analyse the results of the diagnostic procedural steps for accurate decision making. The number of new cases of female breast cancer was 124.9 per 100,000 women per year. The number of deaths was 21.2 per 100,000 women per year. It calls for an urge to increase the awareness of breast cancer and very accurately analyse the causes which may differ in minute variations. Also, the significance of classifying the cancer patients based on their tumour types and levels is growing. This is why the application of computation techniques are widely increasing to support the Diagnostic results. In this work, we present the application of several Machine Learning techniques and models like Neural Networking, SVM in machine learning to quantify the classifications. The techniques that are most reliable, accurate and robust are emphasised. It gives a plethora of explorations into the research field for developing predictive models. To achieve higher reliability on the data, we present the comparison of various Machine Learning techniques which has not been done before on a dataset that is available on the website Kaggle with the goal to model the attributes of cancer risk patients to determine the outcome.
Keywords: Breast Cancer; Machine Learning; Neural Network; FNA; SVM; Kernel; KNN; Naïve Bayes.
Automatic Detection and Classification of Brain Tumors using k-Means Clustering with Classifiers
by Hema Rajini Narayanan
Abstract: A brain tumor detection and classification system has been designed and developed. This work presents a new approach to the automated detection and classification of astrocytoma, medulloblastoma, glioma, glioblastoma multiforme and craniopharyngioma type of brain tumors based on k-means clustering and texture features, which separate brain tumor from healthy tissues in magnetic resonance images. The magnetic resonance feature image used for the tumor detection consists of T2-weighted magnetic resonance images for each axial slice through the head. The application of the proposed method for tracking tumour is demon
Keywords: Magnetic resonance imaging; k-Means clustering; Segmentation; Gray level co-occurrence matrix; Tumor.
Performance and Emission Characteristics of a DI Diesel Engine using Diestrol Blends and diesel as fuel
by BIBIN CHIDAMBARANATHAN, Seeni Kannan Pauldurai, Devan Ponnusamy Kumarasami, Rajesh Rajamoni
Abstract: Biofuels, namely, biodiesel and ethanol produced from renewable energy sources are used as fuels in the blended form along with diesel to investigate the performance and emission characteristics of a DI diesel engine. Diestrol blend consists of diesel, biodiesel/methyl ester and ethanol. In diestrol blends, ethanol percentage is steadily elevated with an incremental factor of 5% culminating into three blends with a maximum percentage of 15% by volume and named as EB5, EB10 and EB15 respectively. A comprehensive analysis of engine performance characteristics such as brake thermal efficiency, brake specific fuel consumption, exhaust gas temperature and emission characteristics such as Carbon monoxide, Carbon dioxide, unburned hydrocarbon, oxides of nitrogen and Smoke opacity were carried out. From the above investigation, it was found that brake thermal efficiency increased by 3%, 5% and oxides of nitrogen emission decreased by 23%, 24.5% when compared to diesel and B20 respectively.
Keywords: diesel engines; diestrol; ethanol; emissions; methyl esters; punnai oil; performance; ternary blends.
A simulated annealing for the cell formation problem with ratio level data
by Kamalakannan Ramalingam
Abstract: In this paper, the cell formation problem is considered with ratio level data with an objective of minimizing the cell load variation. The attempt has been made to propose a Simulated Annealing [SA] based on the perturbation scheme as Random Insertion Perturbation Scheme [RIPS]. The ratio level data is distinguished by utilizing the workload information gathered from process times, production quantity of parts and also from the capacity of the machines. A Modified Grouping Efficiency [MGE] is used to measure the performance of the system. From the results it is observed that the simulated annealing produces the solution does not differ significantly from the optimal solutions for the benchmark problems. The algorithms which we have chosen the benchmark problems are K-means, Modified ART1 and Genetic Algorithm taken from the literature.
Keywords: Simulated Annealing; Cell Formation Problem; Ratio Level Data.
Prediction of carotid atherosclerosis in patients with impaired glucose tolerance A performance analysis of machine learning techniques
by Maruthamuthu A, M. Punniyamoorthy, Swetha Manasa Paluru, Sindhura Tammuluri
Abstract: The focus of this paper is to examine factors associated with carotid atherosclerosis in patients with impaired glucose tolerance (IGT), and to predict the rapid progression of carotid intima-media thickness (IMT). The proposed machine learning methods performed well and accurately predicted the progression of carotid IMT. The linear support vector machine, non-linear support vector machine with a radial basis kernel function, multilayer perceptron (MLP), and Naive Bayes method were employed. A comparison of these methods was conducted using the Brier score, and the accuracy was tested using a confusion matrix.
Keywords: Multilayer Perceptron; Support Vector Machine; Radial Basis Kernel function; Impaired Glucose Tolerance; Carotid Atherosclerosis; Naive Bayesian model; Brier Score.
A Software Reliability Prediction Model based on Fuzzy k-Nearest Neighbor with Glowworm Swarm Optimization Algorithm
by Shailee Lohmor, B.B. Sagar
Abstract: The software reliability prediction aims to estimate the futuristic occurrences of failures in the maintenance and replacement processes of a software. K-Nearest Neighbor (KNN) is a kernel-based learning method that has been successfully implemented for regression problems. However, they have not been widely explored for use in reliability applications. This paper employs a FKNN (Fuzzy k-Nearest Neighbor)-based model for predicting the software reliability so as to capture the inner correlation between software failure time data and the nearest m failure time data. In order to determine the effectiveness of KNN in predicting the time-to-failure for software products, the paper presents comparative analysis with the current models.
Additionally, to optimize the trend of predictive accuracy of KNN as m varies we used the algorithm titled Glowworm Swarm Optimization, whereas the reasonable value range of m is achieved through paired cross fold validation-tests in 4 frequently used failure datasets from NASA software projects.
Keywords: Fuzzy Membership function; Glowworm Swarm Optimization (GSO); K-Nearest Neighbor(KNN); Least Square Estimation (LSE); Maximum Likelihood Estimation (MLE); Software Reliability Prediction (SRE)).
Special Issue on: BDSCC-2018 Big Data Innovation for Sustainable Intelligent Computing
Convergence of Partial Differential Equation Using Fuzzy Linear Parabolic Derivatives
by Shanthi Devi Palanisamy, Viswanathan Ramasamy
Abstract: Discovering solution for Partial Differential Equations (PDEs) is a very difficult. The exact solution can identify only in some particular cases. The numerical method for PDEs have been attains a greater significance growing in recent years. In this paper, we consider the convergence of Partial Differential Equation using Fuzzy Linear Parabolic (PDE-FLP) method on a finite domain. The method is based on the construction of PDE where the coefficients and initial conditions are obtained as fuzzy numbers and solved by linear parabolic derivatives. The linear parabolic derivative serves as the basis where Fuzzy form is considered to be solved as a numerical solution for PDEs. Firstly, the PDE form of two independent variables and the fuzzy representation of the two independent variables are derived. Secondly, Fuzzy Linear Parabolic derivative is provided for the convergence to a numerical solution for PDE. Fuzzy linear parabolic derivatives are employed to describe the wide variety of time dependent development. Parabolic derivatives are used in PDE-FLP because the coefficient is the same as the condition for the analytic solution. Finally, numerical results are given, which demonstrates the effectiveness and convergence of the PDE-FLP method. A detailed comparison between the approximate solutions obtained by these methods is discussed. Also, figurative representation to compare between the approximate solutions is also presented.
Keywords: Partial Differential Equation; Fuzzy; Linear Parabolic; Domain; Numerical solution.
Document Similarity Approach using Grammatical Linkages with Graph Databases
by Priya V, Umamaheswari K
Abstract: Document similarity had become essential in many applications such as document retrieval, recommendation systems, plagiarism checker etc. Many similarity evaluation approaches rely on word based document representation, because it is very fast. But these approaches are not accurate when documents with different language and vocabulary are used. When graph representation is used for documents they use some relational knowledge which is not feasible in many applications because of expensive graph operations. In this work a novel approach for document similarity computation which utilizes verbal intent has been developed. This improves the similarity by increasing the number of linkages using verbs between two documents. Graph databases were used for faster performance. The performance of the system is evaluated using various metrics like cosine similarity, jaccard similarity and Dice with different review datasets. The verbal intent based approach has registered promising results based on the links between two documents.
Keywords: Document similarity;Graph Database; Grammatical Linkages;knowledge graph; Text Similarity.
Special Issue on: Sustainable Computing for Enterprise Resource Planning Applications
MINING MASSIVE ONLINE LOCATION BASED SERVICES FROM USER ACTIVITY USING BEST FIRST GRADIENT BOOSTED DISTRIBUTED DECISION TREE
by Venkatesh M., Mohanraj V, Suresh Y
Abstract: User activity is predicted through the frequency in which the online substances in Location-Based Social Networks (LBSN) are produced and used by the consumer. Users are classified by researchers into a number of groups depending upon the level of their functioning. This takes place by creation of services based on their location, so that maximum number of user gets benefited. This work involves Gradient Boosted Distributed Decision Tree (GBDT) which is optimized on the basis of total iterations and shrinkage on using best algorithm. Implementation of the data is done through Hadoop network. A foursquare dataset is created using work, food, travel, park and shop. One of the most commonly used machine learning algorithm is Stochastic Gradient Boosted Decision Trees (GBDT) at present. Its advantages are that it can be interpreted easily with increased precision. However, most of the implementations are costly computationally and needs all training data in the main memory. The node with lowest lower bound is developed through Best First Search (BFS). Its own filing system is provided through Hadoop which is called Hadoop Distributed File System. The algorithm used is K-Nearest Neighbor (KNN) classifier algorithm.
Keywords: User activity; foursquare dataset; Stochastic Gradient Boosted Decision Trees (GBDT); Best-First Search (BFS) and K-Nearest Neighbor (KNN) classifier.