nach oben

2019 | Buch

Kapitel lesen Erstes Kapitel lesen

Decision Science in Action

Theory and Applications of Modern Decision Analytic Optimisation

herausgegeben von: Kusum Deep, Dr. Madhu Jain, Prof. Said Salhi

Verlag: Springer Singapore

Buchreihe : Asset Analytics

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten

Über dieses Buch

This book provides essential insights into a range of newly developed numerical optimization techniques with a view to solving real-world problems. Many of these problems can be modeled as nonlinear optimization problems, but due to their complex nature, it is not always possible to solve them using conventional optimization theory. Accordingly, the book discusses the design and applications of non-conventional numerical optimization techniques, including the design of benchmark functions and the implementation of these techniques to solve real-world optimization problems.

The book’s twenty chapters examine various interesting research topics in this area, including: Pi fraction-based optimization of the Pantoja–Bretones–Martin (PBM) antenna benchmarks; benchmark function generators for single-objective robust optimization algorithms; convergence of gravitational search algorithms on linear and quadratic functions; and an algorithm for the multi-variant evolutionary synthesis of nonlinear models with real-valued chromosomes.

Delivering on its promise to explore real-world scenarios, the book also addresses the seismic analysis of a multi-story building with optimized damper properties; the application of constrained spider monkey optimization to solve portfolio optimization problems; the effect of upper body motion on a bipedal robot’s stability; an ant colony algorithm for routing alternate-fuel vehicles in multi-depot vehicle routing problems; enhanced fractal dimension-based feature extraction for thermal face recognition; and an artificial bee colony-based hyper-heuristic for the single machine order acceptance and scheduling problem.

The book will benefit not only researchers, but also organizations active in such varied fields as Aerospace, Automotive, Biotechnology, Consumer Packaged Goods, Electronics, Finance, Business & Banking, Oil, Gas & Geosciences, and Pharma, to name a few.

Inhaltsverzeichnis

Frontmatter

π Fraction-Based Optimization of the PBM Antenna Benchmarks

Abstract

Real-world optimization problems often require an external “modeling engine” to compute fitnesses, and these programs often have much longer runtimes than evaluating fitnesses solely with built-in compiler routines. Using a stochastic optimizer on real-world problems can be quite challenging because every run returns a different “best” fitness. This issue is addressed by making many runs, often hundreds, possibly even thousands, in order to generate meaningful statistics, but doing so can be prohibitive with external modeling. And even then the statistical nature of the results may obscure true global extrema. Additionally, real-world problems do not come with well-defined, clearly appropriate objective functions (at least most of the time). The practitioner must define an appropriate function, which in itself can be a daunting task made more difficult using a stochastic optimizer. π fractions mitigate these issues by introducing pseudorandomness in an otherwise truly random metaheuristic, for example, genetic algorithm. This paper illustrates the utility of π fractions by using them in two different optimizers, one deterministic and the other probabilistic. These optimizers are applied with quite good results to the PBM antenna benchmarks, a set of difficult real-world engineering problems, thereby demonstrating the utility of π fractions in all types of optimizers.

Richard A. Formato

Benchmark Function Generators for Single-Objective Robust Optimisation Algorithms

Abstract

Test problems are considered essential when designing optimisation algorithms. The two main conflicting characteristics of a proper test function are simplicity and complexity. The former feature is to allow analysing the behaviour of algorithms, whereas the latter is to mimic real-world problems. Despite the importance of the test functions, however, there are currently neither empirical studies on the suitability of the existing test functions nor benchmark generator to generate them in the field of robust optimisation. This motivates our attempts to analyse the current test functions and propose a new set of benchmark generators to generate test functions with different levels of difficulty. To examine the proposed test functions, robust particle swarm optimisation and robust genetic algorithms are used. The results and analysis first reveal the drawbacks of the current test functions as simplicity, low dimensionality, symmetric search space and lack of scalability. The results then demonstrate the merits of the proposed benchmark generators in alleviating these drawbacks and providing challenging test beds for robust optimisation algorithms.

Seyedali Mirjalili, Andrew Lewis

Convergence of Gravitational Search Algorithm on Linear and Quadratic Functions

Abstract

Convergence characteristic of any optimization algorithm is a very important aspect. Several studies have been performed to discuss the convergence of non-deterministic optimization algorithms. In this article, the convergence of gravitational search algorithm (GSA) is discussed over linear and quadratic functions. A theoretical proof of convergence for GSA is provided for linear and quadratic functions. The article ensures the convergence of GSA over linear and quadratic functions.

Anupam Yadav, Anita, Joong Hoon Kim

An Algorithm of Multivariant Evolutionary Synthesis of Nonlinear Models with Real-Valued Chromosomes

Abstract

We propose a new multivariant evolutionary algorithm for solving the problem of construction of nonlinear models (mathematical expressions, functions, algorithms, and programs) based on the given experimental data, sets of variables, basic functions, and operations. The proposed algorithm of multivariant evolutionary synthesis of nonlinear models includes a linear representation of a chromosome by real variables, simple operations in decoding of a genotype into a phenotype for interpreting a chromosome as a sequence of instructions, and also a multivariant method for presenting a set of models (expressions) using a single chromosome. We compare the proposed algorithm with the standard genetic programming algorithm (GP) and the Cartesian genetic programming (CGP) one. We show that the proposed algorithm exceeds the GP and CGP algorithms both in the time required for search for a solution (more than by an order of magnitude in the most cases) and in the probability of finding a given model.

Oleg Monakhov, Emilia Monakhova

An Artificial Bee Colony Based Hyper-heuristic for the Single Machine Order Acceptance and Scheduling Problem

Abstract

This paper presents an artificial bee colony based hyper-heuristic for solving the order acceptance and scheduling (OAS) problem in a single machine environment. The OAS problem gives the flexibility to accept or reject an order where the systems have limited production capacity and on-time delivery constraints. The OAS problem, which is a typical \(\mathcal {NP}\)-hard problem, becomes more complex when a sequence-dependent setup time is incurred between two consecutive orders. Solving an \(\mathcal {NP}\)-hard problem through exact approaches is computationally expensive and they fail to solve large-size instances. Therefore, we proposed hyper-heuristic in which artificial bee colony (ABC) algorithm is employed as a search methodology for the OAS problem. Hyper-heuristic works on the search space of heuristics, whereas ABC algorithm works on the solution space of the problem. A guided heuristic, which works on search space of heuristics, is developed to search the best heuristic from a set of heuristics residing at the lower level of hyper-heuristic. The proposed approach is compared with the state-of-the-art approaches. The computational results show that the integration of ABC algorithm into hyper-heuristic outperformed the other approaches in terms of average and minimum deviation from the upper bound.

Sachchida Nand Chaurasia, Joong Hoon Kim

A New Evolutionary Optimization Method Based on Center of Mass

Abstract

Physical phenomena have been the inspiration for proposing different optimization methods such as electro-search algorithm, central force optimization, and charged system search among others. This work presents a new optimization algorithm based on some principles from physics and mechanics, which is called Evolutionary Centers Algorithm (ECA). We utilize the center of mass definition for creating new directions for moving the worst elements in the population, based on their objective function values, to better regions of the search space. The efficiency of the new approach is showed by using the CEC 2017 competition benchmark functions. We present a comparison against the best algorithm (jSO) in such competition and against a classical method (SQP) for nonlinear optimization. The results obtained are promising.

Jesús-Adolfo Mejía-de-Dios, Efrén Mezura-Montes

Adaptive Artificial Physics Optimization Using Proportional Derivative Controllers

Abstract

APO (Artificial Physics Optimization) is a physicomimetics-inspired population-based global search and optimization heuristic that can be modeled as a second-order dynamical system. A central concept of physicomimetics is that the tools and techniques of modern physics and engineering may be applied directly to optimization algorithms such as APO. The extended algorithms described in this paper are a realization of this concept. Using the state-space Z-transform, APO’s performance is improved by introducing backward and forward PDCs (Proportional Derivative Controllers). Algorithm APO-PD1 employs a backward PDC architecture that allows each particle to predict its location in the optimization landscape based on its then current state of motion. An error signal computed from the distance between the particle’s predicted position and the swarm-weighted position is used to adjust the particle’s velocity through the decision space (DS) with the result that APO-PD1 is measurably better than APO. APO-PD2 further improves APO by utilizing the same error signal in a forward PDC architecture in which both the particle’s current state of motion and its trajectory history are used to predict its future location. This modification improves performance even more by allowing the swarm’s particles to change trajectories more quickly. Numerical experiments on a suite of widely employed high-dimensionality benchmarks show that APO-PD2 outperforms both APO-PD1 and APO.

Liping Xie, Jianchao Zeng, Qiongqiong Yang, Richard A. Formato

NSGA-II Based Decision-Making in Fuzzy Multi-objective Optimization of System Reliability

Abstract

This paper presents an approach to determine the optimal value of multi-objective optimization of a reliability-based system design problem. For this purpose, an over-speed protection system for a gas turbine is designed with mutually conflicting objectives such as the system reliability and system cost. This is a multi-objective nonlinear mixed integer programming problem subject to the upper limits on design constraints such as weight and volume. To solve the problem, a fuzzy approach is adopted to specify the goals in terms of the membership functions. This approach is effective in modeling the vague and imprecise information involved in the system. NSGA-II is employed to obtain the Pareto solutions efficiently. Finally, one out of these solutions is obtained by the decision-making methods such as TOPSIS and Shannon’s entropy approach. The efficiency of the proposed approach is compared with the existing approach.

Hemant Kumar, Shiv Prasad Yadav

GA-Based Task Scheduling Algorithm for Efficient Utilization of Available Resources in Computational Grid

Abstract

In the grid computing environment, systematic scheduling of tasks/jobs on hand resource is the important parameter for performance evaluation of computational grid. Traditional algorithms cannot produce a load balancing schedule. In the paper, a genetic approach for grid task scheduling has been considered to achieve better solutions within a reasonable period of time. The present study aims at minimizing the make-span and flow-time at the same time and also achieves equiponderant practical application of a set of “n” available computing agents of a grid computing to get the average load balancing. The simulation results show that the proposed approach is more efficient than the GA approach reported in the literature.

Shipra Singh, Anuradha Aggarwal, Harendera Kumar, Pradeep Kumar Yadav

Statistical Feature Analysis of Thermal Images from Electrical Equipment

Abstract

This investigation focuses on intelligent monitoring systems by assessing thermal images from electrical equipment. During modeling of any intelligent system, a variety of attributes are normally used to ensure that all the necessary information is present, which not only increases the computational complexity but also reduces classification accuracies. In this study, widely used features of thermal images like first order histogram, statistical gray level co-occurrence matrix (GLCM) and component based features are considered. The novelty of the work is that the combination of data mining techniques and clustering quality of the data in the selected feature space helps to determine the best classifier independent feature set suitable for thermal monitoring. Interestingly it is found that maximum intensity; average intensity and skewness are identified as the best feature set. Based on experimental verification, it has been demonstrated that the selected feature set gives better classification accuracies than those using all the original features. Therefore, an effective feature selection method is able to greatly improve the performance of classifiers as well as reduce the computational cost.

Tamal Dutta, Deepjyoti Santra, Chee Peng-Lim, Jaya Sil, Paramita Chottopadhyay

Performance of Sine–Cosine Algorithm on Large-Scale Optimization Problems

Abstract

The focus of this paper is the recently proposed sine–cosine algorithm (Mirjalili, Knowl-Based Syst 96:120–1330, [23]) for nonlinear continuous function optimization. The purpose of this paper is to inspect the effect of the sine–cosine algorithm on solving large-scale optimization problems. For this purpose, the algorithm is implemented on five common scalable problems appearing in literature, namely, Ackley, Griewank, Rastrigin, Rosenbrock, and Sphere functions. The dimensions of these problems are varied from 100 to 1000, and results have been recorded for fixed 10,000 iterations. The results are presented in numerical and graphical form. These results indicate that sine–cosine algorithm is a powerful nature-inspired optimization algorithm for solving all of these problems, except Sphere and Rosenbrock functions. Furthermore, the applicability of this algorithm is demonstrated by solving a real-life problem, i.e., gear train design problem.

Puneet Kumar Pal, Kusum Deep, Atulya K. Nagar

Necessary and Sufficient Optimality Conditions for Fractional Interval-Valued Optimization Problems

Abstract

In this paper, we consider the class of fractional interval-valued programming problems. Utilizing the concept of LU optimal solution, the solution concepts of such type of problems have been discussed. Further, the Fritz John and KKT optimality conditions for the nondifferentiable fractional interval-valued functions have also been established.

Indira P. Debnath, S. K. Gupta

Application of Constrained Spider Monkey Optimization to Solve Portfolio Optimization Problem

Abstract

Portfolio optimization problem has attracted the attention of researchers since ages because of its practical application. This problem is constrained in nature and deals with answering the question what amount of wealth should be invested in a particular asset. In this paper, portfolio optimization problem has been solved using Constrained Spider Monkey Optimization (CSMO) algorithm. The objective behind this work is the application of CSMO for solving a real-world optimization problem. For the experiment purpose, basic mean-variance optimization model is considered.

Kavita Gupta, Kusum Deep, Atulya K. Nagar

Optimal Configuration Selection in Reconfigurable Manufacturing System

Abstract

Reconfigurable manufacturing system (RMS) is considered as a major resource of providing variable production capacities and capabilities by different manufacturing companies. For different products needed in small quantities and with short delivery lead time, this is achieved through reconfiguring the system elements over the time. In the present work, various characteristics of RMS have been discussed and formulated. Weighted sum theory has been used for the selection of best manufacturing system. An illustration is given to analyze the applicability of the proposed methodology on a given system.

Kamal Kumar Mittal, Pramod Kumar Jain, Dinesh Kumar

A Comparative Study of Regularized Long Wave Equations (RLW) Using Collocation Method with Cubic B-Spline

Abstract

A collocation technique is successfully formulated for regularized long wave equations (RLW). This method is based on the cubic B-spline finite element. The stability analysis has been discussed by using the Fourier method and shown to be marginally stable. The accuracy, efficiency, and the invariants of motion related to conservation of mass, momentum, and energy are investigated. We have also studied the propagation of single solitary wave motion and two solitary waves interaction. It has been observed that the obtained numerical results are acceptable and more accurate.

Nini Maharana, A. K. Nayak, Pravakar Jena

An Enhanced Fractal Dimension Based Feature Extraction for Thermal Face Recognition

Abstract

Variance in pose during data acquisition poses a serious challenge for any biometric system which uses the human face as a physiological biometric feature. In this paper, we present an enhanced patchwise fractal dimension based feature extraction technique for the purpose of pose-invariant face recognition. We have presented an improved version of the Differential Box Counting (DBC) based fractal dimension computation technique which is used for feature extraction of thermal images of the human face. A Far-Infrared (FIR) imaging based human face database, called the JU-FIR-F1: FIR Face Database, was developed in the Electrical Instrumentation and Measurement Laboratory, Electrical Engineering Department, Jadavpur University, Kolkata, India for testing the accuracy, stability, and robustness of our proposed feature extraction methodology. We have included the results obtained through extensive experimentation to elaborate the superiority of our proposed algorithm over its other well-known counterparts.

Sandip Joardar, Arnab Sanyal, Dwaipayan Sen, Diparnab Sen, Amitava Chatterjee

Seismic Analysis of Multistoried Building with Optimized Damper Properties

Abstract

In today’s scenario where space is an issue, the increase in population has led to a boom in the construction industry. With the lack of land for construction, the buildings are becoming higher and more complex, so with the increase in the number of stories, it is necessary to make them safe under adverse seismic conditions. Dampers are one way to make the structure earthquake resistant and the optimization of their properties is sometimes required. In this study, the damper properties, i.e., damping and stiffness have been optimized using self-organizing migrating genetic algorithm (SOMGA) and genetic algorithm (GA) technique on a model of 10-storey building which has equal mass, stiffness, etc. on all the floors. The optimized damper properties obtained from SOMGA result in the reduction of 52% of the storey displacement while that of GA is 60% as compared to the undamped model. Both techniques provide better optimized damper properties. It is observed that the optimized damper helps in significant reduction of the seismic response of the structure, thus justifying the need of optimized parameters of dampers.

Dipti Singh, Shilpa Pal, Abhishek Singh

Effect of Upper Body Motion on Biped Robot Stability

Abstract

Achieving stability of biped robot during walking is a tough task. In this paper, we generate polynomial cubic spline for ankle joints, hip joints, and upper body so that the resulting walk is stable. Stability is assured by calculating zero momentum point with largest stability margin in Matlabs.

Ruchi Panwar, N. Sukavanam

Ant Colony Algorithm for Routing Alternate Fuel Vehicles in Multi-depot Vehicle Routing Problem

Abstract

A Multi-depot Green Vehicle Routing Problem (MDGVRP) is considered in this paper. An Ant Colony System-based metaheuristic is proposed to find the solution to this problem. The solution for MDGVRP is useful for companies, who employ the Alternative Fuel-Powered Vehicles (AFVs) to deal with the obstacles brought by the limited number of the Alternative Fuel Stations. This paper adds an important constraint, vehicle capacity to the model, to make it more meaningful and closer to real-world case. The numerical experiment is performed on randomly generated problem instances to understand the property of MDGVRP and to bring the managerial insights of the problem.

Shuai Zhang, Weiheng Zhang, Yuvraj Gajpal, S. S. Appadoo

Semidefinite Approximation of Closed Convex Set

Abstract

Approximation of convex sets takes a major role in optimization theory and practice. Approximation by semidefinite representable set draws more attention as semidefinite programming problems can be solved very efficiently using numerous existing algorithms. We contribute a technique by which a closed convex set can be approximated by a compactly semidefinite representable set. Further, we extend the technique of approximation and we prove that a closed convex set can be approximated by semidefinite representable set. These results give new techniques in semidefinite programming.

Anusuya Ghosh, Vishnu Narayanan

Titel: Decision Science in Action
herausgegeben von: Kusum Deep
Dr. Madhu Jain
Prof. Said Salhi
Verlag: Springer Singapore
Electronic ISBN: 978-981-13-0860-4
Print ISBN: 978-981-13-0859-8
DOI: https://doi.org/10.1007/978-981-13-0860-4