Optimization
Chapter 16 provides a comparison of the performance of four machine learning algorithms—Naïve Bayes, Neural Network, Support Vector Machine, and K-nearest Neighbors—in spam classification. The implementation of the algorithms is carried out in R and performance is evaluated by using AUC of the ROC curve, Accuracy, Kappa, and F-Measure. The results revealed that the SVM algorithm performed better than the other algorithms. This work showed that the receiver operating curve–area under the curve (ROC-AUC) is better suited for use in the machine learning world when compared to the accuracy metrics which are generally used in assessing the performance measurement of a classification algorithm.
Chapter 17 deals with an inventory system where urea bags of varying bulk sizes arrive at the warehouse, in which the arrivals follow the Poisson process and the inter arrival times follow exponential distribution. Probability distribution of inventory levels and total expected cost per unit time are obtained, supported with numerical calculations and graphical representations.
Chapter 18 represents a single-objective prototype for supply chain optimization considering disruption scenarios. The goal is to lessen the amount of the setup cost, shipping cost, production price, inventory expense, purchasing cost and scenario cost. A mixed integer linear programming model is developed which as a result of multiple entities is complex. The intention is to find a solution to such a model by developing a solver with the intent of providing a comparative study with different evolutionary approaches and numerical methods like branch and bound.
Chapter 19 studies the tax risk profile of South African construction companies, which is characterized by the book value, cash flow position, headcount, firm earnings, debt size and type of firm. The model that defines this tax risk is a neural network (NN) boosted generalized linear model (GLM). The main aim of this study was to develop an artificially intelligent pricing model. The study, which was conducted using an examination of financial statements of construction companies, highlights the key determinants of the price of tax risk for construction firms. Modeling techniques used to build the pricing model are discussed in detail along with their challenges.
Chapter 20 proposes a design of a simplified Type-A Schiffman phase shifter (SPS) based on microstrip transmission line (TL) technology. This phase shifter (PS) is designed to obtain a phase shift of 90 degrees at the resonance frequency. In this design, the stub matching technique is employed to match the impedance. This design is tunable, thereby obtaining a phase shift from 45 to 90 degrees with a phase deviation of 5 degrees in the resonance frequency of 2.4GHz. A varactordiode is introduced to make the design tunable. The design is carried out using the FR4 substrate, with an operating frequency of 2.4GHz. An IE3D full-wave simulation platform is used for simulation purposes.
Chapter 21 explores manufacturing competencies and sustainability issues for automobile manufacturing companies. The work is based on organizations in north Indian automobile and auto parts manufacturing companies, with various manufacturers being treated alike irrespective of the manufacturing sector. A qualitative model was developed for depicting competency and strategy relation. The research provided an insight into manufacturing competencies and their relation tostrategic success; it also discovered areas that could be improved with further research.
Chapter 22 creates a nonlinear continuous review inventory model for multiple products considering the quantity of the products received as uncertain with controllable lead time. The solution of the model was discussed with and without considering service level constraint. The Lagrangian method is applied for the model with service level constraints and an optimal solution is arrived at.
Part III: Metaheuristics – Applications and Innovations
Chapter 23 proposes a completely innovative metaheuristic optimization algorithm. This new method was inspired by the circular structures on the seafloor which are created by one of the pufferfish species (Torquigeneralbomaculosus). The basic parameters of the inventive method were determined by the parameters observed during the process of the circular structure of the pufferfish. The process of performing this natural phenomenon is detailed in this work. In addition, detailed information about the concept of optimization, development process and activity areas are given in the study. The performance of this proposed method, which is inspired by nature, has been compared with other methods used extensively in the literature.
Chapter 24 proposes a hybrid optimization approach called HGWOSSO based on the integration of two swarm-based approaches, namely grey wolf optimizer (GWO) and sperm swarm optimization (SSO). The aim ofthis hybridization is to merge and enhance the capabilities of exploitation and exploration in both SSO and GWO to generate both in varied strengths. The functions of fixed-dimension multimodal, multimodal, and unimodal benchmarks gleaned from the literature are utilized to check the solution quality and performance of the HGWOSSO variant. The results revealed that the local search in SSO increases the ability of the hybrid variant in solving the benchmark functions, which significantly outperforms the GWO variant in terms of quality of solutions and capability of reaching the global optimum.
Chapter 25 envisions accessible lines of research associated with metaheuristics and highlights less explored areas of considerable concern. The authors concentrate on other metaheuristic approaches, hybrid processes, parallel metaheuristics, metaheuristics under uncertainty and multi-objective optimization. A review of these methods shows that while they are linked to several works, they have not been thoroughly investigated, and there are several open lines of study. The work considered in this chapter is especially beneficial for those researchers looking for novel fields in metaheuristics for multi-objective research and multi-objective optimization.
Chapter 26 deals with the issue of order reduction and controller synthesis in a unified domain for the PMSM drive. Two basic algorithms, viz. the firefly technique and the bacterial foraging optimization technique, are integrated to constitute a new topology known as the hybrid firefly algorithm (HFA). Originally, a PMSM drive consisting of both speed and current controllers created a higher-order system that has been reduced to a lower-order model via an identification method used in signal processing technology. In cascade control with a PI controller, the reduced-order model is there upon compared with that of the reference plant to roughly assess the unknown three-term controller parameters. The control parameters in the unified domain resemble almost accurately the continuous-time parameters at a low sampling limit. A unified controller design framework is thus developed for the drive. The smart algorithm is therefore successfully used both for the order reduction and for the estimation of the controller parameter of PMSM drives.
Chapter 27 presents the three-diode model-based PV module. The Harris hawks’ optimization (HHO) algorithm is used to estimate all the nine parameters of the system for three types of commercially available PV modules; namely, KC200GT multi-crystal, CS6K-280M mono-crystalline, STM6 40-36 mono-crystalline, Pro. SW255 poly-crystalline. The competitive and statistical experimental results show that HHO is advantageous in the sense that the sum of square error is lower as compared to that with other wellknown algorithms. The suggested technique also exhibits better convergence than the salp swarm algorithm (SSA), grey wolf optimizer (GWO), sine cosine algorithm (SCA), and dragonfly algorithm (DA).
