Chapter 3
OPTIMIZATION
3.1 Introduction
Optimization is a chronic and natural process usually witnessed in our daily life events. In various disciplines such as engineering designs, manufacturing systems, agricultural sciences, physical sciences, economics, pattern recognition etc. optimization is observed. Optimization is, thus a process of making best, effective and functional solution out of possible choices no way differs from the structural optimization which is being conceived in the present work. Structural optimization is a decisive and tricky step, where the designers are able to generate better designs saving time and money. Conventional optimization approaches like mathematical programming method, optimality criteria method etc. fails miserably in structural design problems which are highly complex and time consuming in nature. Optimization problems are mathematical models formulated to solve complex designs that may be of multi-objective nature in certain cases. Structural design procedure involves conceptual design and design realization leading to several probable results since higher degree of ambiguity is experienced in every steps. Conceptual design phase is more dependent on decision variables than in advance optimization phase. Optimization is one of the major tool for decision making at the conceptual or realization phase of modern design techniques. In design realization stage optimization is achieved by mathematical and numerical search methods.
Extensive ranges of mathematical programming methods with higher potentials are available for solving wide varieties of engineering optimization especially for structural design problems. These practical structural optimization problems considers discrete mathematic...
... middle of paper ...
...ciated with its parallel processing capability. To solve optimization problems, neural computing techniques have been adopted in the recent past. A neural network is a massively parallel network of interconnected simple processors (neurons) in which each neuron accepts a set of inputs from other neurons and computes an output that is propagated to the output nodes. Thus a neural network can be described in terms of the network connectivity, individual neurons, the weights associated with the interconnections between neurons and the activation function of each neuron. The network maps an input vector from one space to another. In case of the network mapping it is not specified but is learned.
Out of these methods of optimization, mostly chosen and the one chosen in the present study is genetic algorithm, a detailed discussion on which is been given in chapter 4.
Noori, S., Feylizadeh, M. R., Bagherpour, M., Zorriassatine, F., & Parkin, R. M. (2008). Optimization of material requirement planning by fuzzy multi-objective linear programming. Proceedings of the Institution of Mechanical Engineers, 222, 887-900. Retrieved from http://search.proquest.com/docview/195144743?accountid=32521
The advent of neural net with the seminal work of Hopfield , popularized the use of machine intelligence techniques in the pattern recognition. However, the dense and inherent structure of neural networks is not suitable for VLSI implementation. So, researchers in the neural network domain tried to simplify the structure of the neural network by pruning unnecessary connections. Simultaneously, the CA research community explored the advantages of the sparse network structure of cellular automata for relevant applications. The hybridization of cellularity and neural network has given rise to the popular concept of cellular neural networks.
...means and become familiar with K-means clustering and its usage. Then, we finish this part by different method of clustering. The K-nearest- neighbors is also discussed in this chapter. The KNN is simple for implication, programming, and one of the oldest techniques of data clustering as well. There are many applications existing for KNN and it is still growing. The PCA also discussed in this chapter as a method for dimension reduction, and then discrete wavelet transform is discussed. For the next chapter the combination of PCA and DWT, which can be useful in de-noising, come about. In this study, we have examined the neural network structure and modeling that is most of usage these days. The backpropagation is one of the common methods of training neural networks and for the last model, we discussed autoregressive model and the strategies to choose a model order.
Stergiou, C., & Siganos, D. (2011, August 6). Neural Networks. Retrieved August 6, 2011, from
The idea is that the designer first establishes rules and relations by which design components are connected to minimize the time and effort consumed in modifications, and to provide multiple solutions that could not be reachable by traditional methods. The parametric approach has been studied and analyzed by numerous academics and designers (Araya, S., 2006, pp.11-12; Gane, V., 2004, p.54; Hudson, R., 2008, pp.18-19; Llabres, E. and Rico, E., 2016). Most of them coincide describing it as a series of phases, which increase in the level of detail and precision, as they involve from preliminary concept to construction. Herein, the parametric design process starts with Design Exploration, in which background data and design problems are determined, including the design objectives, variables, and constraints. The second phase, Design Development, includes possible solutions for design problems and manipulations of design instances. Generation of alternative solutions are reviewed and evaluated in the Simulation / Evaluation phase, to satisfy project goals, and previously built constraints. After these explorations, a development is considered one single direction in the Manufacturing / Construction phase (Araya, S., 2006, p.12; Gane, V., 2004,
Genetic Algorithms provide a holistic search process based on principles of natural genetics and survivals of the fittest……
Below I will analyse the most important components of an evolutionary computation algorithm and explain how it works.
Digital fabrication is coming into the construction industry to create precisely crafted and complex buildings in response to the new competitive environment and construction market demands. Thanks to advanced technology, the good revolution can be seen in most industrial activities. Almost all of the industry sectors are trying to keep themselves up to date with new related innovation to boost their sustainable growth. However, unfortunately architects and construction engineers have been more conservative despite all of the massive global investment in the construction sector. This conservative trend was started many years after great prosperity and success in the other industries. As an example, in comparison with the more developed automotive industry, the construction industry has been weighing the pros and cons of doing automation, and it is still under development. Despite some limitations and lack of information over the 80’s and even 90’s, but the current status indicates good progress. Today, most of architectural design is no longer possible without computer technologies. The models have become more complex and require advanced tools to understand design codes and implement fabrication processes. It can be said, that the advanced automated tools with a user-friendly programming system can bring incredible solutions for architects and construction engineers. Many researchers have made significant studies to consider all possibilities and limitations of digital design-construction in various types of procedures.
Various learning situations may dictate differing learning processes. The three that will be briefly highlighted in this paper are; learning by induction, through the use of decision rules or decision trees; learning by discovery; and learning by taking advice, explanation-based generalization. The concept of multi-strategy learning in order to handle more complex problems will also be examined.
LINGO is a simples and powerful software that can be used to solve MIP optimization problems. This software can handle tens of thousands of variables and constraints with up to few thousand integer variables (Schrage, 2006). Wong et al. (2010) and Easa and Hossain (2008) used this software to solve MIP problems to find the global optimal solution.
Genetic Algorithm is a sequential procedure developed from the science involved in genetic behaviour organisms for optimization purpose. Working Principle of GA includes the simulation of evolution theory in which, the initial set of “population” is selected in random, and then successive "generations" of solutions are reproduced till the optimal convergence. Existence of the fittest individual and natural selection operators is the main agenda of GA process. Philosophically one can say that GAs are based on Darwin’ theory of survival of the fittest. Genetic algorithm is a method for solving optimization problems that is based on natural selection, the process that drives biological evolution. Being analogous to genetics, it is a long complex thread of DNAs and RNAs containing the hereditary data, by which a traits of each individual can be determined, as chromosomes. Each trait in living organisms is being coded with some combination of DNAs like A (Adenine), C (Cytosine), T (Thymine) and G (Guanine).
Artificial neural networks are systems implemented on computer systems as specialized hardware or sophisticated software that loosely model the learning and remembering functions of the human brain. They are an attempt to simulate the multiple layers of processing elements in the brain, called neurons. These elements are implemented in such a way so that the layers can learn from prior experience and remember their outputs. In this way, the system can learn to recognize certain patterns and situations and apply these to certain priorities and output appropriate results. These types of neural networks can be used in many important situations such as priority in an emergency room, for financial assistance, and any type of pattern recognition such as handwritten or text-to-speech recognition.
Classification algorithms is the process of a computer relating a subject to a category. To best explain this concept, Stephen Marsland states “…consider a vending machine, where we use a neural network to learn to recognize different coins” (Machine Learning, Section 1.4). The computer learns by analyzing large amounts of data and then categorizing the data. This is how a computer system can identify a certain illness to assist medical staff identify a certain type of illness or disease. In addition, supervised learning can also utilize regression
Fortunately, during under-graduation, I got an opportunity to detect the optimum path of a process through my project “Design and Development of PCB Dual Head Drilling Machine”. Additionally, I became aware of a new subject called Six Sigma which aided me in intertwining new optimization techniques into my project to make the process effective. I optimized the creation...
Manufacturing the process design also faces the same challenge in the optimization of the product (ARAUJO, JOSÉ AUGUSTO DA ROCHA DE; Costa, Reinaldo Pacheco da, 2004, p. 4).