Throughout the last decades, one of society’s concerns has been the development of new tools to optimize every aspect of daily life. One of the mechanisms that can be applied to this effect is what is nowadays called Artificial Intelligence (AI). This branch of science enables the design of intelligent systems, meaning that they display features that can be associated to human intelligence, search methods being one of the most remarkable. Amongst these, Evolutionary Computation (EC) stands out. This technique is based on the modelling of certain traits of nature, especially the capacity shown by living beings to adapt to their environment, using as a starting point Darwin’s Theory of Evolution following the principle of natural selection (Darwin, 1859). These models search for solutions in an automatized way. As a result, a series of search techniques which solve problems in an automatized and parallel way has arisen. The most successful amongst these are Genetic Algorithms (GA) and, more recently, Genetic Programming (GP). The main difference between them is rooted on the way solutions are coded, which implies certain changes in their processing, even though the operation in both systems is similar. Like most disciplines, the field of Civil Engineering is no stranger to optimization methods, which are applied especially to construction, maintenance or rehabilitation processes (Arciszewski and De Jong, 2001) (Shaw, Miles and Gray, 2003) (Kicinger, Arciszewski and De Jong, 2005). For instance, in Structural Engineering in general and in Structural Concrete in particular, there are a number of problems which are solved simultaneously through theoretical studies, based on physical models, and experimental benchmarks which sanction and adjust the former, where a large amount of factors intervene. In these cases, techniques based on Evolutionary Computation are capable of optimizing constructive processes while accounting for structural safety levels. In this way, for each particular case, the type of materials, their amount, their usage, etc. can be determined, leading to an optimal development of the structure and thus minimizing manufacturing costs (Rabuñal , Varela, Dorado, González and Martínez, 2005).
At the origin of what is now known as Genetic Algorithms are the works of John Holland at the end of the 1960’s. He initially named them “Reproductive Genetic Planning”, and it wasn’t until the 70’s that they received the name under which they are known today (Holland, 1975).
GA is a search algorithm inspired on the biological functioning of living beings. It is based upon reproductive processes and the principle which determines that better environmetally adapted individuals have more chances of surviving (Goldberg, 1989).
Like living beings, GAs use the basic heritage unit, the gene, to obtain a solution to a problem. The full set of genes (parameters characterizing the problem) is chromosome, and the expression of the chromosome is an individual in particular.
In Computer Science terms, the representation of each individual is a chain, usually binary, assigning a certain number of bits to each parameter. For each variable represented a conversion to discrete valued has to be performed. Obviously, not all parameters have to be coded with the same number of bits. Each one of the bits in a gene is usually called allele. Once the individuals’ genotype (the structure for the creation of an individual in particular) is defined, we are ready to carry out the evolutionary process that will reach the solution to the problem posed.
We start with a random set of individuals, called a population. Each of these individuals is a potential solution to the problem. This would be the initial population or zero generation, and successive generations will be created from it until the solution is reached. The mechanisms used in the individuals’ evolution are analogous to the functioning of living beings:
Key Terms in this Chapter
Deep Beam: A flexural member whose span-to-depth ratio is too low to accurately apply the principles of sectional design through sectional properties and internal forces shear strength the maximum shearing stress a flexural member can support at a specific location as controlled by the combined effects of shear forces and bending moment
Mixture Proportion: The proportions of ingredients that make the most economical use of available materials to produce mortar or concrete of the required properties.
Compressive Strength: The measured maximum resistance of a concrete or mortar specimen to axial compressive loading; expressed as force per unit crosssectional area; or the specified resistance used in design calculations.
Superplasticizer or High-Range Water-Reducing Admixture: A water-reducing admixture capable of producing large water reduction or great flowability without causing undue set retardation or entrainment of air in mortar or concrete.
High Performance Concrete: Concrete meeting special combinations of performance and uniformity requirements that cannot always be achieved routinely using conventional constituents and normal mixing, placing, and curing practices.
Workability: That property of freshly mixed concrete or mortar that determines the ease with which it can be mixed, placed, consolidated, and finished to a homogenous condition.
Slump: A measure of consistency of freshly mixed concrete, mortar, or stucco equal to the subsidence measured to the nearest 1/4 in. (6 mm) of the molded specimen immediately after removal of the slump cone.