Pipeline trace quasioptimum determination

JORGE E. NÚÑEZ MC LEOD; SELVA S. RIVERA

Handbook of Optimization

Springer-Verlag

Año: 2013; p. 673 - 696

This chapter will focus on the development of a system with Artificial Intelligence based on Evolutionary Computation that allow generate a quasioptimum trace of a pipeline integrating Digital Elevation Models and Geographical Information Systems. The algorithm is conceived with optimization purposes based on the relevant characteristics of the trace without prior monetary quantification, although the last was taken into consideration. The chapter consist of three main sections. The first is the description of the basic information: satellite imagery, digital elevation model (DEM) and geographic information system (GIS). Describe the intermediate steps necessary for processing information. Among them are the generation of an image filter, the DEM itself, images with information on routes, railways, rivers, etc. The second section consists of a description of the design of evolutionary algorithm (EA) handling information of the first stage. The algorithm developed is framed as an evolutionary algorithm hybrid, combining features of genetic algorithms and evolution strategies. A number of criteria were defined for the implementation of AE treatment discarding infeasible individuals, a process that slows down considerably to these algorithms. Sampling was used to draw for the selection of individuals for crossover and mutation. It was determined that work with a stable population previously defined. The genetic operators crossover and mutation have been implemented, but with important differences to the classical theory. Finally the third section covers the tasks of adjusting the parameters of AE and obtaining pipeline route quasioptimum in the case of interest. By way of illustration, two examples of paths, which have varied only two features in total and every time. Although the model developed allows the simultaneous evaluation of an unbounded number of desirable features in the path. The tool developed in this chapter allows the pipeline route to obtain a quasi-optimal by using a hybrid evolutionary algorithm. The optimization is based on improving the desired characteristics of the product such as: minimum length, minimization of positive and negative slopes, minimizing lows, slope sections, sections from ridges, etc. This development that exploits modern technologies such as digital elevation models, Landsat imagery and Geographic Information Systems in a transparent, open new perspectives for feasibility studies of paths, reducing the total costs of these and allowing only do field work in the areas of quasioptimum path.