A neural network model to develop actions in urban complex systems represented by 2D meshes

Abstract

The main idea of this work is to present a tool which may be useful to generate a mesh of points where urban actions may be taken after analysing and understanding complex urban situations. By the word complex we mean urban concentrations without precise limits and without a recognizable geometry pattern. In these situations, it is very hard for the architects to understand the system. Therefore, it is very difficult to define an action plan for this type of urban situations. What we propose is an adaptation of a neural network algorithm to work in the context of urban networks. Our objective is to develop a strategy to change this weakness of sparse urban development by activating public spaces with new meanings. A new 2D triangle mesh simplification model is introduced with the central property of preserving the shape of the original mesh. The mathematical model presented consists of a self-organizing algorithm whose objective is to generate the positions of the nodes of the simplified mesh; afterwards, a triangulation algorithm is carried out to reconstruct the triangles of the new simplified mesh. With this algorithm, it is possible to perform specific actions in an urban space, because the urban territory can be considered as a complex mesh with nodes and edges. A real example of an urban action is shown with the introduction of a wireless network in a residential area.

Keywords:

• submission instructions
• source file coding
• artificial neural networks
• urban systems
• mesh simplifications
• triangulation
• GNG3D model
• urban density

2010 AMS Subject Classifications :

• 68T05
• 68Q32
• 68Q25