Intelligent Design Model of Urban Landscape Space Based on Optimized BP Neural Network. (25th March 2022)

• Record Type: Journal Article
• Authors: Wu, Guoping; Miao, Yihua; Wang, Fang
• Other Names: Zeng Wen Academic Editor.
• Is Part Of: Journal of sensors. Volume 2022(2022)
• Publication Date: 2022-03-25
• Subjects: Detectors -- Periodicals; 681.205
• DOI: 10.1155/2022/9704287
• Languages: English
• ISSNs: 1687-725X

Abstract

Abstract : Urban landscape space planning is an important application field of landscape ecology. With the continuous development of the field of architectural optimization, more and more optimization methods have sprung up, including various intelligent optimization algorithms. Such intelligent optimization algorithms usually rely on traditional building performance simulation methods to obtain building performance indicators for optimization in the optimization process. However, intelligent optimization algorithms generally require large-scale calculations. At the same time, the time required for building performance simulation is often limited by the complexity of the building model and the configuration of the computer, which leads to too long performance optimization time for designers in the project. With efficient and accurate feedback, building performance optimization methods based on intelligent optimization algorithms are mainly used in scientific research and are difficult to invest in actual projects. Because the traditional BP neural network has its own limitations and its insufficient sample size and weak generalization ability in complex prediction problems, this paper uses the learning algorithm of optimizing the BP neural network to propose an urban landscape space intelligent design model. This article introduces the artificial neural network, a new application technology with the assistance of a geographic information system, and establishes a BP neural network model for urban landscape ecological planning. Seven elements of distance and number of residential points are used as input variables, patch density, fractal dimension, the Shannon diversity index, and aggregation degree are selected as output changes, and 20 samples are carefully collected to train the network. The results show that the network convergence effect is ideal and the generalization ability is strong, which provides a new simulation analysis method for landscape ecological planning.