سال انتشار: 2022

رشته: معماری، مهندسی کامپیوتر

گرایش: تکنولوژی معماری، معماری منظر، هوش مصنوعی

دانلود رایگان این مقاله:

  دانلود مقاله برنامه ریزی منظره فضای سبز

مشاهده سایر مقالات جدید:

مقالات جدید مهندسی کامپیوتر

مقالات جدید معماری

3. Evaluation system of urban green space landscape planning based on PSO-BP neural network model

3.1. Application feasibility and evaluation index of neural network in urban green space landscape planning scheme evaluation Particle swarm optimization (PSO) is also translated into particle swarm optimization, particle swarm optimization, or particle swarm optimization. It is a random search algorithm based on group cooperation, which is developed by simulating the foraging behavior of birds. It is generally considered to be a kind of cluster intelligence. It can be incorporated into multiagent optimization system. The advantage of evolutionary computing is that it can deal with some problems that cannot be handled by traditional methods. Examples include non differentiable node transfer functions or no gradient information. But the disadvantages are: 1. The performance is not particularly good on some problems. 2. The coding of network weight and the selection of genetic operator are sometimes troublesome. From the perspective of value, it plays an important role in the urban construction system and plays an important and positive role in the construction of urban ecology, economy, society and modern civilization. Landscape is an indispensable part of modern urban planning, whether in the overall urban planning stage or in the detailed planning stage. The purpose of urban green space landscape planning is to redistribute, adjust and construct the urban green space landscape through human influence and planning, so as to restore the scattered ecosystem and environment. There is a close and complex relationship between different landscape patterns in urban green space landscape planning, and this potential relationship can not be expressed by accurate equations or algorithms. Therefore, the artificial neural network with black box characteristics can reflect the relationship between urban green space landscape, as shown in Fig. 1. Although there are many uncontrollable and unforeseen factors in urban green space landscape planning, the evaluation accuracy of the planning scheme can not reach 100%, but it can be achieved as sustainable and reasonable as possible for the green space landscape planning. And the artificial neural network can be used as an important basis for urban green space landscape planning and evaluation by studying the changes of natural landscape, the psychological needs and behavioral characteristics of urban residents. In addition, the ecological of urban green space landscape changes and develops under the joint action of various factors with different weights. The process of influence in various aspects is complex and may overlap with each other. The training and learning purpose of artificial neural network is to determine the weights of these factors, so it can be considered that, The existing urban green space landscape and the planned green space landscape constitute input layer and output layer in structure, and the complex analysis and correction process in the middle constitute the hidden layer. Artificial neural network is mainly used in three aspects, namely, resource evaluation, landscape ecology analysis and recreation analysis. Resource evaluation is to analyze and evaluate the quality, quantity and spatial distribution of green space landscape based on the results of the city field investigation, and to distinguish the different regional values. In the past, the quantitative analysis of urban green space landscape resources was carried out by using expert method and principal component analysis method. Artificial neural network can learn multiple professional standards and aesthetic standards by using super storage function and establish multiple expert analysis database for resource evaluation. Landscape ecological analysis is to analyze the landscape characteristics, landscape pattern, ecological state and sensitivity of the city, so as to judge the suitable area for green space landscape construction and ecological sensitive areas for ecological protection, and further reasonably plan the distribution pattern of green space landscape. At present, the survey data and actual situation error are often found in the process of landscape ecological analysis, which makes the accuracy and scientific of the analysis results decrease. The combination of artificial neural network and remote sensing technology can improve the accuracy and accuracy of image data, and the artificial neural network has certain advantages in image recognition. It can have a more scientific analysis of the pattern of green landscape planning, and can guide the layout and development of urban green landscape.

The analysis process of recreation is to predict and analyze the behavior of urban residents on the basis of simple understanding, so as to obtain the frequency of use of corresponding green space and the form of tourists using green space, and then classify and analyze the data to obtain the corresponding data model carrier. Although it can predict and characterize the development trend of urban residents’ behavior, the model can not be included in the analysis of human psychology and other factors. The difference between artificial neural network and traditional method is that it has no fixed mathematical model and can take into account the factors that can not be added in the past . At the same time, we can learn and correct the errors constantly in the process of training and learning, and get the analysis results that are more consistent with the social behavior in the simulated space.