本文選題：BP神經(jīng)網(wǎng)絡(luò) + 遺傳算法　； 參考：《電子科技大學(xué)》2013年碩士論文

【摘要】：廣州市地處珠江三角洲經(jīng)濟(jì)發(fā)達(dá)密集群地帶，城市化發(fā)展速度極快，由于高強(qiáng)度的政治經(jīng)濟(jì)活動(dòng)造成的城市化擴(kuò)張現(xiàn)象十分明顯，而由此帶來(lái)了一系列環(huán)境問(wèn)題也逐漸引起了各界學(xué)者的高度關(guān)注。但是由于城市化進(jìn)程是一個(gè)影響因子眾多、變化過(guò)程復(fù)雜的城市復(fù)合系統(tǒng)的變化過(guò)程，目前相關(guān)的理論研究依然存在較大的局限。為此，基于HJ-1A/1B遙感影像的數(shù)據(jù)基礎(chǔ)，本文將以廣州市為例學(xué)習(xí)和研究城市擴(kuò)張驅(qū)動(dòng)因子和非線性人工智能算法在城市擴(kuò)張模擬中的應(yīng)用。最終通過(guò)建立廣州市城市擴(kuò)張動(dòng)態(tài)變化模型，結(jié)合分析城市擴(kuò)張影響下的城市熱島效應(yīng)，來(lái)綜合評(píng)價(jià)廣州市生態(tài)環(huán)境的變化趨勢(shì)。本文主要研究成果如下： （1）基于城市土地結(jié)構(gòu)的理論知識(shí)，本文通過(guò)采用緩沖區(qū)分析、主成分分析和聚類(lèi)分析等方法建立城市土地利用類(lèi)型和經(jīng)濟(jì)、政策、人口等因素的相關(guān)關(guān)系，并依此評(píng)價(jià)驅(qū)動(dòng)因子在城市變化過(guò)程中發(fā)揮作用，研究廣州市城市化進(jìn)程的特點(diǎn)。 （2）基于面向?qū)ο蠓椒�，本文提取和分析廣州市2009年至2011年的土地利用覆蓋特征，并完成城市擴(kuò)張環(huán)境因素的定量化描述，最終利用MATLAB實(shí)現(xiàn)基于BP神經(jīng)網(wǎng)絡(luò)和遺傳算法優(yōu)化的元胞自動(dòng)機(jī)模型在廣州市的土地覆蓋變化中的模擬預(yù)測(cè)。通過(guò)對(duì)比模擬結(jié)果可知：BP神經(jīng)網(wǎng)絡(luò)較傳統(tǒng)的Geo-Urban元胞自動(dòng)機(jī)模型，，能更好地模擬分布較為集中的耕地和林地等區(qū)域，精度可達(dá)到70%以上，而對(duì)于面積較為零碎的建筑用地區(qū)域模擬效果較差；遺傳神經(jīng)網(wǎng)絡(luò)優(yōu)化算法則能夠較BP神經(jīng)網(wǎng)絡(luò)總體提高約5%的模擬精度，部分土地類(lèi)型的模擬精度能提高至20%；并且遺傳神經(jīng)優(yōu)化算法還能夠充分考慮影響土地變化的各種擾動(dòng)因素，優(yōu)化選擇驅(qū)動(dòng)因子和縮短迭代次數(shù)，對(duì)于城市土地?cái)U(kuò)張的動(dòng)態(tài)模擬研究具有可行性。 （3）對(duì)于城市擴(kuò)張影響下的城市熱島效應(yīng)研究，本文主要運(yùn)用JMS普適性單通道算法來(lái)定量反演廣州市的城市熱環(huán)境，重點(diǎn)研究廣州市不透水面、土地覆蓋和植被指數(shù)與城市熱環(huán)境的相關(guān)性。研究結(jié)果顯示：廣州市建筑用地的擴(kuò)張導(dǎo)致連續(xù)3年城市熱效應(yīng)顯著加劇；并且城市平均地表溫度與不透水面面積呈現(xiàn)正相關(guān)，與城市的植被指數(shù)和裸土指數(shù)呈現(xiàn)負(fù)相關(guān)，為此加強(qiáng)城市綠地建設(shè)是緩解城市熱島效應(yīng)，實(shí)現(xiàn)城市可持續(xù)發(fā)展的一項(xiàng)重要手段。
[Abstract]:Guangzhou is located in the Pearl River Delta, where the development of urbanization is very fast. The phenomenon of urbanization expansion caused by the high intensity of political and economic activities is very obvious. As a result, a series of environmental problems have gradually attracted the attention of scholars from all walks of life. However, due to the fact that urbanization is a changing process of urban complex system with many influencing factors and complex changing process, there are still some limitations in the relevant theoretical research at present. Therefore, based on the data base of HJ-1A/1B remote sensing image, this paper will take Guangzhou as an example to study and study the application of urban expansion driving factor and nonlinear artificial intelligence algorithm in urban expansion simulation. Finally, by establishing the dynamic change model of urban expansion in Guangzhou and analyzing the urban heat island effect under the influence of urban expansion, the change trend of ecological environment in Guangzhou is evaluated synthetically. The main research results of this paper are as follows: Based on the theoretical knowledge of urban land structure, this paper uses buffer zone analysis, principal component analysis and cluster analysis to establish the correlation between urban land use types and economic, policy, population and other factors. According to this evaluation, the driving factors play a role in the process of urban change, and study the characteristics of the urbanization process in Guangzhou. Based on the object-oriented method, this paper extracts and analyzes the land use and cover characteristics of Guangzhou from 2009 to 2011, and completes the quantitative description of the environmental factors of urban expansion. Finally, the cellular automata model based on BP neural network and genetic algorithm optimization is realized by using MATLAB to predict the land cover change in Guangzhou city. Compared with the traditional Geo-Urban cellular automata model, the comparison of the simulation results shows that the proportion BP neural network can better simulate the distribution of cultivated land and woodland, and the precision can reach more than 70%. But the simulation effect of the area of the building land is not good, and the optimization algorithm of genetic neural network can improve the simulation accuracy by about 5% compared with the BP neural network as a whole. The simulation accuracy of some land types can be improved to 20 parts, and the genetic neural optimization algorithm can also take into account all kinds of disturbance factors that affect land changes, optimize the selection of driving factors and shorten the number of iterations. It is feasible to study the dynamic simulation of urban land expansion. 3) for the study of urban heat island effect under the influence of urban expansion, this paper mainly uses JMS universal single-channel algorithm to quantitatively inverse the urban thermal environment of Guangzhou, with emphasis on the study of the impermeable water surface in Guangzhou. The correlation between land cover and vegetation index and urban thermal environment. The results show that the urban thermal effect is significantly increased due to the expansion of urban land for buildings in Guangzhou for three consecutive years, and the average urban surface temperature is positively correlated with the area of impervious surface, and negatively correlated with the vegetation index and bare soil index of the city. Therefore, strengthening the construction of urban green space is an important means to alleviate the urban heat island effect and realize the sustainable development of the city.
【學(xué)位授予單位】：電子科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2013
【分類(lèi)號(hào)】：TU984.2;X16

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 趙姚陽(yáng);濮勵(lì)杰;胡曉添;;BP神經(jīng)網(wǎng)絡(luò)在城市建成區(qū)面積預(yù)測(cè)中的應(yīng)用——以江蘇省為例[J];長(zhǎng)江流域資源與環(huán)境;2006年01期

2 韋亞平;趙民;肖瑩光;;廣州市多中心有序的緊湊型空間系統(tǒng)[J];城市規(guī)劃學(xué)刊;2006年04期

3 陶嘉;黎夏;劉小平;何晉強(qiáng);;分析學(xué)習(xí)智能元胞自動(dòng)機(jī)及優(yōu)化的城市模擬[J];地理與地理信息科學(xué);2007年05期

4 楊青生;黎夏;;多智能體與元胞自動(dòng)機(jī)結(jié)合及城市用地?cái)U(kuò)張模擬[J];地理科學(xué);2007年04期

5 史培軍,陳晉,潘耀忠;深圳市土地利用變化機(jī)制分析[J];地理學(xué)報(bào);2000年02期

6 楊青生;黎夏;;基于粗集的知識(shí)發(fā)現(xiàn)與地理模擬——以深圳市土地利用變化為例[J];地理學(xué)報(bào);2006年08期

7 劉小平;黎夏;艾彬;陶海燕;伍少坤;劉濤;;基于多智能體的土地利用模擬與規(guī)劃模型[J];地理學(xué)報(bào);2006年10期

8 陳逸敏;黎夏;劉小平;李少英;;基于耦合地理模擬優(yōu)化系統(tǒng)GeoSOS的農(nóng)田保護(hù)區(qū)預(yù)警[J];地理學(xué)報(bào);2010年09期

9 李秀彬;全球環(huán)境變化研究的核心領(lǐng)域──土地利用／土地覆被變化的國(guó)際研究動(dòng)向[J];地理學(xué)報(bào);1996年06期

10 黎夏,葉嘉安;基于神經(jīng)網(wǎng)絡(luò)的元胞自動(dòng)機(jī)及模擬復(fù)雜土地利用系統(tǒng)[J];地理研究;2005年01期

相關(guān)博士學(xué)位論文 前1條

1 陶文芳;西安—咸陽(yáng)地區(qū)土地覆被時(shí)空變化及驅(qū)動(dòng)因子研究[D];西北農(nóng)林科技大學(xué);2010年

相關(guān)碩士學(xué)位論文 前1條

1 劉磊;MODIS數(shù)據(jù)地表溫度反演及其在長(zhǎng)江三角洲都市群熱島效應(yīng)研究中的應(yīng)用[D];南京信息工程大學(xué);2007年

本文編號(hào)：1775577