基于BP神經(jīng)網(wǎng)絡(luò)的高速公路車(chē)流量預(yù)測(cè)研究
發(fā)布時(shí)間：2019-04-27 11:53
【摘要】：隨著我國(guó)改革開(kāi)放不斷深入，人們客運(yùn)和貨運(yùn)需求不斷上升，對(duì)建成的高速公路的通行能力提出了更高的要求，考慮環(huán)境和成本費(fèi)用等問(wèn)題，盲目的擴(kuò)建高速公路是不可取的，高速公路的建設(shè)應(yīng)該從一味的量增轉(zhuǎn)變到合理規(guī)劃、有效益的增長(zhǎng)上來(lái)，這樣可以減少不必要的投資。因此就要求企業(yè)對(duì)原有建成的高速公路車(chē)流量進(jìn)行準(zhǔn)確的預(yù)測(cè)，把預(yù)測(cè)結(jié)果作為交通規(guī)劃決策的依據(jù)和企業(yè)未來(lái)收益預(yù)測(cè)的依據(jù)。 高速公路車(chē)流量的預(yù)測(cè)屬于一種長(zhǎng)期車(chē)流量預(yù)測(cè)，而且容易受社會(huì)環(huán)境各方面的影響，，為了提升預(yù)測(cè)的準(zhǔn)確性，必須選擇一種對(duì)環(huán)境適應(yīng)性更強(qiáng)的預(yù)測(cè)模型。神經(jīng)網(wǎng)絡(luò)模型不僅具有實(shí)行大規(guī)模的并行處理的優(yōu)點(diǎn)，可以在同時(shí)分析大量相關(guān)因素的情況下保證系統(tǒng)能以更快的速度輸出可靠結(jié)果，還具有非線性映射特性，這就大大增強(qiáng)了神經(jīng)網(wǎng)絡(luò)模型適應(yīng)環(huán)境的能力。因此，運(yùn)用神經(jīng)網(wǎng)絡(luò)模型可以對(duì)高速公路車(chē)流量進(jìn)行比較準(zhǔn)確的預(yù)測(cè)。 本文對(duì)現(xiàn)階段高速公路車(chē)流量預(yù)測(cè)方法進(jìn)行了系統(tǒng)的梳理，總結(jié)不同預(yù)測(cè)模型存在的優(yōu)缺點(diǎn)，構(gòu)建了基于BP神經(jīng)網(wǎng)絡(luò)高速公路車(chē)流量預(yù)測(cè)模型，結(jié)合高速公路車(chē)流量數(shù)據(jù)的特點(diǎn)，對(duì)高速公路車(chē)流量樣本數(shù)據(jù)預(yù)處理方法和BP神經(jīng)網(wǎng)絡(luò)預(yù)測(cè)模型的激勵(lì)函數(shù)進(jìn)行了改進(jìn)，確定了預(yù)測(cè)模型中各參數(shù)的初始值的方法，同時(shí)提出了新建或改擴(kuò)建高速公路對(duì)預(yù)測(cè)項(xiàng)目影響的定量化方法，從而結(jié)合影響程度定量化的結(jié)果對(duì)神經(jīng)網(wǎng)絡(luò)模型的預(yù)測(cè)值進(jìn)行改進(jìn)，提高了預(yù)測(cè)結(jié)果的精度。 本文研究的主要結(jié)論有：第一，相比于其它預(yù)測(cè)模型，神經(jīng)網(wǎng)絡(luò)擁有更多的優(yōu)勢(shì)，它可以融合定性和定量?jī)深悢?shù)據(jù)，并且擁有很好的容錯(cuò)性和魯棒性，能對(duì)非線性函數(shù)有很強(qiáng)的映射能力，最后保證系統(tǒng)的大規(guī)模并行處理能力，提高輸出結(jié)果的速度和準(zhǔn)確性。第二，BP網(wǎng)絡(luò)模型的結(jié)構(gòu)設(shè)計(jì)和各參數(shù)選取盡量避免模型自身的缺陷，并結(jié)合所要研究的預(yù)測(cè)項(xiàng)目特點(diǎn)進(jìn)行細(xì)致的分析。第三，路網(wǎng)中如果有新建或者改建高速公路，就會(huì)改變?cè)瓉?lái)的路網(wǎng)結(jié)構(gòu)，對(duì)原有的高速公路就會(huì)產(chǎn)生很大的影響，轉(zhuǎn)移一部分原來(lái)公路上的車(chē)流量，這會(huì)使預(yù)測(cè)模型的預(yù)測(cè)結(jié)果出現(xiàn)偏差，因此為了增加預(yù)測(cè)結(jié)果的準(zhǔn)確性，必須對(duì)新建或改建高速公路的影響程度定量化進(jìn)行相應(yīng)的研究。
[Abstract]:With the deepening of China's reform and opening-up and the increasing demand for passenger and freight transport, it is not advisable to expand the expressway blindly, considering the problems of environment and cost, and putting forward a higher demand for the capacity of the built highway. Highway construction should be transformed from volume increase to rational planning and effective growth, so that unnecessary investment can be reduced. Therefore, the enterprise is required to accurately predict the traffic volume of the original highway, and take the forecast result as the basis of traffic planning and decision-making and the basis of the future profit forecast of the enterprise. The prediction of expressway traffic flow is a kind of long-term traffic flow prediction, and it is easy to be affected by various aspects of social environment. In order to improve the accuracy of prediction, it is necessary to choose a forecasting model which is more adaptable to the environment. The neural network model not only has the advantages of large-scale parallel processing, but also can ensure that the system can output reliable results faster, and it also has the characteristics of nonlinear mapping under the condition of analyzing a large number of related factors at the same time. This greatly enhances the ability of neural network model to adapt to the environment. Therefore, the neural network model can be used to predict the traffic volume of expressway accurately. This paper systematically combs the forecasting methods of expressway traffic flow at present, summarizes the advantages and disadvantages of different forecasting models, and constructs the forecasting model of expressway traffic flow based on BP neural network. Combined with the characteristics of expressway traffic data, the pretreatment method of sample data of expressway traffic flow and the excitation function of BP neural network prediction model are improved, and the initial values of each parameter in the prediction model are determined. At the same time, the method of quantifying the influence of newly built or expanded expressway on the forecast project is put forward, which improves the prediction value of the neural network model combined with the quantitative result of the influence degree, and improves the precision of the prediction result. The main conclusions of this paper are as follows: first, compared with other prediction models, neural network has more advantages, it can integrate qualitative and quantitative data, and has good fault tolerance and robustness. It can map the nonlinear function strongly. Finally, it can guarantee the large-scale parallel processing ability of the system, and improve the speed and accuracy of the output results. Secondly, the structure design of BP network model and the selection of each parameter avoid the defects of the model itself as far as possible, and the characteristics of the prediction items to be studied are analyzed in detail. Third, if there are new or rebuilt highways in the road network, it will change the structure of the original road network, which will have a great impact on the original highway and transfer a part of the traffic flow on the original highway. Therefore, in order to increase the accuracy of the prediction results, it is necessary to make a quantitative study on the influence degree of the newly built or rebuilt highways in order to increase the accuracy of the prediction results.
【學(xué)位授予單位】：武漢理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：U491.14
本文編號(hào)：2466963
[Abstract]:With the deepening of China's reform and opening-up and the increasing demand for passenger and freight transport, it is not advisable to expand the expressway blindly, considering the problems of environment and cost, and putting forward a higher demand for the capacity of the built highway. Highway construction should be transformed from volume increase to rational planning and effective growth, so that unnecessary investment can be reduced. Therefore, the enterprise is required to accurately predict the traffic volume of the original highway, and take the forecast result as the basis of traffic planning and decision-making and the basis of the future profit forecast of the enterprise. The prediction of expressway traffic flow is a kind of long-term traffic flow prediction, and it is easy to be affected by various aspects of social environment. In order to improve the accuracy of prediction, it is necessary to choose a forecasting model which is more adaptable to the environment. The neural network model not only has the advantages of large-scale parallel processing, but also can ensure that the system can output reliable results faster, and it also has the characteristics of nonlinear mapping under the condition of analyzing a large number of related factors at the same time. This greatly enhances the ability of neural network model to adapt to the environment. Therefore, the neural network model can be used to predict the traffic volume of expressway accurately. This paper systematically combs the forecasting methods of expressway traffic flow at present, summarizes the advantages and disadvantages of different forecasting models, and constructs the forecasting model of expressway traffic flow based on BP neural network. Combined with the characteristics of expressway traffic data, the pretreatment method of sample data of expressway traffic flow and the excitation function of BP neural network prediction model are improved, and the initial values of each parameter in the prediction model are determined. At the same time, the method of quantifying the influence of newly built or expanded expressway on the forecast project is put forward, which improves the prediction value of the neural network model combined with the quantitative result of the influence degree, and improves the precision of the prediction result. The main conclusions of this paper are as follows: first, compared with other prediction models, neural network has more advantages, it can integrate qualitative and quantitative data, and has good fault tolerance and robustness. It can map the nonlinear function strongly. Finally, it can guarantee the large-scale parallel processing ability of the system, and improve the speed and accuracy of the output results. Secondly, the structure design of BP network model and the selection of each parameter avoid the defects of the model itself as far as possible, and the characteristics of the prediction items to be studied are analyzed in detail. Third, if there are new or rebuilt highways in the road network, it will change the structure of the original road network, which will have a great impact on the original highway and transfer a part of the traffic flow on the original highway. Therefore, in order to increase the accuracy of the prediction results, it is necessary to make a quantitative study on the influence degree of the newly built or rebuilt highways in order to increase the accuracy of the prediction results.
【學(xué)位授予單位】：武漢理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：U491.14
【參考文獻(xiàn)】
相關(guān)期刊論文 前10條
1 王建軍,劉建超,陳寬民;公路建設(shè)項(xiàng)目交通需求預(yù)測(cè)與分析[J];重慶交通學(xué)院學(xué)報(bào);2004年01期
2 王正武,羅大庸,謝永彰;交通需求預(yù)測(cè)中不確定性的傳播分析[J];系統(tǒng)工程;2005年07期
3 朱從坤,馮煥煥;基于路段交通量的趨勢(shì)增長(zhǎng)——概率分配路網(wǎng)交通量預(yù)測(cè)方法[J];公路交通科技;2005年10期
4 張航;張玲;;基于重力模型預(yù)測(cè)誘增交通量方法研究[J];公路交通技術(shù);2006年01期
5 向前忠;;生長(zhǎng)曲線模型在高速公路誘增交通量預(yù)測(cè)中的應(yīng)用[J];公路交通技術(shù);2007年02期
6 李慶瑞;萬(wàn)發(fā)祥;盧毅;;公路交通量預(yù)測(cè)理論與方法綜述[J];中外公路;2005年06期
7 章錫俏;王守恒;孟祥海;;基于經(jīng)濟(jì)增長(zhǎng)的高速公路誘增交通量預(yù)測(cè)[J];哈爾濱工業(yè)大學(xué)學(xué)報(bào);2007年10期
8 單文勝;宋文;;淺談公路項(xiàng)目誘增和轉(zhuǎn)移交通量的預(yù)測(cè)方法[J];交通標(biāo)準(zhǔn)化;2006年12期
9 彭利人;王樹(shù)東;馮艷春;;公路交通量預(yù)測(cè)可靠性問(wèn)題研究[J];交通標(biāo)準(zhǔn)化;2008年08期
10 王延娟;;誘增交通量計(jì)算模型研究[J];交通標(biāo)準(zhǔn)化;2009年21期
本文編號(hào)：2466963
本文鏈接：http://sikaile.net/kejilunwen/jiaotonggongchenglunwen/2466963.html
最近更新
教材專著
