【作者】 秦元营；

【导师】 孔永健； 武云鹏；

【作者基本信息】 北京交通大学 ， 道路与铁道工程， 2007， 硕士

【摘要】 交通荷载的动态预测是进行路面设计和评价的前提,其预测的正确与否将直接影响到设计和评价的合理性。交通荷载的变化受各种因素的影响,特别是随着时间的推移,交通荷载将呈现出动态变化的趋势。事物之间存在其内部固有的规律,历年的交通荷载变化也应有一定的规律。这种规律可能不是一般的规律,它有其复杂性。在以往的交通量预测方法中都对这种规律进行了研究,但效果并不是特别好。人工智能技术的出现给预测方法的改进提供了更好的理论。神经网络使预测的精度更进一步提高,但它本身也有缺点。为了弥补这些缺点,本文提出了将遗传算法和神经网络相结合,通过遗传算法对神经网络进行优化,利用历年的交通荷载量对网络进行训练。优化后的神经网络能更好的反映出交通荷载量之间的内部关系。利用经过优化、训练好的神经网络再进行预测,就能得到理想的效果。本文采用人工智能的方法开发路面交通荷载预测模型。针对预测系统的非线性,采用将遗传算法与BP网络相结合的算法。利用遗传算法全局寻优的特点来优化BP网络权值和阙值,从而克服了BP网络固有的缺陷。用这样一个较优的GA-BP模型对路面交通荷载进行预测。采用MATLAB语言对所建模型进行仿真,从而验证了模型的可行性。将GA-BP算法与已有的其他算法进行比较,得出GA-BP算法的性能优越性。取河南省1988～1993年的平均交通荷载量进行训练,用1994年的平均交通荷载量进行模型检验。对比各种预测方法的预测结果,得出GA-BP预测的准确性。更多还原

【Abstract】 The dynamic estimate of the transportation load is the premise that carries on the road design and evaluation, whether what it predict is right or not will influence directly to design and evaluate of rationality.The transportation load is under the influence of various factor.Especially along with the change of time, the transportation load will present a dynamic state.All things have its proper regulation. The change of the transportation load that is through the years’ should also have certain regulation.This kind of regulation is not a general regulation, it is a complex. The former traffic volumes predict methods have carried on a research on this regulation, but the result isn’t especially good.The technology of Artificial intelligence provide better theories for improvement of the predict method.The accuracy of the predict is improved by the nerve network. But it also has weakness.For making up these weaknesses, this paper put forward combining together the heredity arithmetic way and the nerve network, using the heredity arithmetic way to optimize the nerve network .The nerve network optimized by the heredity arithmetic is better to reflect the internal relation of the transportation load.Make use of the nerve network which has optimized and trained to predict, we can get ideal result.This paper create the estimate model with the method of the artificial intelligence.Aiming at the nonlinear estimate system, this paper adopt the heredity arithmetic to improve the BP network.The heredity arithmetic way can looks for an excellent characteristics overall situation. Make use of this excellence to optimize the weight and value of the BP network. It can overcome the inherent shortcomings of BP network .Then Use this more excellent GA-BP model to predict the transportation load.This paper adopt the MATLAB language to emulate the model which has set up and verifiy the practicability of the model.By contrast with other predict methods,we get the GA-BP calculate way is superiority.This paper use the average transportation load data from 1988 to 1993 of henan province to train the model, then prove its by use the transportation load data of 1994 . Contrasting with the result of other different predit methods,the GA-BP can get a accurate result.更多还原