【作者】 陈良；

【导师】 李朝奎；

【作者基本信息】 湖南科技大学 ， 地图学与地理信息系统， 2010， 硕士

【摘要】 随着城市的发展和道路基础设施的逐步完善,以及人们对环保意识的增强和对环境质量要求的提高,环境噪声问题已成为越来越受关注的环境问题。因此在项目建设之前进行环境噪声的预测和评估是十分必要的,它对环境污染控制、管理、规划等方面均具有重要意义。住宅小区是人们日常活动的最主要场所,但是却受到越来越多噪声的干扰。对于规划中的项目,我们希望知道其建成后的噪声分布情况,需要进行噪声预测。本文主要学习了通用的噪声预测模型FHWA模式和神经网络的基本原理,选择了BP神经网络作为建立噪声预测模型的基本理论和方法,尝试性的将BP神经网络用于创建噪声预测模型,并且进行了神经网络模型的优化。以某一小区的设计方案为例,运用该噪声预测模型对其进行噪声预测,比较两种模式下的预测精度,经过分析预测结果可知预测结果能较好的模拟该设计方案落成后的噪声分布情况。将噪声预测模型与GIS专业软件相结合,较为直观和逼真的显示了预测结果,根据预测结果调整设计方案,从而到达辅助规划决策的目的。论文的创新点在于:(1)用BP神经网络建立了一个面向小区域的噪声预测模型,补充并发展了噪声预测的新方法;(2)该噪声预测模型与3DCM相结合,在建筑群中对声场进行非线性建模。实验证明基于神经网络的噪声预测模型具有自组织、全局优化与容错学习等能力,通过样本学习可使预测模型有良好的预测能力;噪声预测模型与3DCM的结合,模拟了噪声在建筑群中的分布,更好地将噪声信息进行了表达和分析应用,为政府部门进行城市规划、管理及决策提供科学有效的证据支持。更多还原

【Abstract】 With the urban development and the improvement of road infrastructure gradually, as well as people’s awareness of environmental protection and improvement of environmental quality, environmental noise has become more and more affected by environmental concerns. Therefore, it is necessary to predict and evaluate the environmental noise before the project construction, and it’s of great significance for environmental pollution control, management and planning. etc.Residential area is important for people’s daily activities, but it is suffered by more and more noise. After completion of the planning project, we want to know the noise distribution and predict noise. According to the predictions, analyze spatial distribution, time distribution and the probability distribution characteristics of the noise objectively and effectively with help of computer simulation technology, and adjust the design based on the Urban Regional Environmental Noise Standards, so as to control the noise at Planning Stage effectively, Further more, the purpose of optimizing residential environment can be achieved. There are many problems of noise prediction models though each has its own advantages on specific applications. This paper uses the advantages of the neural network toestablish a prediction model based on the BP neural network. Innovation of this paper are: (1) Base on BP neural network to establishment a noise prediction model for small area, complement and developed a new method of noise prediction; (2) the noise prediction model combined with 3DCM , complex the sound field on nonlinear modeling. Experimental results shows that based on neural network noise prediction model has self-organization, global optimization and fault-tolerant learning abilities, neural networks can learn through the sample noise in the buildings of the acoustic field in non-linear modeling, which has a good predictive ability; The noise prediction model combined with 3DCM is a good simulation of noise in buildings in the distribution. Prediction of noise based on neural network combine with three-dimensional city models (3DCM), is better for information expression and analysis applications, offer scientific and effective supporting evidence to government departments for urban planning, management and decision-making.更多还原