基于神经网络集成的单个飞行事件噪声预测模型

通过分析影响单个飞行事件噪声的各种因素,构建了BP神经网络回归预测模型,并通过自适应遗传算法优选出参与集成的个体神经网络,提出了预测单个飞行事件噪声的神经网络集成预测模型.为了有效保证差异性,设置不同隐藏神经元个数和Bagging算法来构建和训练单个网络.实验结果表明,单个飞行事件噪声的神经网络集成预测模型相对单个BP神经网络模型泛化能力更强,稳定性能更好.本文方法在测试集上误差在3dB以内的平均比率为96.9%,比单个网络高6.8%.     

车库出入口坡道噪声影响因素及预测

选取城市居住区中典型地下车库,在优化其出入口噪声采样点布置和同步采样点分组基础上,通过对噪声样本的声学分析,研究了车库出入口坡道噪声与坡道结构、坡度、线形、上下坡等相关因素的关系.通过对平直混凝土路面车库出入口噪声影响预测值的修正,建立了车库出入口坡道噪声影响预测模式.     

A modeling system for drinking water sources and its application to Jiangdong Reservoir in Xiamen city

Drinking water sources are highly valued by authorities for safeguarding the life of a city. Models are widely applied as important and effective tools in the management of water sources. However, it is difficult to apply models in water source management because water managers are often not equipped with the professional knowledge and operational skills necessary for making use of the models. This paper introduces a drinking water source simulation and prediction system that consists of a watershed model, a hydrological model and a water quality model. This system provides methods and technical guidance for the conventional management of water sources and emergency water event response. In this study, the sub-models of the system were developed based on the data of the Jiangdong Reservoir in Xiamen, and the model validation was based on local monitoring data. The hydrological model and water quality model were integrated by computer programming, and the watershed model was indirectly integrated into the system through a network platform. Furthermore, three applications for Jiangdong Reservoir water protection utilizing the system were introduced in this paper, including a conventional simulation, an emergency simulation, and an emergency measures evaluation.     

基于AGA-BP神经网络的采空区危险性评价

针对采空区危险性评价的影响因素众多且关系复杂的特点,提出了基于AGA-BP神经网络算法评价采空区危险性。将岩体结构、地质构造、岩石抗压强度等13个影响因子作为神经网络输入,采空区危险性等级作为输出,建立一个采空区危险性评价的BP神经网络模型;采用自适应遗传算法(AGA)对BP网络的初始权值和阈值进行全局寻优,将寻优结果回代入网络中进行训练并预测得出采空区危险性等级;利用其它智能算法与该预测结果做出比较,以验证AGA-BP算法的有效性及优越性。结果表明:该算法的优化效果明显,同时在训练时间与预测精度上较其它智能算法有突出的优势,是一种在采空区危险性评价方面值得推广的新方法。     

煤矿瓦斯涌出时序预测的自组织数据挖掘方法

为分析煤矿瓦斯涌出复杂系统时间序列预测方法,提出自组织数据挖掘(SODM)与相空间重构(PSR)相结合的预测建模方法。首先应用C-C方法计算时间序列的最佳嵌入维数和延迟时间后进行PSR;然后以二元二次方程为传递函数,以嵌入维数变量为自变量,以延迟时间后的时间序列为因变量,通过内准则确定传递函数系数和外准则选择最优传递函数,并以最优传递函数的输出为下层迭代传递函数的输入,最后获得最优复杂度预测模型。算例结果表明:该方法对煤矿瓦斯涌出量预测的相对误差为-5.751 7% ~6.049 3%,平均相对误差2.145 7%,预测结果能满足煤矿安全生产实际工程应用要求。     

基于Box Behnken设计的煤导热系数分析与预测

为研究分析水分、固定碳及密度3个因素对煤导热系数的影响,选取多个矿井不同煤质的煤作为实验煤样,使用NETZSCH LFA457型激光导热系数测试仪进行测定,并引入二次响应曲面试验设计思想,采用Design Expert软件进行煤质指标的Box Behnken试验设计,按设计表筛选实验数据,构建了导热系数多因素、多水平影响下二次回归响应曲面模型,并对不同采样地点的煤导热系数进行了分析与预测。结果表明:二次回归方程P值小于0.000 1,极显著,失拟项为0.072 4,不显著,回归方程具有统计学意义;当水分在0.9%~10.9%,固定碳在49.3%~83.6%,密度在1.2~1.92 g·cm³之间,对煤的导热系数一次项影响程度为密度>固定碳>水分;二次项影响程度为固定碳和密度>水分和密度>水分和固定碳,且固定碳和密度之间存在交互作用,水分和密度、水分和固定碳之间无交互作用;应用响应曲面模型预测的煤导热系数误差为4.3%,满足精度要求。     

重庆大气汞人为排放及其预测

利用排放因子法和灰色预测法对重庆大气汞人为排放进行了研究. 重庆大气汞人为排放量从1997年的6.18 t增加到2008年的13.47 t. 重庆自1997年成为直辖市以来,大气汞排放量累计达99.76 t,年均增长率为9.82%,其中燃煤和水泥生产汞排放量分别达58.34和22.37 t. 以1997─2008年重庆大气汞人为排放数据为依据,运用灰色系统理论建立GM(1,1)大气汞人为排放预测模型,预测了2009─2015年重庆大气汞排放趋势. 如果不采取控制措施,预计2015年重庆大气汞人为排放量约30.92 t,年均增长率将高达16.20%.      

Analysis of the global warming dynamics from temperature time series

Global warming is the observed increase of the average temperature of the Earth. The primary cause of this phenomenon is the release of the greenhouse gases by burning of fossil fuels, land cleaning, agriculture, among others, leading to the increase of the so-called greenhouse effect. An approach to deal with this important problem is the time series analysis. In this regard, different techniques can be applied to evaluate the global warming dynamics. This kind of analysis allows one to make better predictions increasing our comprehension of the phenomenon. This article applies nonlinear tools to analyze temperature time series establishing state space reconstruction and prediction. Since noise contamination is unavoidable in data acquisition, it is important to employ robust techniques. The method of delay coordinates is employed for state space reconstruction and delay parameters are evaluated using the method of average mutual information and the method of false nearest neighbors. Afterwards, the simple nonlinear prediction method is employed to estimate temperatures of the future. Temperature time series from different places of the planet are used. Initially, the approach is verified considering known parts of the time series and afterwards, results are extrapolated for future values estimating temperature until 2028. Results show that these techniques are interesting to estimate temperature time history, presenting coherent estimations.     

西安市生态文明建设评价及预测

根据生态文明的涵义和生态市建设的总体思路,提出了生态文明建设3（系统层、子系统层、单项指标层）×3（经济发展,社会进步,环境保护）评价体系,其中包括25个单项指标。对西安市1998-2007年生态文明建设进行了定量评价研究,结果表明西安市生态文明建设处于整体上升的趋势;在此基础上,应用自回归预测模型对西安市经济、社会、环境和生态文明建设发展进行了预测,为西安市全面协调发展提供参考借鉴。