BP神经网络在上海住宅市场需求预测中的应用

人工神经网络是近期发展最快的人工智能领域研究成果之一 ,本文在介绍 BP神经网络的有关原理的基础上 ,建立了一个上海住宅市场的 BP神经网络模型 ,并通过该模型对上海住宅市场的需求进行了预测和分析 .分析结果表明人工神经网络方法在住宅市场需求预测中的应用是可行的并且是有效的 .     

A RAPID APPROXIMATION FOR PREDICTING THE HYDROCARBON DISCOVERY RATE: PART 1. ASSESSING THE ACCURACY

The hydrocarbon discovery prediction problem is important to firms having to make decisions about the deployment of scarce exploration resources. Traditional methods for estimating the discovery rate rely on the completion of time consuming simulation experiments. A rapid approximation that does not require the completion of simulation exists and has been shown to have some promise as a prediction tool. This paper investigates the accuracy of the approximation method under a wide variety of distributional and drilling efficiency assumptions. The results indicate that the approximation produces predictions close to those of simulation under most of the tested conditions. This suggests that resource exploration firms could conveniently use the method for a wide variety of planning purposes without incurring the same costs in time and personnel required for simulation.     

基于多Agent的宏观经济智能预测决策支持系统研究

在分析智能决策支持系统的发展基础上，将人工智能思想，特别是Agent应用于构建智能宏观经济预测决策支持系统。提出面向任务的Agent设计思想，以任务为核心设计多种类型Agent共同完成预测决策任务，给出基于多Agent的系统组成结构，并给出系统中不同类型Agent的具体实现结构，同时探讨了整个系统的运行机制以及系统中多Agent之间的合作。整个系统的构建过程也是面向Agent的程序设计的过程。     

基于新型核与灰度序列的时滞GM(1,N)模型及其应用

为了解决GM(1,N)模型在新型核与灰度的基础上,对驱动项的延迟作用机理不明确的问题,将时滞参数引入到GM(1,N)模型的驱动项中,构建了基于新型核与灰度的时滞GM(1,N)模型,分析了时滞参数的辨识方法,讨论了新模型的建模机理。为了更好地对该模型的有效性进行验证,将优化的时滞GM(1,N)模型对南京市的雾霾进行预测分析,选择GM(1,N)模型、一元回归模型与文中的优化模型进行对比。结果显示,优化模型对PM₁₀浓度的拟合精度更高,且误差均控制在5%之内,从而验证了提出的优化模型适用于具有时滞特征数据的模拟和预测。     

Correlation Analysis of the Runoff of the Rivers and the Seisms in China

In recent hundred years the annual discharge variations of the Changjiang River (represented by the Yichang station) and the Huanghe River (represented by Shanxian and Tangnaihe, respectively) have closely related to the geographical distribution of the earthquakes coming about in China in the same year, Both the occurrence of the destructive seism or seismic swarm in the river basins and the disappearance of the shocks in the east and south of the basins are the conditions that the Changjiang and Huanghe Rivers are the high flow while that the strong earthquake of magnitude 7 or more occurred in North China is the condition for the Changjiang low flow year and that of 6 or more in the Qilian Mountains area is for the Huanghe River. In the latter part of this paper, a 2-year sample is given to explain that the conditions of the 2 rivers being high flow years are that the north-south seismic belt is active and in the meanwhile no seism occurred in South China, and those of the low flow year are that the     

灰色Verhulst神经网络组合预测区域空域流量

能够对区域空域的流量做出准确的预测,可以为空中交通管理部门协调航班运行,减缓空域内航路的拥堵提供重要依据.运用灰色Verhulst模型与神经网络的组合模型对区域航班流量进行预测,并结合华东地区飞行流量的数据样本训练并测试模型.结果显示,由组合模型得出的流量预测值与实际值较吻合,说明组合模型比单种预测方法能够更为准确地对区域空域流量做出预测.     

QuEChERS and solid phase extraction methods for the determination of energy crop pesticides in soil,plant and runoff water matrices

QuEChERS and solid phase extraction (SPE) methods were applied for determining four herbicides (metazachlor, oxyfluorfen, quizalofop-p-ethyl, quinmerac) and one insecticide (α(±)-cypermethrin) in runoff water, soil, sunflower and oilseed rape plant matrices. Determination was performed using gas chromatography mass spectrometry (GC-MS), whereas high-pressure liquid chromatography mass spectrometry (HPLC-MS) was used for quinmerac. In all substrates linearity was evaluated using matrix-matched calibration samples at five concentration levels (50–1000 ng L^?1 for water, 5–500 μg kg^?1 for soil and 2.5–500 μg kg^?1 for sunflower or oilseed rape plant). Correlation coefficient was higher than 0.992 for all pesticides in all substrates. Acceptable mean recovery values were obtained for all pesticides in water (65.4–108.8%), soil (70.0–110.0%) and plant (66.1–118.6%), with intra- and inter-day RSD% below 20%. LODs were in the range of 0.250–26.6 ng L^?1 for water, 0.10–1.8 μg kg^?1 for soil and 0.15–2.0 μg kg^?1 for plants. The methods can be efficiently applied for field dissipation studies of the pesticides in energy crop cultivations.     

基于改进极根学习机的回转窑煅烧带温度预测方法

针对传统算法预测回转窑煅烧带温度存在精度低、速度慢的问题,提出了基于改进极限学习机(ELM)的回转窑煅烧带温度预测方法。对ELM输入权值矩阵定义了变换系数,采用黄金分割法在给定区间内搜寻变换系数的最佳值,改进了ELM网络参数的确定方式,弥补了随机确定输入权值并且不作调整的缺陷,在保证ELM训练速度的前提下提高预测精度、减小模型随机性。实验结果表明,改进的ELM预测精度高、训练速度快、模型性能优,可满足工况恶劣的回转窑的生产需要。     

基于Chebyshev多项式的神经网络中长期负荷预测研究

高精度负荷预测在提高电力系统的安全性和经济性方面有着极其重要的意义,而现有的负荷预测方法因参数有限,难以完全反映其内在规律,因而导致预测结果不够准确.为此提出了一种基于Chebyshev多项式神经网络模型的预测方法.该方法使用递推最小二乘法训练神经网络权值系数,以获得高精度的参数估计,从而实现Chebyshev多项式神经网络模型对负荷量的最优拟合,再利用训练好的Chebyshev多项式神经网络模型实现中长期负荷预测.研究结果表明,该方法能较好模拟负荷变化规律,有效提高了负荷预测精度,在电力系统负荷预测中有较大的应用价值.     

广东省能源需求预测模型构建及实证分析

为了对广东省的能源需求进行准确的预测,首先分析了影响广东省能源需求的各种因素,构建了预测指标体系.在此基础上,针对能源系统非线性等复杂系统特征,结合粒子群算法和BP神经网络的优点,构建了改进的PSO-BP神经网络的预测模型,并通过主成分分析法对指标体系进行数据降维,以降低神经网络的规模和复杂程度.以广东省1985-2013年的能源需求数据进行模拟与仿真,并对2014-2018年的能源需求量进行预测,理论分析和实证研究表明,该方法能够很好的反映广东省能源需求的特征,预测结果较为准确合理.