基于混沌小波与支持向量机的径流预测

以淮河流域沙颍河水系沙河和澧河上游的水库为例,建立了月均流量的混沌小波支持向量机组合预报模型,充分利用了混沌分析的相空间重构、小波分析的多分辨率功能以及支持向量机的非线性逼近能力,并采用NSE、PBIAS和RSR对组合预测模型进行了评价。结果表明:混沌小波支持向量机组合预测模型的识别期和验证期模拟精度均较高,均优于混沌支持向量机模型。     

改进支持向量机模型在城市需水量预测中的应用

以某城市需水量为研究对象,运用改进的支持向量基模型对该地区1991年到2001年的用水量进行模拟计算,并用该市2002年和2003年的用水量进行模型检验,与GM(1.1)模型所得的结果作比较,分析证明了改进的SVR模型方法能取得更好的结果。     

Improving Forecasting Accuracy of Streamflow Time Series Using Least Squares Support Vector Machine Coupled with Data-Preprocessing Techniques

Highly reliable forecasting of streamflow is essential in many water resources planning and management activities. Recently, least squares support vector machine (LSSVM) method has gained much attention in streamflow forecasting due to its ability to model complex non-linear relationships. However, LSSVM method belongs to black-box models, that is, this method is primarily based on measured data. In this paper, we attempt to improve the performance of LSSVM method from the aspect of data preprocessing by singular spectrum analysis (SSA) and discrete wavelet analysis (DWA). Kharjeguil and Ponel stations from Northern Iran are investigated with monthly streamflow data. The root mean square error (RMSE), mean absolute error (MAE), correlation coefficient (R) and coefficient of efficiency (CE) statistics are used as comparing criteria. The results indicate that both SSA and DWA can significantly improve the performance of forecasting model. However, DWA seems to be superior to SSA and able to estimate peak streamflow values more accurately. Thus, it can be recommended that LSSVM method coupled with DWA is more promising.     

基于支持向量机的中长期入库径流预报

采用基于支持向量机的预测模型对水库中长期入库径流进行预报,建立径流预报的SVM模型。预报因子的优劣决定着预测精度的高低。为了提高预报精度,尝试采用模糊优选法对预报因子进行优选。将所建模型应用于新疆雅马渡站的径流预测中,并与没有进行预报因子优选的SVM模型进行比较。结果表明,进行预报因子优化后的SVM模型明显提高了径流的预报精度,具有更好的应用价值。     

大坝监控小波去噪变权支撑向量回归机模型

为有效利用监控模型和指标判别大坝安全性态,实现大坝安全预警。首先利用小波分析对实测数据进行去噪处理,在此基础上利用支撑向量机小样本建模和高泛化能力,考虑不同幅度测值对大坝预警所起的作用不同,从而建立大坝监控变权支撑向量回归机模型。工程实例和理论分析表明,模型具有一定的理论和实用价值。     

An Estimation of the Suspended Sediment Load Using Adaptive Network Based Fuzzy Inference System,Support Vector Machine and Artificial Neural Network Models

Sediment transport in streams and rivers takes two forms as suspended load and bed load. Suspended load comprises sand + silt + clay-sized particles that are held in suspension due to the turbulence and will only settle when the stream velocity decreases, such as when the streambed becomes flatter, or the streamflow into a pond or lake. The sources of the suspended sediments are the sediments transported from the river basin by runoff or wind and the eroded sediments of the river bed and banks. Suspended-sediment load is a key indicator for assessing the effect of land use changes, water quality studies and engineering practices in watercourses. Measuring suspended sediment in streams is real sampling and the collection process is both complex and expensive. In recent years, artificial intelligence methods have been used as a predictor for hydrological phenomenon namely to estimate the amount of suspended sediment. In this paper the abilities of Support Vector Machine (SVM), Artificial Neural Networks (ANNs) and Adaptive Network Based Fuzzy Inference System (ANFIS) models among the artificial intelligence methods have been investigated to estimate the suspended sediment load (SSL) in Ispir Bridge gauging station on Coruh River (station number: 2316). Coruh River is located in the northern east part of Turkey and it is one of the world”s the fastest, the deepest and the largest rivers of the Coruh Basin. In this study, in order to estimate the suspended sediment load, different combinations of the streamflow and the SSL were used as the model inputs. Its results accuracy was compared with the results of conventional correlation coefficient analysis between input and output variables and the best combination was identified. Finally, in order to predict SSL, the SVM, ANFIS and various ANNs models were used. The reliability of SVM, ANFIS and ANN models were determined based on performance criteria such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Efficiency Coefficient (EC) and Determination Coefficient (R²).     

Medium and Long-term Precipitation Prediction Using Wavelet Decomposition-prediction-reconstruction Model

Water Resources Management - To improve the accuracy of medium and long-term precipitation prediction, we propose an innovative application of the wavelet decomposition-prediction-reconstruction...     

Estimation of Scour Downstream of a Ski-Jump Bucket Using Support Vector and M5 Model Tree

Estimation of scour downstream of a ski-jump bucket has been a topic of research among hydraulic engineers. For estimation of scour downstream of ski jump bucket, several empirical models are in use. In recent years, there has been emphasis to develop models which are capable of producing scour with high accuracy. Use of Artificial Neural Network (ANN) approach to model depth, width and length of scour hole indicates that performance of ANN models is far better than existing empirical models. At present, use of Support Vector Machines (SVMs) and M5 Pruned Model Tree are being considered in different disciplines to further improve upon the performance of ANN models as a potential alternate. With this in view, the present study deals with the development of regression models for computing various parameters of scour hole using SVMs and M5 Model Tree. A comparative evaluation of the performance of ANN versus SVMs and M5 Model Tree clearly shows that SVMs and M5 Model Tree can prove more useful than ANN models in estimation of scour downstream of a ski jump bucket. Further, M5 model tree offers explicit expressions for use by design engineers.     

Detecting Multi-Purpose Reservoir Operation Induced Time-Frequency Alteration Using Wavelet Transform

It is well recognized that natural flow variability is an inherent characteristic of rivers. Altered natural flow regime caused by anthropogenic regulations would threaten ecosystem biodiversity and deteriorate riverine health. Wavelet transform is a newly-developed tool that extracts dominant modes of variability by decomposing a non-stationary series into time-frequency space, which can be used to detect hydrologic alteration at various scales caused by reservoir operation. Continuous wavelet transform is simultaneously applied to recorded hourly inflow and outflow series of 1998–2008 for the Feitsui Reservoir located in northern Taiwan. Differences between wavelet power spectrum obtained for outflow and inflow series denote severity of hydrologic alteration. Greater spectral alteration is observed at less-than-1-day scales due to peak-load hydropower releases. The spectral alteration gradually declines with increasing scales. Different variation patterns for the yearly time-averaged spectral difference also reveal that the altered spectrum depends on hydrologic conditions. The index of spectral alteration (ISA), defined as the mean absolute deviations of power spectrum for all scales over a certain time period, is proposed to quantitatively assess severity of altered natural flow regime. ISA of 5 can be roughly recognized as the division of dry and non-dry years for the Feitsui Reservoir case. The obtained results offer decision makers useful information to adopt adaptive operating strategies to mitigate negative impacts of altered natural flow regime and derive optimal trade-off between human and environmental needs.     

相空间重构支持向量机在径流模拟中的应用研究

将相空间重构理论引入月径流模拟中,利用C-C算法进行相空间重构,将一维径流时间序列拓展为多维,基于交叉验证支持向量机(CV-SVM)原理及方法,构建以相空间重构理论与支持向量机相结合的径流时间序列模拟模型,并构建传统BP、双隐层BP及GA-BP径流时间序列模拟模型作为对比模型,以盘龙河龙潭寨月径流时间序列为例进行分析。结果显示:基于相空间重构理论的CV-SVM模拟模型能较好地处理复杂的径流序列,在长达200个月的测试样本模拟中,平均相对误差e MRE、最大相对误差e MaxRE分别为0.571 7%,5.526 7%,决定系数DC和合格率QR分别为0.999 9和100%。表明该模型具有较高的泛化能力和模拟精度,模拟效果明显优于传统BP、双隐层BP模型,甚至优于GA-BP模型;表明研究建立的基于相空间重构理论的CV-SVM模型用于径流模拟是合理可行的,可为径流模拟提供方法和参考。     

基于小波变换的楚雄市近50年降水序列多时间尺度研究

以滇中高原核心城市楚雄市为研究对象,分30个代表性气象站点1950—2009年降水时间序列,通过Morlet小波变换等方法对其进行多时间尺度研究,探讨其演变趋势和变化规律。结果表明楚雄市降水序列均呈较明显的增加趋势,对于各种时间尺度的降水序列均表现出一致的降水变化规律,传统干旱区干湿季节差异性缩小;年降水相关性最好,最大6、24 h降水量相关性非常高,短历时(即最大1、6、24 h)降水与中长时间尺度(即月、年)降水相关性最差,短历时降水对中长时间尺度降水的影响较小,区域使用的数据资料序列小于由气象资料数据序列的长度限制,说明周期性的可靠度不高,还需要进一步根据更长的序列资料进行验证。     

基于改进支持向量机回归的日径流预测模型

日径流预测是水资源优化调度的重要组成部分,日径流预测精度的高低直接影响水资源优化配置的程度。针对日径流序列的特性,研究提出一种改进的支持向量机回归模型,并应用于日径流预测。与基本支持向量机和BP神经网络对比分析的实验结果表明,基于改进支持向量机回归预测模型的日径流预测精度明显高于BP网络,尤其是对于变化剧烈的径流序列表现出较基本支持向量机回归模型更优越的预测性能,为日径流预测分析提供了一种可靠、有效的途径和方法。     

最小二乘支持向量机模型在大坝监测中的应用

提出了一种基于混沌时间序列的最小二乘支持向量机预测方法,该方法的依据是相空间重构技术以及最小二乘支持向量机模型(LS_SVM)。阐述了基于混沌时间序列的最小二乘支持向量机模型的算法步骤,并指出该模型的评价指标为平均绝对误差(MAE)以及预测均方误差(PMSE)。利用该模型对某混凝土大坝5#坝段102号测点的垂直位移进行了预测,结果表明:基于混沌时间序列的LS_SVM模型的预测性能较好,能够很好地体现出模型的实际应用能力;模型的拟合及预报结果能够满足精度要求,与回归模型相比具有预测结果精度较高的优点。     

基于支持向量机的冻融土壤入渗参数预测

以提高冻融条件下农田土壤灌溉水利用率为目的,基于汾河灌区冻融期内土壤入渗试验,运用支持向量机理论,建立了以冻融条件下土壤含水率、质地、容重、地温和灌溉水温为输入变量,Kostiakov入渗模型参数k、α为输出变量的支持向量机回归预测模型。预测模型误差分析表明:入渗系数k与入渗指数α的训练样本相对误差与检验样本相对误差都控制在10%以内,将k与α代入Kostiakov入渗模型得到的90 min累计入渗量I_(90)的相对误差平均值为5.81%。说明冻融土壤条件下,基于支持向量机理论的Kostiakov入渗模型参数的预测是可行的。     

基于小波的宝兴河流域降雨序列近似周期分析

根据宝兴河流域控制水文站宝兴站1958~2007年的降雨资料,采用Morlet小波对其降雨变化的近似周期成分进行分析,结果表明,宝兴河降雨序列具有多时间尺度的变化,具有4年和10年左右的近似周期变化;同时,研究还发现,宝兴河降雨序列变化还存在约10年左右的丰枯交替过程,其中1958~1966年、1975~1980年、1985~1990年、1995~1997和2000~2005年为多水期,其余时段为枯水期,在进入20世纪90年代以后,其丰枯交替现象不甚明显,整个流域的降水量呈现出比较平缓的趋势。     

基于LSSVM的短时风电功率预测

风电功率预测对于风电接入电网后上网限电有非常重要作用,同时也对风电的市场竞争力有很大的支持效应。从现阶段风电功率负荷预测的现状出发,在研究当前风速预测方法和预测效果的基础上提出用LSSVM来进行风速的预测方法,与其他几种风速预测方法的误差进行比较表明,LSSVM在预测风速方面具有一定的优越性。经过实例测算表明,效果较为理想。     

基于进化相关向量机的边坡安全系数估算

针对相关向量机与差分进化优化算法的特点,通过将两种算法有机融合提出进化相关向量机模型E-RVM,并应用于边坡安全系数估算。以留一交叉验证法构建差分进化算法的适应度函数,基于差分优化算法确定相关向量机的最优参数,可有效提高算法预测精度及可靠性。根据模型计算得到的预测均值及预测方差建立预测变量置信区间,分析预测结果的不确定性。以两个边坡数据为例建立基于E-RVM的边坡安全系数估算模型,并与GA-BP、V-SVM、GP方法对比。分析结果表明:E-RVM方法的平均绝对误差、平均相对误差与均方根误差精度指标均明显优于GA-BP、V-SVM、GP。通过95%置信度的置信区间分析,理论安全系数均在置信区间内,并且E-RVM方法具有比GP方法更短的置信区间长度。分析证实E-RVM模型是一种精度高、可靠性强的边坡安全系数预测新方法。     

基于支持向量机方法的水电机组状态识别

模式识别理论是水电机组状态识别与故障诊断的基础,通过选择不同类型的分类器来实现模式识别。支持向量机(SVM)算法有效地改善了传统分类方法的缺陷,具有较强理论依据,适合小样本的模式识别分类。结合水电机组振动信号特点,采用Hilbert-Huang变换对信号做前期处理,以IMF函数能量值为基本元素构造特征向量,将特征向量输入SVM进行模式识别。结果表明对于小样本分类,SVM方法简单有效,切实可用。     

River Stage Forecasting Using Wavelet Packet Decomposition and Machine Learning Models

Water Resources Management - This study develops and applies three hybrid models, including wavelet packet-artificial neural network (WPANN), wavelet packet-adaptive neuro-fuzzy inference system...