共查询到20条相似文献，搜索用时 15 毫秒
1.
Application of support vector machine and relevance vector machine to determine evaporative losses in reservoirs 总被引：1，自引：0，他引：1
This article employs Support Vector Machine (SVM) and Relevance Vector Machine (RVM) for prediction of Evaporation Losses (E) in reservoirs. SVM that is firmly based on the theory of statistical learning theory, uses regression technique by introducing ε‐insensitive loss function has been adopted. RVM is based on a Bayesian formulation of a linear model with an appropriate prior that results in a sparse representation. The input of SVM and RVM models are mean air temperature (T) ( °C), average wind speed (WS) (m/sec), sunshine hours (SH)(hrs/day), and mean relative humidity (RH) (%). Equations have been also developed for prediction of E. The developed RVM model gives variance of the predicted E. A comparative study has also been presented between SVM, RVM and ANN models. The results indicate that the developed SVM and RVM can be used as a practical tool for prediction of E. Copyright © 2011 John Wiley & Sons, Ltd. 相似文献
2.
Ravinesh C. Deo Pijush Samui Dookie Kim 《Stochastic Environmental Research and Risk Assessment (SERRA)》2016,30(6):1769-1784
The forecasting of evaporative loss (E) is vital for water resource management and understanding of hydrological process for farming practices, ecosystem management and hydrologic engineering. This study has developed three machine learning algorithms, namely the relevance vector machine (RVM), extreme learning machine (ELM) and multivariate adaptive regression spline (MARS) for the prediction of E using five predictor variables, incident solar radiation (S), maximum temperature (T max), minimum temperature (T min), atmospheric vapor pressure (VP) and precipitation (P). The RVM model is based on the Bayesian formulation of a linear model with appropriate prior that results in sparse representations. The ELM model is computationally efficient algorithm based on Single Layer Feedforward Neural Network with hidden neurons that randomly choose input weights and the MARS model is built on flexible regression algorithm that generally divides solution space into intervals of predictor variables and fits splines (basis functions) to each interval. By utilizing random sampling process, the predictor data were partitioned into the training phase (70 % of data) and testing phase (remainder 30 %). The equations for the prediction of monthly E were formulated. The RVM model was devised using the radial basis function, while the ELM model comprised of 5 inputs and 10 hidden neurons and used the radial basis activation function, and the MARS model utilized 15 basis functions. The decomposition of variance among the predictor dataset of the MARS model yielded the largest magnitude of the Generalized Cross Validation statistic (≈0.03) when the T max was used as an input, followed by the relatively lower value (≈0.028, 0.019) for inputs defined by the S and VP. This confirmed that the prediction of E utilized the largest contributions of the predictive features from the T max, verified emphatically by sensitivity analysis test. The model performance statistics yielded correlation coefficients of 0.979 (RVM), 0.977 (ELM) and 0.974 (MARS), Root-Mean-Square-Errors of 9.306, 9.714 and 10.457 and Mean-Absolute-Error of 0.034, 0.035 and 0.038. Despite the small differences in the overall prediction skill, the RVM model appeared to be more accurate in prediction of E. It is therefore advocated that the RVM model can be employed as a promising machine learning tool for the prediction of evaporative loss. 相似文献
3.
At the early stages of deep-water oil exploration and development, fewer and further apart wells are drilled than in onshore oilfields. Supervised least squares support vector machine algorithms are used to predict the reservoir parameters but the prediction accuracy is low. We combined the least squares support vector machine (LSSVM) algorithm with semi-supervised learning and established a semi-supervised regression model, which we call the semi-supervised least squares support vector machine (SLSSVM) model. The iterative matrix inversion is also introduced to improve the training ability and training time of the model. We use the UCI data to test the generalization of a semi-supervised and a supervised LSSVM models. The test results suggest that the generalization performance of the LSSVM model greatly improves and with decreasing training samples the generalization performance is better. Moreover, for small-sample models, the SLSSVM method has higher precision than the semi-supervised K-nearest neighbor (SKNN) method. The new semisupervised LSSVM algorithm was used to predict the distribution of porosity and sandstone in the Jingzhou study area. 相似文献
4.
Ani Shabri 《水文科学杂志》2013,58(7):1275-1293
Abstract This paper investigates the ability of a least-squares support vector machine (LSSVM) model to improve the accuracy of streamflow forecasting. Cross-validation and grid-search methods are used to automatically determine the LSSVM parameters in the forecasting process. To assess the effectiveness of this model, monthly streamflow records from two stations, Tg Tulang and Tg Rambutan of the Kinta River in Perak, Peninsular Malaysia, were used as case studies. The performance of the LSSVM model is compared with the conventional statistical autoregressive integrated moving average (ARIMA), the artificial neural network (ANN) and support vector machine (SVM) models using various statistical measures. The results of the comparison indicate that the LSSVM model is a useful tool and a promising new method for streamflow forecasting. Editor D. Koutsoyiannis; Associate editor L. See Citation Shabri, A. and Suhartono, 2012. Streamflow forecasting using least-squares support vector machines. Hydrological Sciences Journal, 57 (7), 1275–1293. 相似文献
5.
使用最小二乘支持向量机分类方法建立了两个砂土液化预测模型,预测结果与野外实际情况全部相符,表明该分类方法用于预测砂土液化是可行的,且预测准确率高。 相似文献
6.
ABSTRACTUnderstanding streamflow patterns by incorporating climate signal information can contribute remarkably to the knowledge of future local environmental flows. Three machine learning models, the multivariate adaptive regression splines (MARS), the M5 Model Tree and the least squares support vector machine (LSSVM) are established to predict the streamflow pattern over the Mediterranean region of Turkey (Besiri and Baykan stations). The structure of the predictive models is built using synoptic-scale climate signal information and river flow data from antecedent records. The predictive models are evaluated and assessed using quantitative and graphical statistics. The correlation analysis demonstrates that the North Pacific (NP) and the East Central Tropical Pacific Sea Surface Temperature (Niño3.4) indices have a substantial influence on the streamflow patterns, in addition to the historical information obtained from the river flow data. The model results reveal the utility of the LSSVM model over the other models through incorporating climate signal information for modelling streamflow. 相似文献
7.
The determination of seismic attenuation (s) (dB/cm) is a challenging task in earthquake science. This article employs genetic programming (GP) and minimax probability machine regression (MPMR) for prediction of s. GP is developed based on genetic algorithm. MPMR maximizes the minimum probability of future predictions being within some bound of the true regression function. Porosity (n) (%), permeability (k) (millidarcy), grain size (d) (μm), and clay content (c) (%) have been considered as inputs of GP and MPMR. The output of GP and MPMR is s. The developed GP gives an equation for prediction of s. The results of GP and MPMR have been compared with the artificial neural network. This article gives robust models based on GP and MPMR for prediction of s. 相似文献
8.
S. S. Kevorkyants 《Izvestiya Physics of the Solid Earth》2006,42(8):702-711
Forward seismic problems are solved for elastic media by rigorous methods (i.e., methods with controllable accuracy). Analysis of the current state of research on this subject suggests that the most promising methods are based on integral and integro-differential equations, notwithstanding the rather modest results of their application to solving forward problems in the theory of elastic vibrations. The second Green integral theorem for seismic waves, formulated and proven in the paper, yields a system of two boundary (surface) integral equations for the displacement vector u(M 0) and the normal (to the boundary surface) vector component of the stress tensor tn(M 0). The integrands of the surface integrals in terms of which the function tn(M 0) is expressed on both sides of the interface between the medium and the heterogeneity contain the second derivatives of the Green’s tensor functions ? e (M 0, M) and ? i (M 0, M), respectively, which are responsible for a cubic singularity (third-order singularity) if the integration point M coincides with the observation point M 0. An original method of eliminating the cubic singularity proposed in the paper involves special tensor normalization of the integrals on the outer and inner sides of the interface and subsequent subtraction of one integral from another in order to construct the second integral equation. 相似文献
9.
Evaluation of liquefaction potential of soil based on standard penetration test using multi-gene genetic programming model 总被引：1，自引：0，他引：1
This paper discusses the evaluation of liquefaction potential of soil based on standard penetration test (SPT) dataset using evolutionary artificial intelligence technique, multi-gene genetic programming (MGGP). The liquefaction classification accuracy (94.19%) of the developed liquefaction index (LI) model is found to be better than that of available artificial neural network (ANN) model (88.37%) and at par with the available support vector machine (SVM) model (94.19%) on the basis of the testing data. Further, an empirical equation is presented using MGGP to approximate the unknown limit state function representing the cyclic resistance ratio (CRR) of soil based on developed LI model. Using an independent database of 227 cases, the overall rates of successful prediction of occurrence of liquefaction and non-liquefaction are found to be 87, 86, and 84% by the developed MGGP based model, available ANN and the statistical models, respectively, on the basis of calculated factor of safety (F s ) against the liquefaction occurrence. 相似文献
10.
The catalogue by Grünthal et al. (J Seismol 13:517?C541, 2009a) of earthquakes in central, northern, and north-western Europe with M w????3.5 (CENEC) has been expanded to cover also southern Europe and the Mediterranean area. It has also been extended in time (1000?C2006). Due to the strongly increased seismicity in the new area, the threshold for events south of the latitude 44°N has here been set at M w????4.0, keeping the lower threshold in the northern catalogue part. This part has been updated with data from new and revised national and regional catalogues. The new Euro-Mediterranean Earthquake Catalogue (EMEC) is based on data from some 80 domestic catalogues and data files and over 100 special studies. Available original M w and M 0 data have been introduced. The analysis largely followed the lines of the Grünthal et al. (J Seismol 13:517?C541, 2009a) study, i.e., fake and duplicate events were identified and removed, polygons were specified within each of which one or more of the catalogues or data files have validity, and existing magnitudes and intensities were converted to M w. Algorithms to compute M w are based on relations provided locally, or more commonly on those derived by Grünthal et al. (J Seismol 13:517?C541, 2009a) or in the present study. The homogeneity of EMEC with respect to M w for the different constituents was investigated and improved where feasible. EMEC contains entries of some 45,000 earthquakes. For each event, the date, time, location (including focal depth if available), intensity I 0 (if given in the original catalogue), magnitude M w (with uncertainty when given), and source (catalogue or special study) are presented. Besides the main EMEC catalogue, large events before year 1000 in the SE part of the investigated area and fake events, respectively, are given in separate lists. 相似文献
11.
The risk formula, expressing the probability of at least one occurrence of earthquakes of greater-than-design-value magnitudes over the economic life of a structure, is modified taking into consideration the probability of no-earthquake years. The annual maximum earthquake magnitudes of three scales: Richter magnitude, also known as local magnitude (ML), body-wave magnitude (Mb), and moment magnitude (MM) in a geographical area encompassing the Bingöl Province in Turkey are taken from two sources: (1) report by Kalafat et al. (2007) [14] and (2) the web site reporting data by Kandilli Observatory which has been recording earthquakes occurring in and around Turkey since 1900. Statistical frequency analyses are applied on the three sample series using various probability distribution models, and magnitude versus average return period relationships are determined. The values of the ML, Mb, and MM series for 10% and 2% risk are computed to be around 7.2 and 8.3. The tectonic structure and seismic properties of the Bingöl region are also given briefly. 相似文献
12.
13.
14.
The various useful source-parameter relations between seismic moment and common use magnitude lg(M 0) andM s,M L,m b; between magnitudesMs andM L,M s andm b,M L andm b; and between magnitudeM s and lg(L) (fault length), lg (W) (fault width), lg(S) (fault area), lg(D) (average dislocation);M L and lg(f c) (corner frequency) have been derived from the scaling law which is based on an “average” two-dimensional faulting model of a rectangular fault. A set of source-parameters can be estimated from only one magnitude by using these relations. The average rupture velocity of the faultV r=2.65 km/s, the total time of ruptureT(s)=0.35L (km) and the average dislocation slip rateD=11.4 m/s are also obtained. There are four strong points to measure earthquake size with the seismic moment magnitudeM w.
- The seismic moment magnitude shows the strain and rupture size. It is the best scale for the measurement of earthquake size.
- It is a quantity of absolute mechanics, and has clear physical meaning. Any size of earthquake can be measured. There is no saturation. It can be used to quantify both shallow and deep earthquakes on the basis of the waves radiated.
- It can link up the previous magnitude scales.
- It is a uniform scale of measurement of earthquake size. It is suitable for statistics covering a broad range of magnitudes. So the seismic moment magnitude is a promising magnitude and worth popularization.
15.
为了借助容易获取的地震相关因素间接预测地震震级,提出基于相关向量机（Relevance Vector Machine,RVM）方法的地震震级预测模型。通过样本学习建立地震震级与地震累积频度、累积释放能量、平均震级、b值、η值和相关区震级等6个主要影响因素之间的非线性映射关系,利用已知影响因素预测地震震级。结果表明:RVM模型预测结果均优于BP神经网络及SOM-BP神经网络预测结果;通过敏感因子分析比较各因素的敏感程度,b值和η值最为突出,在震级研究中应重点分析。综合分析,RVM模型具有精度高和离散性小等优点,对地震震级预测有较好的推广价值。 相似文献
16.
A pronounced increase in seismicity started in and around Longtan reservoir, southwestern China after October 1, 2006 when it began the impoundment, and by the end of May 14, 2010, about 3,233 earthquakes with ?0.6?≤?M L?≤?4.2 had been located. This seismicity which occurred in five clusters mainly concentrated in the areas where few earthquakes had occurred before the first filling. There were four water filling periods in the Longtan reservoir, and the observed reservoir-induced seismicity (RIS) shows a strong correlation with the filling cycles. After the first filling, there appears to be an instant undrained response due to an elastic response to the reservoir load in the third and fourth cluster. Then, this seismicity is followed by a delayed, drained response due to pore pressure diffusion, with the seismicity migrating outwards in one or more directions in the second and third filling period. The seismic diffusivity (α s) we estimated is about 4.54?×?105?cm2/s. The activity levels in the five clusters are different due to differences in the structures and permeabilities of the faults. The delayed seismic response to the filling in the third cluster was due to the combined effects of the lack of local fault intersecting the reservoir and lower permeability of the rock. The b value we obtained for reservoir-induced events was significantly different and higher than that of pre-impoundment natural tectonic earthquakes in the Longtan reservoir. The results of relocated earthquakes based on double difference earthquake location algorithm showed that their focal depths were mainly shallower than about 10 km and the distribution of relocated RIS in four clusters had no relation with these intersecting faults in the Longtan reservoir except the fifth cluster. All these characteristics of RIS in the Longtan reservoir indicate that they may relate to the coupled poroelastic response that includes both pore pressure diffusion and an undrained response, but the pore pressure diffusion and the water permeation appear to play a more important role on inducing the earthquakes in Longtan reservoir. 相似文献
17.
A comparison of performance of several artificial intelligence methods for forecasting monthly discharge time series 总被引：18，自引：0，他引：18
Developing a hydrological forecasting model based on past records is crucial to effective hydropower reservoir management and scheduling. Traditionally, time series analysis and modeling is used for building mathematical models to generate hydrologic records in hydrology and water resources. Artificial intelligence (AI), as a branch of computer science, is capable of analyzing long-series and large-scale hydrological data. In recent years, it is one of front issues to apply AI technology to the hydrological forecasting modeling. In this paper, autoregressive moving-average (ARMA) models, artificial neural networks (ANNs) approaches, adaptive neural-based fuzzy inference system (ANFIS) techniques, genetic programming (GP) models and support vector machine (SVM) method are examined using the long-term observations of monthly river flow discharges. The four quantitative standard statistical performance evaluation measures, the coefficient of correlation (R), Nash–Sutcliffe efficiency coefficient (E), root mean squared error (RMSE), mean absolute percentage error (MAPE), are employed to evaluate the performances of various models developed. Two case study river sites are also provided to illustrate their respective performances. The results indicate that the best performance can be obtained by ANFIS, GP and SVM, in terms of different evaluation criteria during the training and validation phases. 相似文献
18.
Santanu Baruah Saurabh Baruah P. K. Bora R. Duarah Aditya Kalita Rajib Biswas N. Gogoi J. R. Kayal 《Pure and Applied Geophysics》2012,169(11):1977-1988
An attempt has been made to examine an empirical relationship between moment magnitude (M W) and local magnitude (M L) for the earthquakes in the northeast Indian region. Some 364 earthquakes that were recorded during 1950–2009 are used in this study. Focal mechanism solutions of these earthquakes include 189 Harvard-CMT solutions (M W?≥?4.0) for the period 1976–2009, 61 published solutions and 114 solutions obtained for the local earthquakes (2.0?≤?M L?≤?5.0) recorded by a 27-station permanent broadband network during 2001–2009 in the region. The M W–M L relationships in seven selected zones of the region are determined by linear regression analysis. A significant variation in the M W–M L relationship and its zone specific dependence are reported here. It is found that M W is equivalent to M L with an average uncertainty of about 0.13 magnitude units. A single relationship is, however, not adequate to scale the entire northeast Indian region because of heterogeneous geologic and geotectonic environments where earthquakes occur due to collisions, subduction and complex intra-plate tectonics. 相似文献
19.
Aftershock hazard maps contain the essential information for search and rescue process, and re-occupation after a main-shock. Accordingly, the main purposes of this article are to study the aftershock decay parameters and to estimate the expected high-frequency ground motions (i.e., Peak Ground Acceleration (PGA)) for recent large earthquakes in the Iranian plateau. For this aim, the Ahar-Varzaghan doublet earthquake (August 11, 2012; M N =6.5, M N =6.3), and the Ilam (Murmuri) earthquake (August 18, 2014 ; M N =6.2) have been selected. The earthquake catalogue has been collected based on the Gardner and Knopoff (Bull Seismol Soc Am 64(5), 1363-1367, 1974) temporal and spatial windowing technique. The magnitude of completeness and the seismicity parameters (a,??b) and the modified Omori law parameters (P,??K,??C) have been determined for these two earthquakes in the 14, 30, and 60 days after the mainshocks. Also, the temporal changes of parameters (a,??b,??P,??K,??C) have been studied. The aftershock hazard maps for the probability of exceedance (33%) have been computed in the time periods of 14, 30, and 60 days after the Ahar-Varzaghan and Ilam (Murmuri) earthquakes. For calculating the expected PGA of aftershocks, the regional and global ground motion prediction equations have been utilized. Amplification factor based on the site classes has also been implied in the calculation of PGA. These aftershock hazard maps show an agreement between the PGAs of large aftershocks and the forecasted PGAs. Also, the significant role of b parameter in the Ilam (Murmuri) probabilistic aftershock hazard maps has been investigated. 相似文献
20.
Predictive equations based on the stochastic approach are developed for earthquake ground motions from Garhwal Himalayan earthquakes of 3.5≤Mw≤6.8 at a distance of 10≤R≤250 km. The predicted ground motion parameters are response spectral values at frequencies from 0.25 to 20 Hz, and peak ground acceleration (PGA). The ground motion prediction equations (GMPEs) are derived from an empirically based stochastic ground motion model. The GMPEs show a fair agreement with the empirically developed ground motion equations from Himalaya as well as the NGA equation. The proposed relations also reasonably predict the observed ground motion of two major Himalayan earthquakes from Garhwal Himalayan region. For high magnitudes, there is insufficient data to satisfactorily judge the relationship; however it reasonably predicts the 1991 Uttarkashi earthquake (Mw=6.8) and 1999 Chamoli earthquake (Mw=6.4) from Garhwal Himalaya region. 相似文献