潜在地震灾害预测和评价神经网络系统及其若干问题

结合人工神经网络自身的特性和地震灾害预测研究的特点,本文应用神经网络模型,建立了潜在地震灾害预测和评价系统。针对网络模型参数设置、数据归一化、中间层神经元最优数目以及泛化分类评价指标等若干实际问题给出了实际可行的解决方案。通过大样本数据对网络的训练,形成了有识别和记忆功能的非线性预测和评价系统。对网络的测试和检验,论证了该系统在预测潜在地震灾害上的可行性和有效性。同时,从测试精度出发,探讨了这种预测网络存在的不足,并给出了相应的改进建议,为开展进一步的研究工作提供了参考。     

Machine learning for pore-water pressure time-series prediction: Application of recurrent neural networks

Knowledge of pore-water pressure(PWP)variation is fundamental for slope stability.A precise prediction of PWP is difficult due to complex physical mechanisms and in situ natural variability.To explore the applicability and advantages of recurrent neural networks(RNNs)on PWP prediction,three variants of RNNs,i.e.,standard RNN,long short-term memory(LSTM)and gated recurrent unit(GRU)are adopted and compared with a traditional static artificial neural network(ANN),i.e.,multi-layer perceptron(MLP).Measurements of rainfall and PWP of representative piezometers from a fully instrumented natural slope in Hong Kong are used to establish the prediction models.The coefficient of determination(R^2)and root mean square error(RMSE)are used for model evaluations.The influence of input time series length on the model performance is investigated.The results reveal that MLP can provide acceptable performance but is not robust.The uncertainty bounds of RMSE of the MLP model range from 0.24 kPa to 1.12 k Pa for the selected two piezometers.The standard RNN can perform better but the robustness is slightly affected when there are significant time lags between PWP changes and rainfall.The GRU and LSTM models can provide more precise and robust predictions than the standard RNN.The effects of the hidden layer structure and the dropout technique are investigated.The single-layer GRU is accurate enough for PWP prediction,whereas a double-layer GRU brings extra time cost with little accuracy improvement.The dropout technique is essential to overfitting prevention and improvement of accuracy.     

Fuzzy Optimization Neural Network Approach for Ice Forecast in the Inner Mongolia Reach of the Yellow River/Approche d'Optimisation Floue de Réseau de Neurones pour la Prévision de la Glace Dans le Tronçon de Mongolie Intérieure du Fleuve Jaune

Abstract

In ice forecasting, a key problem is the forecast of freeze-up and break-up dates. Ice-water mechanics and the principle of heat-exchange were mainly adopted in previous research. However, the mathematical models in these studies are complex and many parameters are required in relation to upstream and/or downstream gauging stations. Moreover, too many assumptions or simplifications for these parameters and constraints directly lead to low accuracy of the models and limitations as to their practical applications. This paper develops a fuzzy optimization neural network approach for the forecast of freeze-up date and break-up date. The Inner Mongolia reach lies in the top north of the Yellow River, China. Almost every year ice floods occur because of its special geographical location, hydrometeorological conditions and river course characteristics. Therefore, it is of particular importance for ice flood prevention to forecast freeze-up date and break-up date accurately. A case study in this region shows that the proposed methodology may allow obtaining useful results.     

Are artificial neural network techniques relevant for the estimation of longitudinal dispersion coefficient in rivers? / Les techniques de réseaux de neurones artificiels sont-elles pertinentes pour estimer le coefficient de dispersion longitudinale en rivières?

Abstract

Abstract Accurate application of the longitudinal dispersion model requires that specially designed experimental studies are performed in the river reach under consideration. Such studies are usually very expensive, so in order to quantify the longitudinal dispersion coefficient, as an alternative approach, various researchers have proposed numerous empirical formulae based on hydraulic and morphometric characteristics. The results are presented of the application of artificial neural networks as a parameter estimation technique. Five different cases were considered with the network trained for different arrangements of input nodes, such as channel depth, channel width, cross-sectionally averaged water velocity, shear velocity and sinuosity index. In the case where the sinuosity index is included as an input node, the results turned out to be better than those presented by other authors.     

Neuro-fuzzy models employing wavelet analysis for suspended sediment concentration prediction in rivers

Abstract

The study of sediment load is important for its implications to the environment and water resources engineering. Four models were considered in the study of suspended sediment concentration prediction: artificial neural networks (ANNs), neuro-fuzzy model (NF), conjunction of wavelet analysis and neuro-fuzzy (WNF) model, and the conventional sediment rating curve (SRC) method. Using data from a US Geological Survey gauging station, the suspended sediment concentration predicted by the WNF model was in satisfactory agreement with the measured data. Also the proposed WNF model generated reasonable predictions for the extreme values. The cumulative suspended sediment load estimated by this model was much higher than that predicted by the other models, and is close to the observed data. However, in the current modelling, the ANN, NF and SRC models underestimated sediment load. The WNF model was successful in reproducing the hysteresis phenomenon, but the SRC method was not able to model this behaviour. In general, the results showed that the NF model performed better than the ANN and SRC models.

Citation Mirbagheri, S. A., Nourani, V., Rajaee, T. & Alikhani, A. (2010) Neuro-fuzzy models employing wavelet analysis for suspended sediment concentration prediction in rivers. Hydrol. Sci. J. 55(7), 1175–1189.     

Modelling combined open channel flow by artificial neural networks

Combined open channel flow is encountered in many hydraulic engineering structures and processes, such as irrigation ditches and wastewater treatment facilities. Extensive experimental studies have conducted to investigate combined flow characteristics. Nevertheless, there is no simple relationship that can fully describe the velocity profiles in a turbulent flow. The artificial neural network (ANN) has great computational capability for solving various complex problems, such as function approximation. The main objective of this study is to evaluate the applicability of the ANN for simulating velocity profiles, velocity contours and estimating the discharges accordingly. The velocity profiles measured by an acoustic doppler velocimeter in the open channel of the Chihtan purification plant, Taipei, with different discharges at fixed measuring section and different depths are presented. The total number of data sets is 640 and the data sets are split into two subsets, i.e. training and validation sets. The backpropagation algorithm is used to construct the neural network. The results demonstrate that the velocity profiles can be modelled by the ANN, and the ANN constructed can nicely fit the velocity profiles and can precisely predict the discharges for the conditions investigated. Copyright © 2005 John Wiley & Sons, Ltd.     

Knowledge integration in sustainability governance through science-based actor networks

Understanding how science, technology and innovation can best help to accelerate progress in achieving sustainable development remains a grand challenge for researchers and practitioners. In the context of the global consultation process for preparing a post-2015 Sustainable Development Agenda, various science-based actor networks have emerged, aiming to translate research into political decision-making and to inform transformations towards sustainability. Over the last years, these networks seem to have taken an ever-growing role in structuring the science-policy interface in global sustainability governance. The question arises, however, how they understand and organize ‘scientific knowledge integration’ in sustainability politics.This study offers a structured comparison of twelve global science-based actor networks engaged in the implementation of the Sustainable Development Goals. It shows that these networks use two types of strategies to foster scientific knowledge integration in sustainability governance. A new framework emerges, in which each strategy corresponds to two main approaches of scientific knowledge integration: The entrepreneurial strategy generally seeks to advance advice-oriented and solution-oriented knowledge processes, while assessment-oriented and learning-oriented processes in scientific knowledge integration are mainly promoted through a mediating strategy.     

Asymptotic Rayleigh instantaneous unit hydrograph

The instantaneous unit hydrograph for a channel network under general linear routing and conditioned on the network magnitude,N, tends asymptotically, asN grows large, to a Rayleigh probability density function. This behavior is identical to that of the width function of the network, and is proven under the assumption that the network link configuration is topologically random and the link hydraulic and geometric properties are independent and identically distributed random variables. The asymptotic distribution depends only on a scale factor, , where is a mean link wave travel time.     

Explorations into the formal structure of drainage basins

Every basin of higher than first order is drained by a channel network composed of two subnetworks. Their basins are separated by a drainage divide line, called the basin divider, which is the primary organizing feature of the main basin. Each basin of magnitude n contains n – 1 subnetworks of higher order, and is therefore organized by a set of n – 1 dividers. The dividers and the basin boundary are interconnected in a graph called the divider network of the basin; in graph-theoretic terms this network forms a tree and has the same magnitude and link numbers as the channel network draining the basin. While the subbasins and subnetworks of a drainage basin form a nesting hierarchy, the corresponding dividers do not; indeed, any two dividers share at most one node in common, and whether they do so is independent of whether the corresponding subbasins are nesting or disjoint. However, the dividers of nesting basins are linked by recursive relationships which permit the derivation of a set of algebraic equations; these equations relate the dividers of a basin to other basin components; for example, their combined length is equal to half the length of all first-order basin boundaries minus the length of the main basin boundary. The second part of the paper explores the dependence of the divider length on other basin parameters. The expected length, as predicted by the assumption of topological randomness, is clearly rejected by the data. An alternative approach (regression) is based on the observed magnitudes of the subbasins separated by each divider, and is reasonably successful in estimating divider length. The last section introduces the concept of the standardized basin defined by a boundary length of unity; the estimated lengths of the basin divider and the basin boundary permit an approximate reconstruction of the idealized basin shape and the location of the divider in it.     

Comparing Human and Neural Network Learning of Climate Categories

This study compares how humans and neural networks classify climate types. Human subjects were asked to classify climates from monthly temperature and precipitation patterns. To model their learning process, the same data were used to produce input vectors that trained a pattern associator neural network. Both human subjects and the neural network classified climates accurately after 10 rounds of supervised learning. The neural network successfully modeled the rate of human learning and the ability to learn specific climate categories. Moreover, the neural network weights used to classify climates correspond to distinct visual characteristics in temperature and precipitation. These results suggest that neural networks can model the formation of visual categories.