Probabilistic neural network algorithm for using radon emanations as an earthquake precursor

The investigation throughout the world in past two decades provides evidence which indicate that significance variation of radon and other soil gases occur in association with major geophysical events such as earthquake. The traditional statistical algorithm includes regression to remove the effect of the meteorological parameters from the raw radon and anomalies are calculated either taking the periodicity in seasonal variations or periodicity computed using Fast Fourier Transform. In case of neural networks the regression step is avoided. A neural network model can be found which can learn the behavior of radon with respect to meteorological parameter in order that changing emission patterns may be adapted to by the model on its own. The output of this neural model is the estimated radon values. This estimated radon value is used to decide whether anomalous behavior of radon has occurred and a valid precursor may be identified. The neural network model developed using Radial Basis function network gave a prediction rate of 87.7%. The same was accompanied by huge false alarms. The present paper deals with improved neural network algorithm using Probabilistic Neural Networks that requires neither an explicit step of regression nor use of any specific period. This neural network model reduces the false alarms to zero and gave same prediction rate as RBF networks.     

Deepwater helium in Lake Baikal as an earthquake precursor

In this paper, attention is paid to the importance of short-term prognosis of earthquakes. The variability of determination methods is noted. One of the geochemical methods, based on study of the helium content in deep water of Lake Baikal, is considered; such a method has not been used for open deep-water basins within the zones of high seismic danger. It is established that in the period of earthquake preparation, variations in the helium content deep underwater in Lake Baikal are recorded. A sharp decrease in the helium content two days before the earthquake was recorded first time for a long period of observation, as well as the consequent increase. Further study of the helium content deep underwater in Lake Baikal is recommended, and, should these data be proved, it is recommended as a short-term precursor of earthquakes.     

Seismic human loss estimation for an earthquake disaster using neural network

In Iran, earthquakes cause enormous damage to the people and economy. If there is a proper estimation of human losses in an earthquake disaster, it could be appropriately responded and its impacts and losses will be decreased. Neural networks can be trained to solve problems involving imprecise and highly complex nonlinear data. Based on the different earthquake scenarios and diverse kind of constructions, it is difficult to estimate the number of injured people. With respect to neural network’s capabilities, this paper describes a back propagation neural network method for modeling and estimating the severity and distribution of human loss as a function of building damage in the earthquake disaster. Bam earthquake data in 2003 were used to train this neural network. The final results demonstrate that this neural network model can reveal much more accurate estimation of fatalities and injuries for different earthquakes in Iran and it can provide the necessary information required to develop realistic mitigation policies, especially in rescue operation.     

Evidence of precursor phenomena in the Kobe earthquake obtained from atmospheric radon concentration

Atmospheric ²²²Rn concentrations were determined over a 10a period, which included the date of the Kobe, Japan earthquake, on January 17th 1995. It was found that the seismically related ²²²Rn anomaly was higher than the 99% confidence limits for the residual value of atmospheric ²²²Rn which had been observed 2 months before. The residual ²²²Rn concentration, in which residual values of the daily minimum are the difference between each normal ²²²Rn concentration (calculated from January 1984 to December 1993) and the daily minimum ²²²Rn concentration (January 1994 to January 1995), was calculated by applying the exponential smoothing method to the residual values for each day. It was found that the fluctuations of the residual values can be fitted very well to a log-periodic oscillation model. The real residual values stopped increasing at 1994.999 (December 31st 1994), which corresponds with the critical point (t_c) of best fit model. This anomalous ²²²Rn variation can be seen as the result of local stresses, not primary stresses which directly lead to the Kobe earthquake. On the other hand, when the critical exponent (z) and the radial frequency (ω) of the model were simultaneously fixed 0.2 ⩽ z ⩽ 0.6 and 6 ⩽ ω ⩽ 12, t_c (critical point) was between January 13th 1995 and January 27th 1995. The Kobe earthquake occurrence date (January 17th 1995) is within this range. Therefore this anomalous ²²²Rn variation can also be seen as the result of primary stresses which possibly led to the Kobe earthquake. There is a distinct possibility that similar statistical oscillations will be detected in other measurements such as microseismicity, tectonic strain, fluctuation in the ground level, or changes in groundwater elevations and composition.     

A neural network approach to geostatistical simulation

Neural networks offer a non-algorithmic approach to geostatistical simulation with the possibility of automatic recognition of correlation structure. The paper gives a brief overview of neural networks and describes a feedforward, back-propagation network for geostatistical simulation. The operation of the network is illustrated with two simple one-dimensional examples which can be followed through with hand calculations to give an insight into the operation of the network. The convergence of the network is described in terms of the variogram calculated from the values at each of the output nodes at each iteration.     

Prediction of pile capacity from stress-wave measurements: A neural network approach

A neural network approach for the prediction of pile bearing capacity by the stress-wave matching technique is presented. The main advantage of this approach over the traditional manual or automated matching approach is that it avoids the time-consuming process of iterative adjustment. This makes it feasible to determine the static pile capacity in real time in the field. Another benefit of this approach is that as more case histories become available, the neural network can be improved by learning from these new examples. A three-layer back-propagation network is set up to illustrate the capability of the proposed approach for 70 dynamically tested concrete bored piles. A wave equation model developed at the National University of Singapore and coded in the NUSWAP computer program is used to formulate the problem. Up to 14 of the 70 piles (20 percent) are used in training the network. The NUSWAP program is used to generate simulation training examples based on the manually fitted training examples for further training of the network. Different network configurations are examined. The trained network produces results exhibiting good stress-wave matching qualities compared to those obtained by manual fitting. The pile bearing capacities predicted by the two approaches agree very closely. The load-settlement curve and axial load distribution in the pile computed using the network-predicted soil parameters are in good agreement with the field measurements obtained from a maintained load test.     

模糊神经网络在矿震预测中的应用

矿震同天然地震一样会给矿山生产及人身安全等带来重大灾难。也是目前尚不能准确预测的矿山灾害现象之一。根据现有的研究成果可知，矿震是一个多输入、多干扰、单输出的复杂系统。由于干扰项的存在，使利用建模、神经网络等手段对系统进行预测时会导致很大误差。模糊神经网络系统在建立对象输入、输出关系时与传统数学方法不同。即可以建立在无模型基础上，并利用其较强的学习训练特性，自动获取对象的输入、输出关系表达；可以将专家的评价语言作为系统的干扰项引入。这在一定程度上缓解了人为因素对预测结果的影响，且平滑了观测数据的随机性。文章利用改进的模糊神经网络及抚顺老虎台矿的矿震资料，对矿震最大震级的预测方法进行了探索。‘初步探讨了改进的模糊神经网络在矿震预测中的应用。得出在运用模糊神经网络进行预测时，为减小预测误差，应综合多种因素并提高专家评判语言的精确度的结论。指出在建立矿震系统预测模型时，利用干扰项将人为因素引入系统是必需的。通过实际应用证明其可行性。     

砂土地震液化的神经网络预测

在简要分析BP算法的基础上,应用BP网络的理论与方法,选取砂土的平均粒径(d50／mm)、相对密度(Dr／％)、标准贯入击数(N63.5／击)、上覆有效应力(σv／kPa)、地震烈度(I0)作为指标,预测砂土在地震作用下液化的可能性,取得了较好的预测效果。     

Variations of the cosmic ray fluxes as a possible earthquake precursor

Variations of the air pressure, cosmic ray fluxes, sunspot numbers, and interplanetary magnetic field in connection with strong earthquake occurrences are studied. The results of this investigation permits one to consider the variations of the cosmic rays as one of the possible cause of air pressure variations and one of the possible earthquake precursors.     

Application of neural network and ANFIS model for earthquake occurrence in Iran

This study examined the spatial-temporal variations in seismicity parameters for the September 10^th, 2008 Qeshm earthquake in south Iran. To this aim, artificial neural networks and Adaptive Neural Fuzzy Inference System (ANFIS) were applied. The supervised Radial Basis Function (RBF) network and ANFIS model were implemented because they have shown the efficiency in classification and prediction problems. The eight seismicity parameters were calculated to analyze spatial and temporal seismicity pattern. The data preprocessing that included normalization and Principal Component Analysis (PCA) techniques was led before the data was fed into the RBF network and ANFIS model. Although the accuracy of RBF network and ANFIS model could be evaluated rather similar, the RBF exhibited a higher performance than the ANFIS for prediction of the epicenter area and time of occurrence of the 2008 Qeshm main shock. A proper training on the basis of RBF network and ANFIS model might adopt the physical understanding between seismic data and generate more effective results than conventional prediction approaches. The results of the present study indicated that the RBF neural networks and the ANFIS models could be suitable tools for accurate prediction of epicenteral area as well as time of occurrence of forthcoming strong earthquakes in active seismogenic areas.     

A hybrid Bayesian back‐propagation neural network approach to multivariate modelling

There is growing interest in the use of back‐propagation neural networks to model non‐linear multivariate problems in geotehnical engineering. To overcome the shortcomings of the conventional back‐propagation neural network, such as overfitting, where the neural network learns the spurious details and noise in the training examples, a hybrid back‐propagation algorithm has been developed. The method utilizes the genetic algorithms search technique and the Bayesian neural network methodology. The genetic algorithms enhance the stochastic search to locate the global minima for the neural network model. The Bayesian inference procedures essentially provide better generalization and a statistical approach to deal with data uncertainty in comparison with the conventional back‐propagation. The uncertainty of data can be indicated using error bars. Two examples are presented to demonstrate the convergence and generalization capabilities of this hybrid algorithm. Copyright © 2003 John Wiley & Sons, Ltd.     

Identification of sedimentary facies with well logs: an indirect approach with multinomial logistic regression and artificial neural network

Taking K-successions of the H-Zone of the Pearl River Mouth Basin as a testing example, we used two kinds of approaches to implement the microfacies identification. One is a direct identification, the other is an indirect approach in which we conducted the lithofacies classification first and then identified the microfacies based on previously estimated lithofacies. Both approaches were trained and checked by interpretations of experienced geologists from real subsurface core data. Multinomial logistic regression (MLR) and artificial neural network (ANN) were used in these two approaches as classification algorithms. Cross-validations were implemented. The source data set was randomly divided into training subset and testing subset. Four models, namely, MLR_direct, ANN_direct, MLR_indirect, and ANN_indirect, were trained with the training subset. The result of the testing set shows that the direct approaches (MLR_direct and ANN_direct) perform relatively poor with a total accuracy around 75%. While the indirect approaches (MLR_indirect and ANN_indirect) perform much better with a total accuracy of around 89 and 82%, respectively. This indirect method is simple and reproducible, and it could lead to a robust way of analyzing sedimentary microfacies of horizontal wells with little core data or even are almost never cored while core data are available for nearby vertical wells.     

A neural network approach for attenuation relationships: An application using strong ground motion data from Turkey

This paper presents an application of neural network approach for the prediction of peak ground acceleration (PGA) using the strong motion data from Turkey, as a soft computing technique to remove uncertainties in attenuation equations. A training algorithm based on the Fletcher–Reeves conjugate gradient back-propagation was developed and employed for three sample sets of strong ground motion. The input variables in the constructed artificial neural network (ANN) model were the magnitude, the source-to-site distance and the site conditions, and the output was the PGA. The generalization capability of ANN algorithms was tested with the same training data. To demonstrate the authenticity of this approach, the network predictions were compared with the ones from regressions for the corresponding attenuation equations. The results indicated that the fitting between the predicted PGA values by the networks and the observed ones yielded high correlation coefficients (R²). In addition, comparisons of the correlations by the ANN and the regression method showed that the ANN approach performed better than the regression. Even though the developed ANN models suffered from optimal configuration about the generalization capability, they can be conservatively used to well understand the influence of input parameters for the PGA predictions.     

A preliminary design of a radon potential map for Canada: a multi-tier approach

Canada is a vast country with most of the population living on a small portion of the land. However, for a national radon potential map, it is mandatory to cover the entire country including sparsely populated areas. Because of these characteristics, the radon map development for Canada is challenging. After briefly reviewing of radon map development in the world, this study considers a multi-tier approach to best use available however limited resources and to generate a national radon map in a timely fashion. In summary, radon potential maps for highly populated areas should be determined by direct indoor radon measurements where enough indoor radon data are available. In areas where indoor radon measurements are limited or not yet available, the radon potential maps could be developed from various data sources with a multi-factor scoring system including geological information on soil permeability, soil gas radon concentration and ground uranium concentration. In sparsely populated areas, radon potential maps can only be generated with geological predictive tools, especially in those areas where no houses have yet been built. Because indoor radon measurement data and geological information relevant to radon are very limited in Canada, a multi-step strategy is also worth considering in addition to the multi-tier approach.     

A probabilistic approach for earthquake risk assessment based on an engineering insurance portfolio

This paper presents the construction classification of the existing engineering covers in Taiwan. The exposure profile and variable vulnerability during different construction phases are established for some kinds of classes of construction. Finally, we present a method and framework to estimate the probable maximum loss of engineering insurance portfolio during an earthquake with consideration of the dynamic nature of structural changes and exposure values during a construction project.     

基于神经网络理论的三峡水库诱发地震预测研究

在地震地质背景研究的基础上,运用神经网络理论中改进的BP算法对三峡水库诱发地震强度进行了预测研究.预测结果表明,秭归盆地高桥断裂近库段有可能诱发Ms＞6.0级的地震,可能诱发4.5＜Ms＜6.0级地震的库段有庙河口-九湾溪、九湾溪-路口子断层沿线、瞿塘峡北岸库段、大宁河西岸近库段;仙女山断层沿线、高桥断裂灰岩区远库段、巴东以南灰岩区库段等;可能诱发3.0＜Ms＜4.5级地震的库段有巴东-官渡口砂页岩区、巴东以南灰岩区库段,其它库段均为Ms＜3.0级.     

Integrated model for earthquake risk assessment using neural network and analytic hierarchy process:Aceh province,Indonesia

Catastrophic natural hazards,such as earthquake,pose serious threats to properties and human lives in urban areas.Therefore,earthquake risk assessment(ERA)is indispensable in disaster management.ERA is an integration of the extent of probability and vulnerability of assets.This study develops an integrated model by using the artificial neural network–analytic hierarchy process(ANN–AHP)model for constructing the ERA map.The aim of the study is to quantify urban population risk that may be caused by impending earthquakes.The model is applied to the city of Banda Aceh in Indonesia,a seismically active zone of Aceh province frequently affected by devastating earthquakes.ANN is used for probability mapping,whereas AHP is used to assess urban vulnerability after the hazard map is created with the aid of earthquake intensity variation thematic layering.The risk map is subsequently created by combining the probability,hazard,and vulnerability maps.Then,the risk levels of various zones are obtained.The validation process reveals that the proposed model can map the earthquake probability based on historical events with an accuracy of 84%.Furthermore,results show that the central and southeastern regions of the city have moderate to very high risk classifications,whereas the other parts of the city fall under low to very low earthquake risk classifications.The findings of this research are useful for government agencies and decision makers,particularly in estimating risk dimensions in urban areas and for the future studies to project the preparedness strategies for Banda Aceh.     

地震与地震前兆数字化观测基础

本文介绍了数字观测中的模数转换、采样、量化和传输的基本原理并以地形变数字观测为例介绍了采样定理的应用     

Flood forecasting based on an artificial neural network scheme

Natural Hazards - Nowadays, floods have become the widest global environmental and economic hazard in many countries, causing huge loss of lives and materials damages. It is, therefore, necessary...