Demarcation of potential seismic sources on integration of genetic algorithm and BP algorithm

In this paper potential seismic sources in coastal region of South China are identified by integration of genetic algorithm (GA) and back propagation (BP algorithm). GA is used for finding the best parameter combination rapidly in an infinite solution space for artificial neural networks (ANN). The results show that the distribution of potential seismic sources with different upper magnitude demarcated by this classifier is mostly satisfied the intrinsic relationship between seismic environment and earthquake occurrence, with less effect from subjective judgment of human being.

Artificial Neural Network Method Based on Expert Knowledge and Its Application to Quantitative Identification of Potential Seismic Sources

In this paper,an approach is developed to optimize the quality of the training samples in the conventional Artificial Neural Network(ANN)by incorporating expert knowledge in the means of constructing expert-rule samples from rules in an expert system,and through training by using these samples,an ANN based on expert-knowledge is further developed.The method is introduced into the field of quantitative identification of potential seismic sources on the basis of the rules in an expert system.Then it is applied to the quantitative identification of the potential seismic sources in Beijing and its adjacent area.The result indicates that the expert rule based on ANN method can well incorporate and represent the expert knowledge in the rules in an expert system,and the quality of the samples and the efficiency of training and the accuracy of the result are optimized.

Potential rupture surface model and its ap-plication on probabilistic seismic hazard analysis

Potential sources are simplified as point sources or linear sources in current probabilistic seismic hazard analysis （PSHA） methods. Focus size of large earthquakes is considerable, and fault rupture attitudes may have great influence upon the seismic hazard of a site which is near the source. Under this circumstance, it is unreasonable to use the simplified potential source models in the PSHA, so a potential rupture surface model is proposed in this paper. Adopting this model, we analyze the seismic hazard near the Chelungpu fault that generated the Chi-Chi （Jiji） earthquake with magnitude 7.6 and the following conclusions are reached. （1） This model is reasonable on the base of focal mechanism, especially for sites near potential earthquakes with large magnitude; （2） The attitudes of potential rupture surfaces have great influence on the results of probabilistic seismic hazard analysis and seismic zoning.

Expected magnitude and distance of potential source area and the estimating method

Magnitude and distance of major potential source are needed in order to determine duration time of artificial ground motion and to determine the type of response spectrum (near field or far field) when using the seismic intensity zonation map. The magnitude probabilistic distribution function of seismic belt and the magnitude and space joint distribution function for given intensity of the site in a potential Source are provided. Then the basicformula of calculating expected magnitude and expected distance are developed. Several examples for calculating expected magnitude and expected distance in northern China are discussed. These results show that expected magnitude and expected distance are related not only to geometry of potential source and magnitude but also to the intensity of the site with certain exceeding probability.

Interrelation among several important links in seismic risk analysis

In order to further reveal the interrelation among division of seismic statistical regions, delimitation of potential seismic sources and estimation of seismicity parameters, we select 21 representative sites located in different places within the range of 100°-120°E, 29°-42°N to study the influences of seismicity parameter uncertainties of statistical regions on seismic risk estimations of these sites in the inhomogeneous and homogeneous distribution models. Combining the results from this study and previous ones, we can see that different schemes for dividing seismic statistical regions can change the seismic data in a statistical region. The uncertain data and additional uncertainty in selecting time intervals for seismic statistics will result in uncertainty of seismicity parameters estimation in a statistical region. For the homogeneous model, the larger the variation of this uncertainty is, the greater the uncertain influence on the seismic risk estimation of a site will be, which means that the division of seismic statistical regions makes a major contribution. In a seismic statistical region, the delimitation of potential sources and variant weight assignment of spatial distribution functions can raise the estimated values of ground motion parameters in the place where great earthquake might occur and its vicinity. In these places, the influence of uncertainty in potential source delimitation is very obvious, especially on the absolute magnitude of ground motion parameters （e.g., intensity）, which means that the link of potential source delimitation makes a major effect. Generally speaking, the link of potential source delimitation affects mainly the sites located in the potential sources with the highest and second-high upper-limit earthquake magnitudes or in the vicinity of those with the highest upper-limit magnitude. While for the sites located in the potential sources with low upper-imit magnitudes, the uncertainty influence of statistical region division is larger than that of potential source delimitation.

Probability-consistent spectrum and code spectrum

In the seismic safety evaluation (SSE) for key projects, the probability-consistent spectrum (PCS), usually obtained from probabilistic seismic hazard analysis (PSHA), is not consistent with the design response spectrum given by Code for Seismic Design of Buildings (GB50011-2001). Sometimes, there may be a remarkable difference be-tween them. If the PCS is lower than the corresponding code design response spectrum (CDS), the seismic fortifi-cation criterion for the key projects would be lower than that for the general industry and civil buildings. In the paper, the relation between PCS and CDS is discussed by using the ideal simple potential seismic source. The re-sults show that in the most areas influenced mainly by the potential sources of the epicentral earthquakes and the regional earthquakes, PCS is generally lower than CDS in the long periods. We point out that the long-period re-sponse spectra of the code should be further studied and combined with the probability method of seismic zoning as much as possible. Because of the uncertainties in SSE, it should be prudent to use the long-period response spectra given by SSE for key projects when they are lower than CDS.