Pre-stack-texture-based reservoir characteristics and seismic facies analysis

Seismic texture attributes are closely related to seismic facies and reservoir characteristics and are thus widely used in seismic data interpretation.However,information is mislaid in the stacking process when traditional texture attributes are extracted from poststack data,which is detrimental to complex reservoir description.In this study,pre-stack texture attributes are introduced,these attributes can not only capable of precisely depicting the lateral continuity of waveforms between different reflection points but also reflect amplitude versus offset,anisotropy,and heterogeneity in the medium.Due to its strong ability to represent stratigraphies,a pre-stack-data-based seismic facies analysis method is proposed using the selforganizing map algorithm.This method is tested on wide azimuth seismic data from China,and the advantages of pre-stack texture attributes in the description of stratum lateral changes are verified,in addition to the method＇s ability to reveal anisotropy and heterogeneity characteristics.The pre-stack texture classification results effectively distinguish different seismic reflection patterns,thereby providing reliable evidence for use in seismic facies analysis.

Combining Self-organizing Feature Map with Support Vector Regression Based on Expert System

A new approach is proposed to model nonlinear dynamic systems by combining SOM(self-organizing feature map) with support vector regression (SVR) based on expert system. Thewhole system has a two-stage neural network architecture. In the first stage SOM is used as a clus-tering algorithm to partition the whole input space into several disjointed regions. A hierarchicalarchitecture is adopted in the partition to avoid the problem of predetermining the number of parti-tioned regions. Then, in the second stage, multiple SVR, also called SVR experts, that best fit eachpartitioned region by the combination of di?erent kernel function of SVR and promote the configura-tion and tuning of SVR. Finally, to apply this new approach to time-series prediction problems basedon the Mackey-Glass di?erential equation and Santa Fe data, the results show that SVR experts hase?ective improvement in the generalization performance in comparison with the single SVR model.

Modeling and Mining the Temporal Patterns of Service in Cellular Network

Recent emergence of diverse services have led to explosive traffic growth in cellular data networks. Understanding the service dynamics in large cellular networks is important for network design, trouble shooting, quality of service(Qo E) support, and resource allocation. In this paper, we present our study to reveal the distributions and temporal patterns of different services in cellular data network from two different perspectives, namely service request times and service duration. Our study is based on big traffic data, which is parsed to readable records by our Hadoop-based packet parsing platform, captured over a week-long period from a tier-1 mobile operator's network in China. We propose a Zipf's ranked model to characterize the distributions of traffic volume, packet, request times and duration of cellular services. Two-stage method(Self-Organizing Map combined with kmeans) is first used to cluster time series of service into four request patterns and three duration patterns. These seven patterns are combined together to better understand the fine-grained temporal patterns of service in cellular network. Results of our distribution models and temporal patterns present cellular network operators with a better understanding of the request and duration characteristics of service, which of great importance in network design, service generation and resource allocation.