Optimization Of Seismic Attributes And BP Wavelet Neural Network In Prediction Of Reservoir Parameters

The seismic attribute is the important means of horizontal reservoir prediction. They are key problem which is decide to keep the reservoir prediction success that how to accurate withdraw a purpose layer seismic attribute and how to carry out seismic attribute optimization and how build up the relation of reservoir parameter.This paper introduces the basic theory of seismic attributes and its optimization methods; the analysis of kernel principal component method and the differences with the general principal component analysis methods are also described. The kernel principal component method was used in seismic attribute optimization. BP wavelet neural network is a new type wavelet neural network based on BP algorithm, in which the BP algorithm uses gradient descent to reduce the error function. BP wavelet neural network forecasting method has the advantage of strong anti-interference, fault tolerance, not strict requirements of independence of the parameter, and can study the characteristics of a large number of earthquake parameters. Prediction of reservoir parameters has broad application prospects.In this paper, according to log data, combined with interpretation of the results of drilling, we use BP wavelet neural network to forecast single well sand body of the work floor area for three wells, and gave the optimization of seismic attributes. With the single well logging data sand body results, we used the BP wavelet neural network to predict sand body horizontally. Finally, we gave the qualitative prediction results of the work area storage layer plane distribution based on root-mean-square amplitude and planar distribution of sandstone thickness...