Combine The Neural Network Method With The High SNR Method To Pick Up First Arrival Detecting

First arrivals detecting on Seismic Record is important at all time.It has significant function on the refraction static correction,VSP interpretation,shallow refraction and seismic tomography surveys.First arrival detecting is accurate or not can directly affect the result of processing.The workload of first arrival detecting is prodigious in seismic data processing,so its speed has crucial effect to the efficiency of seismic data processing.By far,many methods and theories have been put forward,such as correlation method,ratio of energy method,maximal amplitude method,fractal dimension method and neural network.For some areas that have simple first arrivals,it can reach the require using normal method,but for complicated areas,first arrivals detecting is difficulty is a difficult problem for us for a long time, ANN(Artifical Neural Network) is a cross subject, which can complete difficult intelligent analysis and process effectively nonlinear, fuzzy and uncertain problems with simple mathematic methods. Because of its massive parallel processing, distributed storage, self-adaptation and error allowance, ANN has been widely applied in biology, electronics, computers, mathematics and physics。In this paper,we research and analysis of the classical BP neural network theory. First, we do wavelet plastic to the CSP records and the COP records .Then we pick up first arrival detecting by the way of the high SNR method for the BP neural network learning samples . Finally, we analysis the BP neural network method to pick up the first arrival detecting which based on the CSP records and the COP records , through actual data it shows that the BP neural network method which based on the COP records to pick up first arrival detecting is more better, more efficient.