The Oil-Gas Prediction of Seismic Reservoir Based on Rough Set and PSO Algorithm

In the oil-gas prediction of seismic reservoir, the traditional method directly classify by attribute. However, the dimension of input information is so large that the calculation is time-consuming, the storage capacity demanding and the network structure complex. Moreover it is easy to be caught in local minimum in the sample learning. Therefore, a method of oil-gas prediction in seismic reservoir based on rough set and PSO algorithm is presented. The main process is to reduce the seismic attributes by the method of attribute reduction in rough set, which can simplify the input structure and reduce the time needed to train those involved. The prediction system of neural network based on PSO algorithm can overcome many disadvantages in traditional BP network, and improve the training process. The simulation experiments and actual examples show the network structure constructed by attribute reduction not only can achieve the prediction precision, but also can save cost, improve process speed and have notable effect on oil-gas prediction.