Abstract:Although the self-organizing mapping neural network is capable of self organination,self learning and side associative thinking,it has some defects: •Slow convergence in learning course •The variations of initial condition and sample input sequence are very sensitive to the learning course and the learning result •No full use of reliable priori knowledge during its learning without supervision.To remove the above defects,we improve here the neural network in the following aspects: •In initializing the neural network,we take the known sample vectors as typical samples,initialize the weight vectors of given output nodes,and revise unforcedly the network after each iteration.,thus improving the classification accuracy. •I.earning rate is adaptively regulated to speed up the training of the neural network. •The criterion for judging if the iteration converges is offered to quicken computation. •According to Euclidean distance between the weight vectors of the corresponding nodes in an output layer,the classification numbers of output nodes are rearranged to achieve desirable sequential classification.The improved neural network algorithm has been used to predict the reservoir of a seismic section that passes three boreholes in eastern Zhungeer basin,and the prediction brought good effect.
罗立民, 王允诚. 自组织特征映射网络的改进及在储层预测中的应用[J]. 石油地球物理勘探, 1997, 32(2): 237-245.
Luo Limin, Wang Yuncheng. Improvement of self-organizing mapping neural network and the application in reservoir prediction. OGP, 1997, 32(2): 237-245.