论文部分内容阅读
利用基于模拟退火算法的神经网络技术进行测井约束的波阻抗反演，可根据数据本身之间的内在联系建立一个自适应非线性认知系统，只要在输入端输入特征数据，便能在输出端得到期望输出值，而不必关心系统本身的内部机理。在反演前，从测井资料中整理出地层波阻抗参数，用神经网络建立起地震波特征和地层波阻抗参数的映射关系，然后再利用这种映射关系进行外推，得到其它地震道所对应的波阻抗参数。在训练过程中，引入了模拟退火算法，使网络能有效地避开局部极小，这样可以提高收敛速度和拟合精度。
Using the neural network technology based on simulated annealing algorithm to conduct logging impedance inversion, an adaptive nonlinear cognitive system can be established according to the intrinsic relations between the data itself. As long as the input characteristic data is input, End to get the desired output value, without having to care about the internal mechanism of the system itself. Before inversion, stratigraphic wave impedance parameters are sorted out from the well logging data, and the mapping relationship between the seismic wave characteristics and the formation impedance parameters is established by neural network. Then, this mapping relationship is used to extrapolate the corresponding stratigraphic waveforms The wave impedance parameters. In the process of training, the simulated annealing algorithm is introduced to make the network effectively avoid local minima, which can improve the convergence speed and the fitting accuracy.