Estimating one-dimensional models from frequency-domain electromagnetic data using modular neural networks

摘要: An artificial neural network interpretation system is being used to interpret data from a frequency-domain electromagnetic (EM) geophysical in near real time. The integrates 45 separate networks visualization shell. produce interpretations at three different transmitter-receiver (Tx-Rx) separations for half-space and layered-Earth interpretations. Modular (MNNs) were found be the only paradigm that could successfully perform MNN with 16 inputs, five local experts, each seven hidden processing elements, outputs was trained on 4795 patterns 200 epochs. For two-layer models resistivity contrast greater than 2:1, estimates had 96% accuracy first-layer resistivity, 98% second-layer thickness of first layer. If less accuracies are unaffected but layers 2 m unreliable. A Tx-Rx separation maximum depth penetration 8 assumed example cited.