Summary:
Artificial Intelligence is generally divided into two basic categories, rule-based (expert) systems and adaptive (neural) systems. Neural network, a biologically inspired computing scheme, is an analog, adaptive, distributive, and massively parallel system.
The science of pattern recognition is concerned with three major issues:
1. The appropriate description of objects, physical or conceptual, in terms of representation space;
2. The specification of an interpretation space;
3. The mapping from representation space into interpretation space.
GRNN (General Regression Neural Networks) was used to identify the design structure of the optimum network and then this design structure was used in a back propagation network for the final result.
Three input neurons, namely each well’s relative location as well as relative depth were dedicated to geophysical well log responses such as gamma ray, bulk density and deep induction.
Spatial coordinates, gamma-ray, bulk density, and deep induction log responses, as well as some geological interpretations, were used as input to the neural networks.
Bulk density, porosity
Gammy-ray, clay content, conduct fluid
Deep induction, water saturation
Back-propagation neural network used well log responses and geological sub-divisions as inputs and rock parameters as outputs.
The prediction results are as follows:
It
is interesting to note that for comparable amount of efforts to neuro-process
permeability, porosity, water and oil saturations, better results have been
achieved for the first three ones in comparison to that of oil saturations. This
is also true in traditional calculations of these rocks parameters from resistivity
and density logs.
Tomorrow, I will read more papers on ANN.
No comments:
Post a Comment