Neural Network Ensemble Input Subsets for Well Log Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Optimizing the selection of input parameters for neural networks to achieve stable and accurate predictions is challenging, particularly in applications like well logging where converting cased hole logs to synthetic open hole logs is necessary for effective data comparison and reservoir characterization.
Innovation Solution
Creating a neural network ensemble by training multiple networks on diverse subsets of input signals and varying complexities, then selecting the best ensemble based on validation error, complexity, and negative correlation to enhance prediction robustness and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple input parameters are selected for neural network training, then prediction accuracy is improved, but input optimization becomes difficult and complex
Solution Approach 1:
The patent divides the single complex optimization problem into multiple simpler sub-problems by training multiple neural networks, each with a different subset of input parameters. This segmentation allows each network to be optimized independently for its specific input subset, avoiding the complexity of optimizing all inputs simultaneously for a single network.
Solution Approach 2:
The patent changes the parameter configuration by varying the input parameter subsets across different neural networks in the ensemble. Each network is trained with a different combination of input parameters, allowing the system to explore multiple parameter configurations and select the most effective ones for accurate prediction.
2Device complexity
If a single neural network is used, then the system is simple, but prediction stability and accuracy deteriorate
Solution Approach 1:
The patent merges multiple neural networks into an ensemble system where each network contributes to the final prediction. By combining the predictions of multiple networks trained on different input subsets, the system achieves improved stability and accuracy while maintaining reasonable complexity through modular architecture.
Data Source
AI summary
Methods of creating and using robust neural network ensembles are disclosed. Some embodiments take the form of computer-based methods that comprise receiving a set of available inputs; receiving training data; training at least one neural network for each of at least two different subsets of the set of available inputs; and providing at least two trained neural networks having different subsets of the available inputs as components of a neural network ensemble configured to transform the available inputs into at least one output. The neural network ensemble may be applied as a log synthesis method that comprises: receiving a set of downhole logs; applying a first subset of downhole logs to a first neural network to obtain an estimated log; applying a second, different subset of the downhole logs to a second neural network to obtain an estimated log; and combining the estimated logs to obtain a synthetic log.