This course is designed to provide participants with a basic understanding of neural networks and how they can be used for predicting reservoir and well properties. The course will use software from the Matlab Neural Networks Toolbox. Laptop recommended.

This course is designed to provide participants with an overview of environmental problems and solutions in the petroleum industry.

• Introduction to Neural Networks (NN)
  • What they can be used for
  • Why use NN
  • Basics of NN use
• Types of NN
  • Hopfield, Hamming, Perceptron
  • Example problem solutions using NN
• Language of NN
  • Basic linear algebra
  • Definition of NN components
  • Structure of feed-forward NNs
• Learning rules – how NN work
  • Decision boundaries
  • Linear vs. nonlinear problems
  • Simple perceptron linear example
  • Simple perceptron nonlinear example
• Building a multi-layer perceptron NN (back propagation)
  • Data and populations
  • Inputs
  • Targets
• Training a multi-layer perceptron NN (back propagation)
  • Practice examples
  • Monitoring the training process
  • Understanding the output
  • Applying the trained NN
  • Discussion of NN error
• Examples
  • McCormack problem
  • Well log problem
  • Lithology problem
  • Porosity prediction
  • Permeability prediction
  • Combination NNs
• Interpreting results
  • Visualization/mapping
  • Preparing for simulator input
  • Other
• Other software – ease of use & cost
  • Freeware
  • Shareware
  • Commercial software
• Further applications
  • Nasty prediction problems
  • Will NN solve all of the world’s problems?
  • Wrap-up