Bayesian Resource Characterization Reducing Conditional Bias
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Kriging methods used in mining plans often systematically over- or underestimate resource properties, leading to conditional bias, which affects the accuracy of predicting ore grades and other properties in mining operations.
Innovation Solution
A method utilizing Bayes' theorem and Gaussian or Gaussian-like probability distributions to model changes in resource properties across a block, calculating parameters associated with these changes, and estimating property characteristics with error estimates, thereby reducing conditional bias and improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If kriging is used to predict resource properties, then statistical prediction can be performed, but conditional bias occurs causing systematic over- or underestimation
Solution Approach 1:
The patent changes the statistical parameters and methods used for prediction by implementing a bootstrap resampling approach with multiple iterations. Instead of using traditional kriging parameters, the system generates multiple simulated datasets through resampling, calculates properties for each iteration, and aggregates results to produce unbiased predictions with associated uncertainty estimates.
2Measurement precision
If traditional statistical methods are used, then prediction can be made, but error estimates are not provided
Solution Approach 1:
The patent implements a feedback mechanism through bootstrap resampling where the system continuously refines predictions by comparing results across multiple iterations. The variation in results across iterations provides feedback about the uncertainty and reliability of predictions, generating natural error estimates that inform subsequent analysis and decision-making.
Data Source
AI summary
The present disclosure provides a method of characterizing a resource in a block located in an area. The method comprises the step of providing information concerning a property of the resource for a plurality of sample positions in, at or in the environment of the block. The information is obtained from analyzes of samples from the sample positions. The method further comprises modelling a change in the property along a distance at or within the block using the provided information. The method also comprises calculating parameters associated with the modelled change using Bayes' theorem and calculating a property characteristic of the resource for the block using the calculated parameters.