Neural Computing Analysis of Multi Layered Geophysical Data

Ross Corben


Neural Mining Solutions, 3rd Floor, 1 Alfred Street, Sydney NSW 2000


Key Words: neural networks, geophysics, anomaly, cluster analysis, fuzzy search



Mining exploration companies are increasingly making use of geophysical prospecting methods in order to extend their search into areas under cover that are inaccessible to more traditional exploration methods.  As these tools become more advanced and are used more extensively, the amount of data being generated and cost of obtaining it is increasing significantly.   The process of managing these large multi layered data sets and analysing them to detect the complex relationships and subtle patterns that may exist is critical in terms of realising the full potential of the data.


Neural computing techniques have traditionally been used in the financial industry to analyse large amounts of data for relationship identification, pattern recognition, make associations, highlight anomalies and make predictions automatically.  Applied to the various forms of exploration data, they provide a powerful analysis tool that is orders of magnitude more time efficient than current methods which rely heavily on the ability of experts and are time consuming.


In order to help demonstrate the use of neural network techniques, the Northern Parkes Discovery 2000 data set covering the Nyngan-Narromine 1:250,000 sheet areas have been processed and analysed using the commercially available neural network software package Prospect Explorer.  The two main types of neural networks searches are demonstrated.  An unsupervised neural network is first run in order to organise the multi layered geophysical data sets into clusters and to identify any anomalous patterns.  A supervised neural fuzzy search is then run looking for North Parkes lookalikes beneath the black soil plains in the northern part of the study area.

PowerPoint Presentation