Geological evaluations of coal deposits have been traditionally
undertaken in a somewhat manual manner. Information is collected
downhole by chip sampling, coring and using downhole geophysical
logging. This information is then interpreted by an experienced geologist
to define the coal and noncoal lithological units. Coal units were
subsequently correlated to provide data for structural modelling. In recent
years a number of advancements in interpretation techniques have been
made based on Artificial Intelligence. This paper describes an integrated geological interpretation system
based on maximised use of Al techniques. The first part describes the
lithological interpretation which uses a number of boreholes with both
logged lithology and downhole geophysical logs. This information
provides a learning set of data to the neural network. The second part concentrates on the correlation of both coal and
noncoal units from borehole to borehole. Problems in this area, in a
manual sense, are interpretation of correlation across weathering zones
and other geological anomolies. The technique is based on pattern
recognition and dynamic programming techniques where a correlation
matrix is developed relating all holes with their maximum correlations.