A decision scheme for the interpretation of spectra from wavelength di
spersive X-ray fluorescence spectrometry is described that encompasses
elements from three areas of artificial intelligence: fuzzy logic, ru
le based expert systems and neural net technology. After transforming
the recorded spectra to line spectra by appropriate back-ground correc
tion a reasoning scheme is applied that takes into account not only th
e observed spectra, but also the recording conditions and prior spectr
oscopic information regarding the relative emission probabilities and
the usefulness of the different lines for the purpose of element ident
ification. The latter is done on the basis of a previously described s
cheme to compute conditional a posteriori Bayes probabilities for a ''
mean matrix''. These different pieces of information are then assemble
d into a battery of fuzzy rules. The importance of the rules as well a
s the importance of the X-ray lines is determined in a training proces
s, similar to the one in a feedforward back-propagation network. To fu
rther stabilize the results this network is pruned in a second trainin
g cycle. This, however, had little effect on the quality of interpreta
tion. The advantages of this approach to the interpretation of X-ray s
pectra over older ones are numerous: the system adapts itself to bette
r interpret spectra that are of greater importance to a laboratory as
these are better represented in the training set; the fuzzy logic is c
apable of working with incomplete and uncertain knowledge, and the neu
ral network results based on these fuzzy rules is readily interpretabl
e by the X-ray spectroscopist as every rule can be expressed also in n
atural language as in any classical rule based system.