Artificial intelligence is incorporated into a sequential injection an
alyzer by means of an expert system which is able to derive all the ne
cessary parameters for the flow system to be automatically configured.
The knowledge base behind the expert system is described. The expert
system is applied and validated for the sequential injection determina
tion of Ca and Mg in natural waters with spectrophotometric diode arra
y detection and multivariate calibration. The analytical system is sim
ple and robust and allows Ca and Mg to be simultaneously determined in
the range 0-280 mg l(-1) and 0-140 mg l(-1), respectively. The method
has the additional advantage of including interferences in the establ
ished multivariate model and so previous separation or masking steps a
re unnecessary. The throughput of the method is 20 samples per hour. T
he expert system gives both general and specific information so that s
equential injection methodologies for different analytes and samples c
an be easily developed.