The introduced method of pattern recognition is a tool for on line control purposes. It is based on the detection of complex states of (bio-)technological processes.
The process data are treated in a comprehensive way which allows the formulation and detection of interrelations between different signals. Not only single signals rather than a multiple set defines the state of a process (more) completely. This allows to perform “advanced control” of the “physiological state” of a population and additionally the on line supervision of the technical equipment.
Differences in absolute values between corresponding states do not impair the inter-signal associations seriously if the general character of the trajectories remains unchanged. Hence, this approach is very similar to a human view of processes and is suited to serve in expert systems.
First results are shown concerning stability of the recognition procedure and a practical approach is introduced performing the automatic detection of defective sensors.