An autonomous and portable multisensor system has been constructed, which comprises four SAW devices, four semiconductor metal oxide sensors, and a microscontroler in order to relalize the sensor array data treatment using a back-propagation neural network. These processing modules were based on a shifting temporal window technique, and thus permit to implement a dynamic detection unit. Using the RS232 port, the system can be recalibrated and thus the network topology and the learning parameters can be adapted. To validate this approach, GB gas SAW responses have been applied and permonstrate the performances of the system.
展开▼
