As was already mentioned, some time ago, participants of the current project from Warsaw University of Technology have built prototype of e-NOSE with the use of neural networks and then successfully applied it to identify forest pests as well as pheromones of different concentrations. This e- NOSE was capable to determine the concentration of pheromone of the order approaching to that of pheromone concentration emitted by insects. Analysis of the testsresults have led to the conclusion that in order to increase research capabilities by using e-NOSE it is necessary to use appropriately selected very sensitive and less or more selective sensors in conjunction with the methods of artificial intelligence. The aim of the project is to build an electronic nose based on the artificial intelligence using selective sensors with very high sensitivity. The main task is to manufacture a matrix of appropriately selected and miniaturised sensors, less or more selective, with different parameters. Research work will be concentrated on proper selection of ion selective sensors with performance appropriate to our goals. This will allow receiving much more comprehensive and complex information, which subsequently permit, using artificial intelligence, to create a neural classifier. The essential part of the task will be to develop the software concerning generated signals and their analysis. To this end there will be adapted the state-of-the-art seeking algorithms to analyze the data, including methods of artificial intelligence (AI). This applies in particular to the development of databases, because the information system should allow cooperation of instrument and the central database. The end product of this will be a standalone computer system for acquisition and data processing, resulting from sensory matrix. This system will have the ability to communicate with the central database. Dedicated sensory matrices will be made for applications referred to the project, in other words, to the analysis of specific smells. The sensory matrices will undergo further actions in the following activities. These actions will lead to collect the data necessary to perform any design changes in matrices. At the same time, the software will be modified. Research on the development of electronic smell sensors are reflected in the literature presented below.

Close Menu