Samra M. Neuronetworks methods and means for gas recognition for medical diagnostics and ecology monitoring

As a result of dissertation work implementation have done the next items. The tasks of medical diagnostics and ecological monitoring, solved by gas multisensor signals recognizers, were systemazed. The systems analysis of known gas multisensorr types and their output signals were reviewed. The state-of-the-art review of the methods and facilities for multisensor signals recognition were conducted. The method for gas multisensor signals recognition was developed with the use of pulsed neural net-work (PNN). The structure of PNN was improved. The variants of PNN's neuron realization were of-fered on the basis of photosensitive bispin-devices. The specified mathematical model of pulsed neu-ron, based on bispin-device, was developed with the purpose of determination of its adequacy to mathematical model of formal pulsed LIF-neuron. A method of PNN learning was improved with the purpose of its adaptation to optoelectronic implementation. Principles of pulsed neuron technical im-plementation, based on bispin-device, were investigated. PNN technical implementation, based on op-toelectronic elementary basis, were proposed. Experimental researches of bispin-based pulsed neuron parameters were conducted. Computer simulation of the proposed neurocomputer system for gas mul-tisensor signals recognition was conducted.