VARIANT OF ODOR SYSTEM FOR THREATS RECOGNITION
Annotation
Subject of Research. The paper deals with a new design solution for the odor recognition system aimed at detecting threats. An algorithm for neural network training that solves the problem of recognizing dangerous substances by odor is developed. We present experiment results on the selection of neural network hyper parameters, its architecture and testing. Methods. The detection system is a comprehensive solution that gives the possibility to ensure the safety of life and human activity applying odor detection of threats. The solution complexity is realized by technology platform choice of the Internet of things and the neural network that solves recognition problem. Main Results. We propose a new approach to creation of odor system for detecting threats that makes it possible to improve technologies for ensuring the safety of life and people's activities. The working capacity of the proposed solution is demonstrated on the layout. It shows the possibility to use the technology of the Internet of things at the system implementation stage and to deploy their work on any territory, including hard-to-reach areas. The convergence of the results of the trained neural network with test sets of concentrations of hazardous substances in the air is shown. Practical Relevance. The odor system for threats detection can be useful as an element of an integrated solution to ensure the safety of people on any territory, depending on the tasks assigned. The odor detection system for dangerous substances has been brought to the layout, which makes it possible to detect such threats as the leakage of benzene, butane, methane, propane and the ignition at an early stage.
Keywords
Постоянный URL
Articles in current issue
- ON MODERN APPROACH TO AIRPLANE-TYPE UNMANNED AERIAL VEHICLES DESIGN WITH SHORT TAKEOFF AND LANDING
- SEARCH AUTOMATION OF BINARIZATION OPTIMUM LEVEL FOR SYNTHESIZED HOLOGRAMS
- INSCRIPTION PROCESS RESEARCH AND OPTIMIZATION FOR SUPERIMPOSED FIBER BRAGG GRATINGS
- SPATIALLY SELECTED SPECKLE-CORRELOMETRY OF TEMPERATURE DEPENDENT GELATION KINETICS
- CONVERGENCE SPEED INCREASE OF A FINITE-TIME STABLE CONTROL SYSTEM
- ROBUST PROPERTIES OF SYSTEMS WITH NEGATIVE DEGREE OF HOMOGENEITY WITH RESPECT TO DELAY
- COMPUTER SIMULATION OF INTERACTION OF LYSINE DENDRIMERS WITH STACK OF AMYLOID PEPTIDES
- INVESTIGATION ON CORRECTING PROCESS OF SPHERICAL ROTOR IMBALANCE AT THIN FILM DEPOSITION STAGE
- ALGEBRAIC MANIPULATION DETECTION CODES WITH PERFECT NONLINEAR FUNCTIONS UNDER NON-UNIFORM DISTRIBUTION
- DISTRIBUTION OF RESOURCES BASED ON HYBRID MODELS OF SWARM INTELLIGENCE
- SHAPE RESTORATION OF THREE-DIMENSIONAL OBJECT FROM TWO-DIMENSIONAL IMAGES
- AUTOMATED SIMULATION TECHNOLOGY OF MULTI-HOP WIRELESS NETWORKS IN HETEROGENEOUS MODELING ENVIRONMENT
- VARIANT OF ODOR SYSTEM FOR THREATS RECOGNITION
- ADAPTIVE SELECTION OF ARTIFICIAL IMMUNE SYSTEMS AND LOCAL SEARCH MUTATION OPERATORS
- NONLINEAR DYNAMICS AND STABILITY OF MICROSYSTEMS ENGINEER-ING ELEMENTS
- IDENTIFICATION OF PIEZOACTUATOR PARAMETERS
- SPECIAL DESIGN ASPECTS OF GRIPPING DEVICES FOR INCREASING THE FLEXIBILITY OF AUTOMATED AND ROBOTIC MANUFACTURING LINES AT INSTRUMENT MANUFACTURING FACILITIES
- AN APPROACH TO RELIABILITY INCREASE OF TURBINE SAFETY SYSTEM BASED ON HIDDEN RESERVES
- OPERATIONAL CONTROL METHOD FOR TRAIN INTEGRITY MONITORING BASED ON OPTICAL COHERENT REFLECTOMETRY DATA
- LOWER LIMB EXOSKELETONS: BRIEF REVIEW
- MODELING OF LOCOMOTOR SYSTEM DYNAMICS
- EXPERIMENTAL INVESTIGATION OF WINDING DIAMETER EFFECT ON OPTICAL PROPERTIES OF HI-BI FIBER WITH TIN COATING
- DIGITALIZATION OF PROJECT AND PRODUCTION PROCEDURES AS A TOOL FOR INSTRUMENTATION DESIGN AUTOMATION IN INDUSTRY 4.0