What priority area of science and technology does it correspond to: Fundamental scientific research on the most important problems of developing scientific, technical, socio-economic, socio-political, human potential to ensure the competitiveness of Ukraine in the world and the sustainable development of society and the state.
Research: targeted fundamental.
Perspective of further implementation: will be continued as applied.
Research level: no analogues in the world.
What additional actions require further research: Implementation requires further funding.
Brief description, advantages, further prospects for application.
A prototype of a new optical infrasonic sensor has been created, in which the feedback is calculated by a built-in microprocessor. A set of optimal algorithms has been determined that implements the gradient descent method, the method of neural networks and adaptive Kalman filtering of the functioning of an acoustic reconnaissance and signal system. An algorithm and software for the procedure for recognizing acoustic images were created, which uses an artificial neural network with three layers.
The input vector is a line that envelopes the spectra of the amplitude-frequency characteristics of signals from infrasound sensors that are part of the acoustic reconnaissance and signal system (ARSS). It is shown that the use of digital infrasound sensors makes it possible to use only one digital communication channel (wired, wireless or combined) in ARSS, which significantly reduces its cost and reduces installation time. The accuracy of determining the parameters (distance, azimuth to the source and time) of the event by the experimental group (acoustic antenna) is sufficient for location. It has been established that the indicated technical means register signals of natural and technogenic origin with sufficient quality for determining the parameters of disturbances. It is shown that sensors that distort input signals to a lesser extent provide significantly better performance in algorithms for detecting, locating, recognizing and tracking targets.
It has been established that for the problem of locating objects in on-line mode, it is advisable to use the CHAP solution, which has significant advantages in the speed of algorithm execution than the statistical method according to the Fisher criterion.
It has been established that the Kohonen neuron grid with one hidden layer is able to effectively recognize acoustic and seismic images by the envelope spectrum of oscillations. According to experimental data, the efficiency is approximately 95%, which meets modern requirements for reconnaissance and signal systems for the needs of the army and security agencies. The directions and measures for the creation of miniature precision sensors of mechanical vibrations for the needs of the army, security systems and geophysical research are determined. Created within the framework of the project, new ultrasound sensors with improved amplitude-frequency characteristics (AFC) and digital transmission and data processing can significantly improve the efficiency of ARSS in locating, tracking and identifying targets at low altitudes compared to existing systems. Using a precision calculation of the feedback signal using a microcontroller and adding it to the input signal, sensors with significantly improved characteristics in the infrasonic region have been created.