Forming a Method for Determining the Coordinates of Sound Anomalies Based on Data From a Computerized Microphone System

Abstract

Special features in the process of determining the coordinates of sound anomalies according to the sound series were considered. Sound anomalies have been shown to be a source of information about events, phenomena that are already occurring, or are their harbingers. It has been stated that audio interception systems complement thermal imagers and provide savings in financial and human resources when used in combination with the benefits of unmanned aerial vehicles. The methods facilitating solution of the problem of surveillance and prediction by finding coordinates of sound anomalies were presented. Indirect methods for solving problems of searching for sound anomaly coordinates with three microphones in accordance with a linear scheme of approximants and linear and quadratic approximation were proposed. Solutions were reduced to analytical complete expressions that make it possible to calculate coordinates according to the input conditions for three or four microphones. Also, a problem of finding coordinates of a sound anomaly for three and four microphones was set and solved by direct methods. Solutions were presented as expressions that make it possible to calculate coordinates of sound anomalies. Numerical experiments were performed during which coordinates of sound anomalies, the absolute error of their determination at each iteration, and total time spent for calculation were found. Systems with coordinates of microphones and sound sources almost identical or coinciding have been shown to have the greatest error. Under these conditions, for direct methods, values of the equation coefficients decreased to almost zero or turned to zero and the difference of values of the sought coordinates between iterations increased sharply which slowed down the process of solution convergence. It was shown that the application of approximate methods to the search for coordinates by solving the minimization problems with the involvement of the recurrent approximation method makes it possible to construct simple algorithms. Their implementation to solve the problems of numerical experiments has given quick and practically accurate coordinate values. It was found, that the application of algorithms to the methods of logical analysis and rules of logical inference reduces the number of iterations and the total calculation time