Helicopter radio system for low altitudes and flight speed measuring with pulsed ultra-wideband stochastic sounding signals and artificial intelligence elements

Abstract

The subject matter of this study is algorithms for measuring the components of an aircraft speed vector and altitude. The goal of this study is to improve algorithms for processing wideband stochastic pulse signals in helicopter low-altitude and flight-speed radio systems by introducing secondary signal processing based on artificial intelligence elements. The tasks to be solved are as follows: to develop an optimal algorithm for determining the speed and altitude of flight for a helicopter radio complex; to supplement the signal processing algorithm with an artificial intelligence-based processor to determine the "safety" of the current trajectory; provide the pilot with relevant information about possible options for further actions based on an analysis of the current position of the helicopter and flight parameters; and to analyse the efficiency of the proposed complex when using various artificial intelligence-based algorithms. The methods used are as follows: methods of mathematical statistics and optimal solutions for solving problems of statistical synthesis of active radio complex structure; methods of machine learning; and methods of computer simulation. The following results were obtained. The algorithms for signal processing in a helicopter radio complex are obtained by the method of maximum likelihood, and the use of three radio channels to calculate the full vector of speed and altitude is argued. The structure of a secondary information processing system, using algorithms based on artificial intelligence is proposed. The effectiveness of determining the safety of the current landing trajectory using various algorithms based on artificial intelligence (LinearSVC, GaussianNB, DecisionTreeClassifier, RandomForestClassifier, KNeighborsClassifier, MLPClassifier and RidgeClassifier) was analysed. Conclusions. The simulation results show that in the presence of accurate (noise-free) information on the current location of the helicopter, its axial velocities, and a map of the terrain with defined areas dangerous for landing, the DecisionTreeClassifier and RandomForestClassifier algorithms can provide a high probability of correctly determining the safety of the current landing trajectory. At the same time, in the presence of instability in the measurements of helicopter movement parameters, only the RandomForestClassifier algorithm maintains high accuracy.