MODELING OF RADIOPHYSICAL PROCESSES IN UAV MICROCONTROLLER SYSTEMS USING APPROXIMATION SCHEMES

Abstract

The paper presents deterministic and stochastic models of radiophysical processes in the microcontroller (MCU) systems of unmanned aerial vehicles (UAVs). The study aims to identify the parameters of a stochastic model based on empirical oscillographic measurements of the MCU supply voltage under various operating modes of the propulsion system. The methodology integrates two complementary stages. At the theoretical stage, a deterministic model of signal dynamics in state space is constructed, followed by a justification for including random perturbations in the form of Markov switching processes. Applying the averaging principle and diffusion approximation in the asymptotic limit of fast switching (ε → 0) leads to an Ornstein–Uhlenbeck stochastic differential equation for the supply voltage component P(t), which defines two key parameters: the relaxation rate α and the diffusion intensity σ. At the experimental stage, these parameters are identified through direct oscillographic measurements of the MCU supply voltage (using an FNIRSI DS215H oscilloscope and a SpeedyBee F405 V3/STM32F405 quadcopter) across four propulsion system load profiles (0 %, 5 %, 10 %, 40 %). The relaxation parameter α is determined by approximating the exponential decay of the transient response during a 0 % → 20 % → 0 % cycle, while the diffusion coefficient σ is derived from the stationarity condition (2 ). tms s =V ⋅ α The measurement results confirmed a statistically significant correlation between the propulsion system load and the intensity of stochastic voltage fluctuations. The identified values of α = 63,0 с-1 and σ ranging from 315,6 ⋅ 10-3 В ⋅ с-1/2 (10 % load) to 632,0 ⋅ 10-3 В ⋅ с-1/2 (40 % load) provide a quantitative basis for evaluating the reliability of control algorithms under stochastic electromagnetic disturbances. The stationary variance Var∞(U) = σ2/(2α) increases more than fourfold when transitioning from 10 % to 40 % load, serving as a direct quantitative indicator of decreased system reliability. The key contribution of this work is the synthesis of stochastic modeling with experimental parameter identification under realistic electromagnetic interference conditions.