Real time data-based wind model for a Venus Aerobot: development and testing

The surface of Venus is characterized by extreme temperature and a crushing pressure. However, at altitudes between 51 to 62 km, conditions resemble Earth's surface, making it an ideal location for a proposed mission concept that involves deploying a balloon with a suspended science gondola, known as an aerobot. Due to the planet's extreme wind conditions, with gusts reaching up to 100 m/s, and the complexities of balloon dynamics in the atmosphere, detailed models of wind gusts are critical to predict the aerobot’s dynamics. Unfortunately, previous missions have only collected surface wind data and vertical gusts, leaving horizontal gusts data missing. The goal of this paper is to present the development of a systematic method to generate wind gust model capable of capturing the highly non-stationary and random characteristics of wind gusts from real-time data. The model is fitted to data collected during terrestrial flight tests, so it can be used in predictive simulations of the Venus aerobot behavior in Earth field experiments. The proposed approach relies on a set of stochastic differential equations, specifically a bidimensional Ornstein-Uhlenbeck process, to accurately represent the autocorrelation function and probability density function of a measured wind signal. The first function captures the “memory effect” of the signal, while the second one serves as a tool for reproducing the observed instantaneous distribution of wind direction and speed. The proposed method was tested using real-world wind speed measurement data collected by Jet Propulsion Laboratory during flight tests in the Mojave Desert. An extended Kalman filter was used to process the real-time wind signal captured by measurement instruments before incorporating it into the model. This filter adeptly captures the dynamics of an atmospheric balloon, and efficiently fuses data from inertial measurement units and wind measurement instruments. Results indicate that the proposed method is simple to implement and can accurately capture simultaneously the autocorrelation and probability distribution of wind speed measurement data, and holds promise as a tool for design of future Venus aerobots.