A Real-Time Computing Platform for UAS Dynamic Simulation and Control Testing

This paper presents a low-cost, multi-rate, real-time distributed computing platform for integrated dynamic simulation and controller testing of unmanned aerial systems (UASs) with coupled electro-mechanical (or magnetic), aero-, thermal and control dynamics. This platform consists of modules for dynamic simulation, control implementation, data visualization, and communications, by leveraging off-the-shelf design and testing tools, and is suitable to rapid virtual testing of aircraft system dynamics and control, including mission/path planning, flight control, propulsion, and energy management. Testing of these functional components in a high-fidelity virtual environment will help make corrections or refined updates to the control algorithms and software and provide confidence to the user, before finally moving to field test, because the UAS testing is very costly and involve safety concerns. This platform is based upon heterogeneous computing hardware, including real-time multi-core processors (running Linux system and PX4-based flight simulation/control/visualization) and field programmable gate array (FPGA - a reconfigurable, parallel computing device). The mathematical models of different components are solved by a set of multi-rate computational solvers and running on multiple hardware units with different simulation time steps (or rates), depending on the model's time scale. The innovative computational solvers are based on the first principles and can also capture fast electromagnetic transient processes occurring in aerospace systems. A universal power converter model will be discussed and can be implemented for different power converters. Also presented is a generalized electrical machine model that can solve a system of differential and algebraic equations in parallel on FPGA, and its main components include input data sampling, mechanical dynamics calculation, abc-to-dq transformation, flux and current calculations, torque or horsepower calculations, dq-to-abc transformation, and data output. The controllers can be implemented on either real-time processors running Linux system, or FPGA computing modules, or remote hardware controllers. An FPGA-based control prototype is discussed in detail. A case study will demonstrate the use of this platform and show the results of a flight simulation case, involving dynamics of several key components in an autonomous flight mission. An advanced UAS power generation system including power converters and electrical machines will also be studied on the developed real-time computing platform to validate the solver accuracy.