FILTERING ALGORITHM OF ACCELEROMETRICAL SENSORS MEASUREMENTS IN PLATFORMLESS INERTIAL NAVIGATION SYSTEMS UNMANNED AIRCRAFT

In unmanned aerial vehicles of low mass, platformless inertial navigation systems are widely used. They are implemented on the basis of accelerometers and gyroscopes, which are made according to the technology of microelectromechanical systems. The low accuracy of microelectromechanical systems determines the use of additional stages of navigation measurement processing in navigation systems. To increase the accuracy of navigation determinations, stochastic filtering algorithms are used, namely the Kalman filter and its various modifications. Existing filtering algorithms are characterized by high computational complexity due to the abstract form of presentation, which does not reflect the details of implementation. That is why the task of synthesizing filtering algorithms that will meet the requirements of guaranteed convergence of the filtering process and minimal computational complexity regarding its implementation is relevant. This requirement is extremely important for navigation systems of small unmanned aerial vehicles, as their on-board equipment must be cheap and low-energy. The work is devoted to the synthesis and research of the algorithm of polynomial filtering of measurements of accelerometric sensors in platformless inertial navigation systems of unmanned aerial vehicles. The synthesis of the algorithm is performed according to the method based on the presentation of smoothing filters as dynamic systems described by discrete transfer functions, which are determined by the application of the third form of invariance conditions. A distinctive feature of the synthesized algorithm is the consideration in the filtering process of not only current, but also previous measurement results, which are weighted by their own smoothing coefficients. The conditions for the convergence of the filtering process are defined for the synthesized algorithm. Due to the scalar form of the algorithm implementation, low computational complexity is inherent. The effectiveness of the algorithm was confirmed by the results of computer simulation based on the results of real measurements of the ADXL345 accelerometer, which is part of the Arduino UNO R3.