Research On The Attitude Estimation And Fusion Algorithms For Initial State To Level Flight Of Small Parachute Aircraft

In recent years, the small aircraft has been profoundly researched in the United States, Britain, Israel and other countries, and particularly in America, related projects have been already launched in fields including army, navy and airforce. Compared with conventional tactical weapons and small UAVs(Unmanned Aerial Vehicle), the small parachute aircraft has incomparable advantages and great research significance. For the small parachute aircraft, the initial state to level flight process is the initial stage before the small aircraft begins mission flight. During this process, firstly the navigation system completes the initial attitude estimation, and subsequently provides the attitude information for the use of controlling the initial-level flight and the stable flight. The purpose of this paper is to conduct researches on attitude estimation and fusion algorithms. The main contents are as follows:(1)Initial-level flight process includes parachute steady stage and initial-level stage. Firstly, based on the exterior ballistics theory for small aircraft, the trajectory environment of the two stages was analyzed and motion feature was obtained. Furthermore, the external trajectory of one certain type of small aircraft was simulated, and the simulation results proved the correctness of the motion feature results. Secondly, according to the trajectory simulation results, the mission demand and the design requirements, the prototype of the navigation system was designed. And the random error types of the sensors, which affects the accuracy of attitude estimation, were analyzed. Finally, based on the error propagation equation, the attitude estimation errors of the two stages were calculated, and the attitude estimation accuracy of navigation system was also verified preliminarily.(2)Regarding the initial attitude estimation for parachute steady stage, an improved optimal vector attitude estimation algorithm was presented. Firstly, the judgment basis of steady stage was proposed, and the bounded-variable Gauss Newton algorithm was studied. Secondly, the new unit vector measurement model was established, and the statistical characteristics of the unit vector measurement was obtained; according to the Wahba problem, the weight of each unit vector measurement for attitude estimation was deduced, for the purpose of minimizing the attitude estimation error variance. Finally, combining the new unit measurement vectors, weights and common optimal vector attitude estimation algorithm, the improved optimal vector attitude estimation algorithm was derived.(3)Focusing on the stability control requirements after the initial-level flight stage, in order to improve the attitude estimation accuracy, the gyro-free multi-sensor quaternion fusion algorithm and the gyro multi-sensor attitude fusion algorithm were proposed. Firstly, the discrete Kalman filter was introduced, and the effects of noise produced in filter process was analyzed in detail. Secondly, the relatively independent “loose combination” fusion structure among multi sensors and the attitude fusion modelling were introduced. Thirdly, in order to reduce navigation system cost, the gyro-free multi-sensor quaternion fusion algorithm was researched. Finally, considering the disadvantages of navigation systems without gyros, the gyro was taken as the main sensor and the gyro multi sensor attitude fusion algorithm was derived. Simulation and test results showed that the two attitude estimation algorithms were both able to improve the attitude estimation accuracy, and the gyro multi sensor attitude fusion algorithm was proved to have higher accuracy and to be more suitable to estimate attitude of the small aircraft during stable flight stage.(4)Considering the effects of noise model on results during the fusion filter process, the characteristics of the MEMS IMU random error were researched, and the indirect inference wavelet variance method was derived to estimate the random error model parameters. Firstly, the statistical characteristics in the time domain of the Allan variance method were analyzed, and the advantages of this method in identification of the MEMS IMU random error source were obtained. Secondly, regarding the disadvantages of the Allan variance method in parameter estimation, the wavelet theory was adopted to analyze the random error, and the method for determining scale of wavelet multi-resolution analysis was presented. Finally, combining the wavelet variance and indirect inference theory, a new method to estimate the random error parameters was suggested. Simulation and test results indicated that, compared with estimated results by Allan variance method, the new parameter estimation method not only improved the accuracy of parameter estimation, but also effectively handled the issue concerning parameter estimation of complex random error statistical model.(5)Based on the above work, the research results were utilized in the small parachute aircraft navigation system, and the vehicle test and the flight test were separately designed to verify the system performance. Vehicle test results showed that, indirect inference wavelet variance method was more applicable to estimate the random error model parameters. In flight test, the initial state to level flight process was simulated, and the results indicated that the designed navigation system and the proposed algorithm in this paper well satisfied the control system requirements of small parachute aircraft in the initial state to level flight process.