Research On Algorithms And Applications Of Flush Airdata Sensing System

The flush airdata sensing system is a certain kind of flight data sensing system.The system uses a matrix of pressure sensors, arranged orderly, to measure the air pressure all around to estimate the airadata indirectly. Before entering into the airdata computer, the air pressure needs to transform and be sampled. And then the air pressure will be computed to get the final airdata with a certain kind of algorithm, memorized in the airdata computer.The aerodynamic model of flush airdata sensing system is nonlinear. On that basis, the final airdata can be acquired. In addition, the airdata can also be estimated by the nenual network trained by datas from the wind tunnel testing and flight testing, which needs to be researched further. Conventional algorithms of the FADS system are based on the aerodynamic model. In these algorithms, the iterative least squares algorithm has some problems with the stability, and by reason of this, the algorithm is risky when verhicles are at a high speed, what's more, some accurate means need to be used to acquire the initial airdata. The"triples"algorithm has some problems with ill-conditioning points and an array of"triples"frames when it is used to estimated the attack angle and sideslip angle, what's more, the algorithm is not as robust as the least squares algorithm to small purtations. The process of the iterative algorithm, used to estimate the dynamic pressure and static pressure, is complicated, owing to the participation of the shape and compressibility parameter and calculations of matrix inverse.However, improved algorithms for the flush airdata sensing system are presented in this paper.1. The combination of the least squares algorithm and"triples"algorithm is developed. Three ways of acquiring initial airdatas are given, and the flow of the failure detection algorithm is analysized.2. The idea of estimating the optimal solution of attack angle and sideslip angle is developed, and some emulational caculations are carried out. The ill-conditioning points' problem is discussed carefully with an example.3. A means, based on the calculation of overdetermined linear equatons, is developed to estimate the dynamic pressure and static pressure. The means is proved that the matrix is full rank, so the singular value decomposition is avoided. Finally, the airdata is obtained by means of the trained neural network before flying.4. A means, based on the Kalman algorithm in which the shape and compressibility parameter is avoided from the iterative algorithm, is developed to estimate the dynamic pressure and static pressure. The algorithm is not as complicated as the iterative algorithm. And the algorithm is robust to small putations.5. The algorithm based on blocking neural networks is developed. It is used to train the function relationship between the parameter t and the shape and compressibility parameter, so as to estimate the dynamic pressure and static pressure. The algorithm is used to calibrate airdatas to improve the calibration precission. And the algorithm can also improve efficiency of searching some datas for FADS system.Finally, the electroniclly scanned pressure module is used to measure air pressures in the wind tunnel testing. And the frequency response model of the measureing air pressure system is developed, and the designing rule of the measureing air pressure system is put forth.