Research On Muti-sensor Information Fusion Technology Of Flight Control System

For the purpose of autonomous flight and executing related tasks for unmanned aerial vehicle (UAV), all states must be accurately acquired by the flight control system. Thus, the performance of UAV can be ensured. Due to the limitation of technology and conditions, the measuring precision of aeronautic sensors were limited, so single sensor had difficulty to obtain information of high precision, this can influence the performance of flight control system. In order to solve this problem, multiple sensors were installed on UAV to measure the same signal, obtained more measuring information, then the information was synthetically handled to get more precise values. This process that multi-sensor measuring values were synthetically handled was called multi-sensor information fusion technology.According to the theory of the information fusion and the requirement of the flight control system, this paper took the multi-sensor information fusion technology apply to the flight control system of an UAV. In view of the characteristic of altitude and attitude angles, adopted the Kalman filter method to design the altitude multi-sensor information fusion, and adopted BP neural network method to design the attitude angles multi-sensor information fusion. To satisfy the design requirement, the fusion algorithms were implementated, using C code, integrated with original flight control code to apply to the flight control system.Both altitude and attitude angle fusion algorithms were validated using the sensor simulation data and flight test data, the result indicated that the accuracy of information acquisition were improved in the system, consequently enhanced the performance of the flight control system. By calculating the algorithm execution time, the execution time of Kalman filter algorithm one time was in the range of 10-6～10-5 second magnitude, and BP network algorithm was about in 10-6 second magnitude. According to the system real-time requirement, it concluded that the information fusion algorithm completely matched the realtime request. In addition, the algorithm didn't increase the hardware structure, thus the system integrity and reliability were ensured.