Application Of Multi-Sensor Information Fusion Technology In UAV Height

With the rapid development of information and intelligence in modern society,UAVs and related technologies have been applied and studied in a wider range,which also poses new challenges to the control accuracy and safety performance of UAV flight control system.The UAV flight control system is a kind of multi-sensor control system,so multi-sensor information fusion method,as an effective means of processing multi-source data,has been widely used in UAV altitude control system.Kalman filter is one of the most common methods in information fusion technology because of its stable performance and simple calculation.However,traditional Kalman filter is difficult to develop in most practical applications due to various limitations.In order to improve the accuracy of the UAV height control system in the height direction and ensure that the UAV can effectively track the real state in the event of a fault,based on the principle of Kalman filter,this paper proposes a multi-layer multi-source information fusion method of improved Kalman filter and a strong tracking fusion method of improved extended Kalman filter,which ensures the accuracy and robustness of UAV altitude flight system.The work of this paper is mainly reflected in the following two aspects.(1)The improved Kalman filter multi-layer information fusion method is introduced.Three kinds of altimetric sensors commonly used in UAV flight control systems are selected,and the corresponding altitude measurement models are established by combining the measurement characteristics of different sensors,so as to do the preliminary work for altitude data fusion;Then,the recursive weighted least square method is used to fuse the de-noising height data of the three sensors to achieve the first level fusion of the improved algorithm.Then,the Kalman filter is used to fuse the fusion results of the first three sensors in the second level,Compared with the data without fusion algorithm,the root mean square error of the height estimation is reduced by 50.7%,and the maximum deviation is reduced by 59.8%.It can be demonstrated that the positioning accuracy of the final fusion result in the vertical direction is effectively improved,And initially have the ability to deal with the abnormal situation.(2)An improved strong tracking filtering method based on extended Kalman is introduced and applied to fault detection and diagnosis of UAV flight system.In order to effectively track the real flight state of UAV,the theory of strong tracking filtering algorithm is introduced by extending Kalman filter: One is to introduce a new kind of strong tracking filter with single suboptimal fading factor.The new information sequence is used to calculate the time-varying fading factor,and the filter gain matrix is modified in real time.The algorithm focuses on skipping the update estimation of model parameters,and directly realizes the modified estimation of state;Secondly,a new kind of strong tracking filter with multiple fading factors is proposed,which can eliminate different data channels through multiple fading factors to further improve the tracking ability of the system.The calculation of multiple fading factors does not take the traditional method of using prior information to design the proportion,but uses a new relatively accurate suboptimal algorithm to calculate the fading factors,which not only realizes the online update of fading factors,but also greatly saves the calculation workload,The results show that the improved sub optimal fading factor strong tracking filtering algorithm has better performance and accuracy.