UAV-based Real-time Tracking Method For Dynamic Target In Oilfield Inspection

As an emerging inspection technology,UAV target tracking technology has been widely used and promoted in the fields of weather forecasting,agricultural production,network communication,and power production.There are two crucial problems in target tracking which are target trajectory prediction and real-time tracking trajectory optimization.With the transformation of oilfield production management to digitization,automation and intelligence,the traditional methods of manual inspection and fixed-point video surveillance cannot meet the needs of modern oilfield management.Thence,considering the character of oilfield production,it is significantly meaningful to utilizing UAV to conduct the real-time tracking for dynamic target in oilfield.Technically speaking,dynamic target trajectory forecast is the problem that the current position and speed information of dynamic target is acquired by airborne-sensor so as to predict its trajectory in the next several sampling periods,which is one of the vital jobs in UAV real-time tracking.Under this background,the dynamic target tracking prediction methods are studied in this paper based on the environment of oilfield production field,and the main research content are as follow:First,the oilfield production environment is modelled,and the UAV motion model and UAV target tracking model are constructed by analyzing the flight constraints of UAV and the motion state of dynamic target.Besides,a single-step trajectory prediction method of dynamic target is proposed and there are three evaluation indicators used to estimate the forecast accuracy of target trajectory prediction.Then,consider the more complex oilfield production environment,that is,the dynamic target trajectory prediction problem when the oilfield production environment is composed of multiple working platforms.Through further analysis of the motion characteristics of the intrusion dynamic target,a two-dimensional mathematical model that can capture the behavior of the intrusion target is constructed.Besides,according to the Kalman filtering algorithm,a multi-step Unscented Kalman Filter(UKF)algorithm is proposed to forecast the multi-step trajectory of dynamic target in next several sampling periods.What's more,the optimization of UAV real-time tracking trajectory which is the most crucial difficulty in dynamic target real-time tracking.For real-time tracking trajectory optimization based on single UAV,BA algorithm is taken by comparing several optimization algorithms,and the initialized mechanism of population in FOA algorithm is utilized to modified classical BA algorithm,so as to propose a new algorithm called FOBA algorithm,which is verified its effectiveness through the comparison between different simulation of diverse algorithms.Finally,based on single UAV dynamic target tracking,multi-UAV cooperation real-time tracking trajectory optimization is further considered.In this situation,due to more UAV and more complexed oilfield production environment,the calculation amount of real-time tracking trajectory rises.And FOBA algorithm have shown some drawbacks such as poor optimization accuracy and larger optimization time,when it was taken to solve above problem.Therefore,a new population updated mechanism is proposed to modified traditional FOA algorithm,so as to present a new algorithm called MIEFOA algorithm,which acquires superior trajectory optimization accuracy and shorter optimization time that meet the real-time constraints of real-time trajectory optimization.Besides,the effectiveness of MIEFOA is verified by simulating the optimization effects of eight different algorithms,at the same time,the method based on infinite Markov chain is used to verify the convergence of the MIEFOA algorithm and analyze the complexity of the algorithm.