Airborne Multi-sensor Data Fusion Target Tracking Technology Research And Implementation

In the military field, in order to acquire information superiority, master activepower, and obtain the best campaign effect, it needs multi-sensor collaborative workinstead of a single sensor. And the development of multi-sensor data fusion technologyjust satisfies these requirements. As a classical application of multi-sensor data fusion,target tracking is the precondition for drivers to obtain war information such as targetinformation, assistant decision-making information in time and availably, since it hasimportant research significance and value. For this, based on the basic theory ofmulti-sensor data fusion for target tracking, the airborne aircraft is used as the researchplatform to achieve some correlative works in this dissertation. And the main works aresummarized as follows:(1) Data characteristic of several common airborne sensors are analyzed concretely,namely fire-control radar, electronic support measurement, infrared sensor, lasertelemeter. The airborne multi-sensor data fusion tracking system is analyzed anddesigned in detail. And the system is the research basis of the design andimplementation for the following simulation platform.(2) The distributed multi-sensor track fusion algorithms based on statistical theoryare researched. And several track fusion algorithms are introduced. Then, a weightedfusion algorithm based on local state estimations is proposed by using multi-sensorlocal state estimations so that the influence of uncertain and correlative measurementsfor the original weighted algorithms is decreased, and the problem of the decreasingvalidities of fusion algorithms by using multi-sensor measurements directly is solved.Besides, combining with the idea of the weighted fusion, a multi-layer fusion algorithmbased on supporting degree matrix is put forward, and it has useful value for theresearch of the asynchronous multi-sensor track fusion. In many scenes, the simulationresults show that these two algorithms not only decrease the influence of uncertain andcorrelative measurements, but also have well performance in target tracking.(3) The asynchronous multi-sensor track fusion in the distributed fusion structure isstudied. For the multi-sensor limitations of working and functionary range and the random noises in campaign environment, multi-sensor are working asynchronously inmany cases, such as the different sampling periods among multi-sensor and thechangeful sampling periods of a certain sensor. An asynchronous multi-sensor trackfusion algorithm based on filtering prediction is proposed to obtain accurate and reliabletarget information and solve these asynchronous problems. It not only avoids the errorof synchronous fusion by using data matching indirectly, but also achieves to solve thecomplex asynchronous issue of the changeful sampling periods for multi-sensor well.The simulation results show the algorithm is effective. In addition, in the asynchronousfusion tracking, the performance of the algorithm proposed in this dissertation is betterthan a step by step prediction fusion algorithm based on asynchronous multi-sensorsystem.(4) The designed requirements of the simulation platform are analyzed. Theairborne multi-sensor fusion tracking system simulation platform is designed, and itssubsystems are analyzed and designed in detail. At the same time, in order to operateabundant data conveniently, the data-base is designed to manage data. The systemsimulation platform is realized and tested by using Borland C++Builder (BCB)6.0IDEand airborne multi-sensor observations. The testing results show the system works well,and the global fusion tracking performance of the airborne multi-sensor is evaluateddirectly. Moreover, according to the varieties of sensors, it finds that the system hassome expansibilities and adaptabilities.