A New Method Of Information Fusion Evaluation System Based On Feedback Optimization

Data fusion,a crucial technology of multi-target tracking system,has been widely utilized in modern warfare.Generally,it is important to conduct performance testing and evaluation before being put into practice.However,the conduction of field test is often constrained by high cost,uncontrollable accident and high design difficulty.Thus,developing effective test scene and evaluation system is significantly important.This thesis focuses on establishing a fusion evaluation system and designing a simulation test scene.Firstly,a novel method of establishing track complexity function is proposed based on fusion system and evaluation system.Then,by utilizing track complexity function,the evaluation system,fusion filtering and scene generation are integrated into a system in order to realize scene parameter adjustment.From the perspective of engineering practice,a complete information fusion system performance evaluation system is designed.Last but not least,a new method of building a benchmark library for fusion evaluation system is proposed based on the above algorithms.The main contents of this thesis are presented as follows:(1)A scene generation algorithm based on track complexity is proposed.Both the single flight simulation and multi-objective mixed scene simulation are realized by utilizing kinematics and dynamics track generation algorithm.Meanwhile,warfare environment simulation method is also designed.Fusion simulation system is developed to establish the track complexity function which reflects the relationship between scene parameters and track difficulty.Based on obtained function,the scene parameter can be adjusted efficiently.(2)Fusion evaluation index establishment algorithm and evaluation software are proposed.From the perspective of engineering practice,a complete information fusion performance evaluation system is designed.Meanwhile,comprehensive evaluation result of the evaluation system is achieved by AHP.Then,the performance evaluation software platform is developed based on the above evaluation index system.Sensor data analysis,sensor comparative analysis and data association evaluation analysis constitute the main function.(3)By synthesizing the algorithm proposed above,a new method is proposed to establish benchmark library of fusion evaluation system.Six classic tracks along with the typical tracks are integrated into the benchmark library.Finally,a software which contains the function of scene customization generation,benchmark scene extraction and scene display evaluation is developed based on this algorithm.This thesis detailedly presents the above research results in core chapters.In order to show the practical engineering value of those presented algorithm,in the end of each chapter,the performance of proposed algorithms are tested by simulation experiments.