Research On Key Technologies Of Fusion Recognition And Sensor Management Based On Satellite Constellation

Fusion recognition and sensor management is the key technologies of space-based optical warning system. According to the characteristics of satellite constellation in space-based optical warning system, this paper conducts research on the fusion recognition algorithm and sensor management strategy from the following three aspects.1. Aiming at the recognition process of space targets in multi-sensor system, this paper puts forward a decision level fusion recognition algorithm based on adaptive fuzzy integral. According to the actual circumstance of satellite communication of space-based optical warning system of satellite communication, decision level fusion recognition is chosen. The basic definition and the nature of fuzzy measure and fuzzy integral is given, and combining with the information fusion theory, the paper analyzes the uncertainty of the sensor, and defines the adaptive fuzzy measure at this time for decision level recognition. Based on this, the paper puts forward a decision level fusion recognition algorithm based on adaptive fuzzy integral, which uses static prior knowledge, as well as the judgment results of each sensors uncertainty, and corrects the fuzzy measure adaptively, effectively improves the fusion recognition rate of multi-sensor system.2. After the IR dual band characteristic extraction of space targets, the paper puts forward a feature level fusion recognition algorithm based on adaptive fuzzy integral. Considering the space-based optical warning system relies mainly on the infrared dual band sensors on satellite for space information acquisition, the paper analyzes the dual band infrared radiation characteristics of space targets, compares with the radiation characteristics of the bait and other interference, and extracts the average gray level distribution and the gray level change rate distinguished features. Based on this, the paper proposes a feature level fusion recognition algorithm based on adaptive fuzzy integral, in which using the description quality and decision quality of the characteristics to define fuzzy measure, and by for adaptive processing of fuzzy integral, making the recognition results can be corrected dynamically, at last further improves the ability to recognize the space target from the bait and other interference of the space-based optical warning system.3. According to the characteristics of the space-based optical warning system, the paper establishes a sensor management model in fusion recognition process, and puts forward two solutions to solve the problem: the method based on greedy algorithm and the method based on particle swarm optimization algorithm. On one hand, the model of sensor management based on satellite constellation is established. The paper defines the optimization objective function through the quantization of target priority and the combined performance of target and sensor, and introduces the constraint conditions by observability analysis and so on. On the other hand, to solve the optimization model, the paper introduces the greedy algorithm and particle swarm optimization algorithm, and proposes the sensor management method based on greedy algorithm and the sensor management method based on particle swarm optimization algorithm. Both algorithms can carry on the reasonable distribution of sensor resources and make full use of them, and the analysis of the characteristics of these two methods, and the comparison of the respective advantages and disadvantages of performance, provide the accordance of the selection of sensor management strategy in practice.