Learning system of multi-source track association machine

Abstract

The invention discloses a learning system of a multi-source track association machine, belongs to the field of multi-source information fusion, and mainly solves the problem that the existing track association model requires a lot of manual repeated debugging and is difficult to be directly applied in the actual engineering application. Firstly, historical track data of information sources are collected and an association relation is manually analyzed and studied and judged to form an original database. Then, vector formation of training samples is set, data of associative and non-associative samples are calculated and generated, track association training data are formed, a training data set is preprocessed and then a standard training data set is generated. Finally, the model is trained, verified and overparameterized and tuned by using a learning model of a binary classification machine and adopting appropriate training and verification methods, so that the track association model is generated. The system automatically trains and generates the track association model, completely avoids a large amount of manual debugging for model parameters, and has the advantages of high model generation speed, good practical effect and the like.

Description

technical field [0001] The invention belongs to the field of multi-source information fusion, relates to the learning and generation of multi-source flight track correlation models, and is suitable for radar networking systems and multi-source information fusion systems. Background technique [0002] Track association is to discriminate and process multi-target tracks reported by different detection methods or systems, in order to achieve correct merging of multi-source tracks for the same target, and provide a basis for simple de-duplication or further fusion of subsequent tracks to ensure the presentation of the situation One target uniquely corresponds to one track, which guarantees the unique characteristics of the target track. If track correlation processing is not performed or the processing effect is poor, it is inevitable that one target corresponds to multiple tracks, which seriously affects the accuracy of subsequent situation analysis and judgment. Therefore, tra...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to propose a multi-source track correlation machine learning system, aiming to solve the problem that the existing track correlation method needs to be manually modified and debugged repeatedly using measured data in practical application

Images