Intelligent Target Clustering Technology Based On Deep Learning

With the application of a large number of battlefield observation sensors,the battlefield target data inputted into command and control system is increasing rapidly.The large scale,high dimension and complex structure of target information data bring new challenges to the target clustering technology.However,the traditional clustering method has been unable to effectively cluster the high-dimensional battlefield target data.Aiming at the current difficulties of target clustering technology,this paper first proposes a general clustering model framework based on deep learning.This model framework embeds clustering technology into the deep neural network model,which can extract features from high-dimensional battlefield target data sets and realize clustering.On this basis,a target clustering model based on deep stack self-coding network is constructed.Target clustering algorithms DAE-k and DAE-G are designed to synthesize the advantages of traditional clustering algorithm.Through designing the validation experiment of intelligent target clustering technology,the evaluation index of target clustering effect is established.In the experiment,several battlefield target datasets from different fields and space are selected.After data preprocessing,target clustering algorithm is implemented on Tensor Flow artificial intelligence development platform.The experimental results of DAE-k and DAE-G demonstrate the validity and applicability of the two target clustering algorithms for large-scale and high-dimensional target clustering problems.The research of intelligent target clustering technology based on in-depth learning can make the command and control system more efficient and intelligent in dealing with large-scale high-dimensional data,and also provide technical support for the intelligent construction of command and control system.