Target Intention Recognition Of Electromagnetic Threat

Intent identification is a key technology in the military field,which is the basis of situation assessment and risk judgment,and plays an important role in assisting commanders to make rapid and accurate decisions.In order to improve the ability to study and judge the threat target intention,the study and judge of threat target intention is developing in the direction of intelligence.With the rapid increase of battlefield information and the complexity of confrontation between the two sides,it is difficult to obtain the mapping relationship in the knowledge graph solely by manual method to be capable of real-time and accurate determination of the tactical intention of the threat target from the multi-source battlefield data.In the background of electromagnetic countermeasure,this paper takes electromagnetic target as the research object and studies the intention recognition method of electromagnetic threat target based on the existing open literature results,combined with data analysis and knowledge reasoning.The main contents and experimental conclusions of this paper are as follows:(1)Study the electromagnetic threat target,assume the electromagnetic threat target's confrontation scenario,and make a detailed analysis of the electromagnetic threat target's parameters,attributes,characteristics and other aspects.Combining the two perspectives of electromagnetic target behavior and external target related intelligence,the behavior rules under the corresponding intention are set and the characteristic sequence of time series events is mined.(2)Electromagnetic threat target intention recognition based on data analysis.From the point of view of data analysis,according to the timing characteristics of the threat target behavior,a cyclic neural network which can mine the timing characteristics is established to realize the electromagnetic threat target intention recognition.Firstly,the timing characteristics of electromagnetic threat targets are established,the network method is used to optimize the super parameters,and the combination of super parameters with the best comprehensive performance is calculated.The validity of the combination is verified by experiments,and the advantages and disadvantages of the model are mined.(3)Knowledge-based inference for electromagnetic threat target intention recognition.From the perspective of knowledge reasoning,the target attributes and factors of electromagnetic threat are analyzed.Through knowledge extraction and knowledge mining,the deep features of simulation samples are mined,and the generalized knowledge graph of electromagnetic threat target is established.By integrating the knowledge in the knowledge base and mining the relationship between knowledge and intention,the target intention recognition of electromagnetic threat is realized.Based on the analysis of data and knowledge reasoning intention recognition model was constructed,the super parameter configuration to the network,the optimization results with test samples to test intention recognition method of data analysis,using the knowledge base for knowledge reasoning intention recognition method to test and evaluate all intention recognition method,analysis the advantages and disadvantages of different approaches,by analyzing the results,An intention recognition method based on knowledge and data association is presented,and the performance of the new model is evaluated by using the multi-index evaluation method.The simulation results show that both BI-LSTM network and collaborative filtering intention recognition method have good performance in electromagnetic threat target intention recognition.Although the recognition rate is very close,they have different performance for different intentions.Combining the above two methods to establish the intention recognition model combining data and knowledge can complement each other.More features can be noticed in the recognition process,and the recognition results are more accurate.The effectiveness of the intention recognition model based on knowledge and data association is verified by comprehensive evaluation of the new pattern.