Analysis And Prediction System Of Electromagnetic Environment Effects Based On Machine Learning

With the continuous development of military science and technology,unmanned equipment combat technology has gradually matured,and its advantages such as high survivability,low cost and high cost ratio are widely used in modern warfare.A large number of information technology equipment into the battlefield,making the electromagnetic environment has become more complex and diverse.UAVs are composed of multiple electronic components,which are highly integrated and susceptible to electromagnetic interference.Many countries have experienced UAV navigation electromagnetic decoys and electromagnetic interference crashes,which have caused widespread international concern.In order to ensure that the effectiveness of UAV operations in complex electromagnetic environments can be effectively played,it is necessary to fully understand and recognize the impact of complex electromagnetic environment effects on UAVs,so as to ensure the reasonable application of UAVs.Therefore,predicting interference under different electromagnetic environment effects becomes a key issue for conducting experiments on electromagnetic environment effects on UAVs.The paper focuses on the following aspects:In terms of data acquisition.Radiation interference experiments were conducted for the UAV Lidar detection system.The lidar in the UAV was used as the test device in the experiment,and the strong electromagnetic field was used to radiate interference to it.The data collected were detection angle,detection distance,interference frequency and interference field strength,and the data were clustered using the K-means algorithm.The analysis of the data reveals that there are problems of unbalanced samples,outliers and inconsistent magnitudes in the data.The SMOTE algorithm is used to equalize the samples,the box-line plot is used to find outliers and remove them,and the zero-mean standardization method is used to unify the magnitudes of the data.In electromagnetic environmental effect prediction,traditional methods of electromagnetic environmental effect prediction mostly use mathematical function models,but the traditional prediction methods have poor ability to deal with nonlinearity and weak generalization ability.The paper uses machine learning algorithms for electromagnetic environmental effect prediction research.The theory of fusion algorithm is introduced in the research process,and the electromagnetic environmental effect prediction model of Stacking model fusion algorithm is constructed.The model uses grid search and cross-validation to optimize the parameters,and uses Ada Boost(Adaptive Boosting),SVM(Support Vector Machines),DT(Decision Tree),XGBoost(Extreme Gradient Boosting),GBDT(Gradient Boosting Decision Tree),KNN(K-nearest Neighbor)and the constructed Stacking model fusion algorithms are compared for experiments.Three model evaluation metrics,Accuracy,Hamming distance and Kappa coefficient,are used to evaluate the prediction effectiveness of each model.The experimental results show that the prediction of electromagnetic environmental effects using Stacking model fusion algorithm is better.The paper designs and implements an electromagnetic environmental effects analysis and prediction system,which is written in Python language,built with Py Qt5 framework,and graphically displayed using Matplotlib visualization technology.