Intelligent Analysis Of Movement Law Of Radar Signal Target

The analysis of the movement law of the radar signal target is one of the key technologies of the cognitive electronic warfare.The traditional analysis technology relies too much on the experience of operators and lacks integrity.Combining the intelligent method,this thesis designs an intelligent analysis method for the movement law of the radar signal target.The method analyzes radar full pulse signals to handle targets mining,anomaly detection and trajectory prediction.Specifically,the main work of this thesis are as follows:1.A key targets mining method suitable for Pulse Description Word（PDW）is proposed.This method firstly extracts the time stamp and TOA in PDW,uses TDOA technology to locate the targets.Secondly,the data is sliced into time slots,which are in the order of seconds,then a set is built according to the target ID that appears in each time slot.Finally,the Apriori and FP-Growth algorithms are applied to mine the key targets for a period of time.The experimental results show that the proposed method can accurately obtain 8 frequent items and 9 association rules set in the experimental scene.The problem of mining and monitoring key targets in the fixed area through long-term observation data is solved by the proposed method.2.A target trajectory anomaly detection method based on Deep Convolutional Generative Adversarial Network（DCGAN）is proposed.This method extracts normal movement trajectory information based on DCGAN,learns its movement law,and completes target anomaly detection according to the probability output by the discriminator.The experimental results show that,compared with the method based on Support Vector Machine（SVM）,the proposed method can improve the Accuracy and F1-Score by 10% and 9% respectively when detecting abnormal trajectories of offshore targets,both reaching 96%.The problems of low accuracy and high probability of false alarm in traditional methods are solved by the proposed method.3.A trajectory prediction method based on Long Short Term Memory neural network（LSTM）is improved by combining with attention mechanism（ATTENTION）This method expands the input data into eight dimensions（speed,heading,and 6 forward trajectory points）,and inputs them into the LSTM model combined with ATTENTION for training,and obtains the best weight distribution of the input data in this scene to predict subsequent trajectories.The experimental results show that,compared with the LSTM method,under the same conditions,the proposed method reduces the average error by about 50% when predicting the trajectory of abnormal targets.The problem of large prediction error in traditional methods is solved by the proposed method.