Research On Radio Frequency Interference Suppression Method Of Solar Radio Dynamic Spectrum

Solar radio radiation carries a large amount of information and has a rich representation in the spectrum.The fine structure of the radio burst is related to the physical origin and can be used to diagnose the shock process and particle acceleration.Observing and studying solar radio burst process has important scientific value and also has unique application value in space physics-space weather research.The project team developed a high-resolution radio spectrum analyzer for fast and efficient analysis of solar radio bursts.However,in the meter band,there are a large number of radio frequency interference signals in the space,mainly due to various radio signal interferences,which seriously interfere with the reception and analysis processing of the solar radio signals.How to suppress interference and get a clear solar radio dynamic spectrum map is the problem that this thesis tries to solve.The method for filtering interference in the existing solar radio spectrum is mainly processed by the hardware circuit filtering method before the AD conversion of the receiver,and the signal and the interference are simultaneously filtered without distinction;the subsequent software processing appears,when there is strong interference,it is difficult to distinguish between the two signals.According to the characteristics of interference signals,this thesis proposes a software suppression method on the digital level.According to the magnitude relationship between the interference signal strength and the burst signal strength,the interference signal is divided into strong interference,target interference station and weak interference.Analyze various signal characteristics and screen and process them.Among them,the target interference radio is the main processing target of this thesis,and the other two interference signals are processed in combination with the existing methods.In view of the timing characteristics of the target interference signal,it is treated as a timing signal and predicts the possible strength of the radio station signal in the burst area.Assuming that the signal is linearly addable,it is numerically operatedwith the intensity of the corresponding region,and the burst signal strength is obtained in the reverse direction.In this way,the purpose of suppressing interference can be achieved under the premise of retaining the explosive information as much as possible.After step-by-step prediction of the signal using the traditional time series prediction model-Autoregressive Integrated Moving Average Model(ARIMA)and Recurrent Neural Network(RNN),it is found that the RNN performs better.But the error caused by prediction method will continue to affect the results of the network prediction.After that,according to the numerical characteristics of the target interference radio,the RNN method based on digital mapping is designed to transform the time series prediction problem into the time segment classification problem,and the prediction of the radio value is realized.The comparison of the experimental results of the three prediction methods shows that the RNN method based on digital mapping can effectively suppress the step error and improve the accuracy of prediction.In order to further embody the advantages of the RNN method based on digital mapping,it is applied to the simulated burst event separately from the traditional RNN method.The simulated burst event is constructed according to the linear addition principle of the signal,The comparison results show that the former performs better in suppressing interference.Apply the designed method to the actual outbreak event,briefly analyze the excessive suppression situation in the actual situation,and compensate the interference suppression area with the data of the event itself.After the other two kinds of interference signals are processed by a common method,a spectrum with little interference is obtained by combining the image enhancement method.Finally,comparing the results of the outbreak event processing between the commonly used algorithm and the design method of this thesis,it is proved that the latter has better performance in interference suppression and can achieve the expected goal of this thesis.The research work in this thesis can effectively suppress the interference in the solar radio spectrum.After processing the radio values in the thesis,it can be found that the prediction problem is transformed into the classification problem for processing,which can effectively improve the prediction accuracy.Compared with the processing at the image level,the interference suppression processing on the digital level not only retains more effective information,but also leaves more room for subsequent event analysis.At the same time,it provides new ideas and directions for deep learning in astronomical anti-jamming processing.