Research On Comprehensive Storage And Analysis Technology Of Test Data For Traveling Wave Tube

Because of its high power,high efficiency and high gain in microwave power amplification,the traveling wave tube(TWT)has become a core device widely used in weapon system and communication system.With the development of national defense,higher requirements are put forward for the reliability of TWT.Burn-in test is an important screening method to ensure the reliability of TWT.It tests the various indicators of TWT by providing the corresponding voltage and applying various necessary stresses to the TWT in the appropriate environment.It improves the reliability of TWT by removing the unqualified products and reduces the loss caused by the failure of TWT after leaving the factory.However,due to the long time of burn-in test and various types of test data,the current manual monitoring and recording data is very labor-intensive.And it can not detect the abnormal state of TWT in time,and there are also have problems in data storage.Consequently,it is very important to establish a set of automatic monitoring program for the burn-in test of TWT.This paper mainly studies the burn-in test of TWT.Combining the communication protocol of the test instrument for data collection,designing an effective algorithm for data analysis,and adopting an appropriate method for data storage,a set of automatic monitoring procedures for the burn-in test of TWT is established.The program can collect the test data periodically,carry on real-time analysis to detect the changes of state for TWT in time,and store the data in a structured way,which overcomes some problems existing in the current burn-in test.The main work is as follows:First,the automatic collection of test data is completed.Through the TCP/IP communication protocol of the high-voltage program-controlled power supply,the collection of voltage and current data at all levels in the burn-in test of TWT is completed.Based on the visa library and some SCPI commands,the collection of power data is completed.Second,the analysis algorithm of test data is designed.Through the investigation of data feature extraction and pattern representation and anomaly detection methods,a real-time analysis algorithm which can effectively detect the abnormal changes of TWT is designed by combining the piecewise linear representation and the LOF algorithm.And the effectiveness of the algorithm is verified by the actual test data.At the same time,the segmented error limit of piecewise linear representation and the reference set of LOF algorithm are determined by using the R test method.The control limit of LOF algorithm is determined by kernel density estimation.The KD tree is used to improve the calculation speed of LOF algorithm and is applied to the overall analysis algorithm.Meanwhile,the effectiveness of the overall analysis algorithm in abnormal state detection is verified by the actual test data.Third,the structured storage of test data is completed.Through the analysis of the data in burn-in test,the storage format and specification of the data are determined,the data table and the sequence information table for burn-in are established.Instead of the current local storage formats,the local storage of data is completed using the CSV format.About databases,the structured storage of data is completed using the HTTP interface of the time series database InfluxDB.