Research On Signal Sorting Algorithm In Cognitive Reconnaissance

In order to deal with the complication of radar system and the development of cognition,the structure and concept of cognitive electronic warfare have been put forward in many countries.In the process of Cognitive Electronic Warfare processing radar pulse signal,when the received radiation source signal mixing degree and the complexity is high,the traditional electronic reconnaissance in the aspect of processing efficiency and processing accuracy will be gradually cannot meet the requirements,in this context,many researchers on the basis of the cognitive structure of the Electronic Warfare puts forward the concept of cognitive reconnaissance,paper focuses on the cognitive reconnaissance module were studied.The cognitive reconnaissance proposed according to the current structure and function,the paper from the aspects of electronic warfare database and signal separation to improve the existing electronic reconnaissance,and database with cognitive ability is put forward based on convolution neural network parameters of single pulse signal sorting concept,relative to the cognitive reconnaissance of the existing electronic reconnaissance in the signal processing efficiency and improved signal processing time.In order to promote the processing efficiency of signal sorting in cognitive reconnaissance,the thesis puts forward a database concepts of cognitive ability,combining with the clustering algorithm based on hierarchical from index to spy on the existing database call form and improvements on the store content,through the use of algorithm,simulation shows that the database with a cognitive signal separation process has improved the efficiency of data to be obtained compared with the existing reconnaissance database,and with the increase of amount of data the more obvious advantages.The thesis to solve the existing for preassigned selected error is too big lead to the final separation of error signal,current signal sorting preassigned selected and used by the main sorting parameters in combination with,put forward the concept of single pulse parameter signal sorting and will be the arrival time difference of matrix,as a new separation parameters will eventually sorting accuracy improved.In this paper,the arrival time difference matrix calculated in the signal sorting of monopulse parameters is further processed to obtain the maximum common multiple vector,and an improved fuzzy C clustering algorithm is proposed and simulated to solve some clustering fuzzy problems.Different from the pulse descriptors used in current signal sorting,the pulse descriptor difference matrix calculated by the single-pulse parameter signal sorting method proposed in this paper can obtain a larger amount of single-pulse data.On this basis,the signal sorting method can be combined with the convolutional neural network algorithm in deep learning,so that the final sorting accuracy can be stabilized at more than 90% and around 97% under the condition of sufficient data.In the whole paper,the problems of the two key links in the cognitive reconnaissance module,namely,the reconnaissance database and the signal sorting,are improved,and the traditional signal sorting and the algorithms used are simulated.Finally,the research on radar signal sorting in cognitive reconnaissance is completed by comparing the running time and running accuracy.