Radar Emitter Signal AFMR Section Intelligent Search And Sorting Feature Extraction

For the wide application of the new radar system, exploring effective sorting method of radar emitter signal is the key issue that must be addressed. A long time, the radar signal sorting method is mainly dependent on the time of arrival (TOA), the radio frequency (RF), pulse width (PW), pulse amplitude (PA), and the direction of arrival (DOA) five classic parameters, for today’s complex and variable electromagnetic signal environment, these methods can not achieve satisfactory sorting results. Therefore, to explore the characteristic parameters of the emitter signal to compensate for the lack of the classic five parameter’s sorting capacity is considered to be one of the effective ways to solve the problem of signal sorting.Because ambiguity function can describe the structure of the signal information more completely, Pu Yun-wei proposes an extraction method based on characteristic parameter of the ambiguity function of the main ridge (Ambiguity Function Main Ridge, AFMR) slice. The parameters extracted by this method have strong compactness in a cluster and good separateness between cluster and good anti-noise performance; so that separation methods based on these parameters have good sorting results. However, the method to extract ambiguity function of the main ridge slice require excessive calculation, is not conducive to the real-time processing.Because the method mentioned by Pu Yun-wei has the shortcoming of the large time complexity, this paper proposes a quick and intelligent method to search ambiguity function main ridge slice. Then, study the AFMR slice depth, explore new sorting parameters, and design the two extracting methods based on features of AFMR slice. Finally, the clustering results research the performance of the extracted characteristic parameters. The main work and research results are as follows:1. This paper introduces the theory of ambiguity function for radar emitter signal, the relationship between self-correlation function and ambiguity function, the method that calculating radial section through the origin of ambiguity function by fractional autocorrelation. Then research the method of extracting ambiguity function main ridge slice quickly and accurately. With simplicity and robustness of the genetic algorithm, this paper designs an intelligent search method based on superiority inheritance to find AFMR slice, and gives specific flow chart of the method. Introduce the details of the design process of the method, and describe several aspects of the design ideas, for example,uniform initialization, the adaptive genetic crossover and mutation probability, the s neighborhood elitist preservation, as well as secondary search. Finally, using the theory of statistical hypothesis testing, analyze and text AFMR slice’s data extracted by the algorithm. The experimental results show that the intelligent search method compared with the method proposed by Pu Yun-wei can be faster search AFMR slice. As accuracy has been greatly improved, the AFMR slice more ideal. As the signal-to-noise ratio (SNR) is not less than OdB, the AFMR slice search success rate is more than96.5%.2. With studying the AFMR slice in depth, because of feature information of its graphics, this paper designs a sorting algorithm based on the cumulative angle and unit cumulative length feature. In different SNR conditions, analyze the similarity of the same signal’s AFMR slices, extract two characteristic parameters——accumulated angle and units accumulated length. Then use the fuzzy K-means clustering methods to cluster signal, and evaluate clustering results. Experimental results show that, when SNR is not lower than OdB, the signal clustering success rate is94.3%or more. The characteristic parameters have strong compactness in a cluster and good separateness between cluster and good anti-noise performance.3. Because of the energy distribution information of the AFMR slice, this paper designs a extracting algorithm based on energy accumulation point. When SNR is OdB,6dB,12dB,20dB, extract energy accumulation point feature of the signal. Use these parameters to cluster signals by the fuzzy Kmeans clustering method. The experimental results show that, when SNR is not lower than6dB, can achieve better clustering results.