Zaman Frekans Analiz Yöntemleri ile İmha Değerlendirme
xmlui.mirage2.itemSummaryView.MetaDataShow full item record
Kill assessment is an capability that evaluates the success or failure of a missile interception of a threat by providing the basis for further decisions in a defense system. The most important advantage of the assessment is to reduce the costs of the defense system by reducing the number of missiles that will be fired to destroy a threat. The need for an assessment for small ranges is much less with respect to long ranges because assessment can be made visually by the operator in small ranges . In long ranges, the fact that the kill or miss point is too far from the observation point causes the need for a separate algorithm to make the assessment. In this study, radar data was used as input of assessment and since there is no radar data of the moment of kill in the literature, the study started with a simulator design and that produces synthetic radar data for different miss and kill scenarios. In order to detect both the range and speed of the air targets, pulse Doppler searching radar was based on. Pulse compression methods were compared to increase range resolution performance, and the linear frequency modulation pulse compression method was decided by evaluating the advantages and disadvantages. LockHeed Martin's RRP-117 Radar is referenced in the technical data used in the simulator. For the processing of radar data, some of the time-frequency analysis methods were compared based on couple of parameters such as resolutions and computation time. The studies about Ambiguity Analysis in the literature are summarized by emphasizing the similarity with the Wigner Ville distribution, which was chosen as the favorite time frequency method. Matched filter, Wigner Ville distribution and ambiguity analysis signal processing methods were compared by using example target signals. As a result of processing, it is aimed to obtain delay and Doppler frequency, so range and speed, of the air targets detected by the radar. It was decided to use ambiguity functions in the study due to the distortions in the matched filter output in the presence of a moving target, the interferences encountered in the use of Wigner Ville and the loss of resolution performance caused by the filtering used to prevent interference. In order for the thresholding to be adaptive, the CFAR algorithm was applied to the ambiguity function outputs and feature matrices were extracted on the scenarios by applying feature extraction algorithms to these outputs. Analysis graphs were drawn using noiseless and noisy environment assumptions and interpreted in the performance tables considering the detection rate and accuracy parameters in order to assess kill. Also these feature matrices were given as input to the another thesis work in order to give the kill assessment decision using machine learning and deep learning methods. The analysis graphs provided the representation of the reflected radar signals of the air targets in the delay and Doppler frequency domains and made it possible to make some assumptions about miss/kill scenarios. For example, it has been interpreted that reflections with varying delay values at the same Doppler values may belong to the same target, while reflections with reduced delay difference values at two different Doppler values may be two targets converging to each other. It has been evaluated that after the point where the delay difference between the two targets is minimum, the increase in the delay differences without the change in the Doppler values may be a result of the miss scenario. Also after same point, revealing of more than two air targets with different Doppler and delay values may be result of the kill scenario. In the noiseless environment assumptions, only the errors arising from the analysis were emphasized, and in the noisy environment assumptions, first, the effect of the different antenna gains, so indirectly SNR value, and after the effect of thresholding on the analysis was compared and interpreted due to the CFAR application applied for different false alarm rates. In this way, it was concluded that with the selection of an appropriate false alarm rate, the kill assessment evaluation can be made with the analysis graphics extracted from the radar signals.