LEAST SQUARES AND CLUSTER ANALYSIS BASED METHODOLOGY FOR SONIC BOOM LOCALIZATION OF ROCKET DEBRIS

Aiming at the transonic sonic boom localization problem generated by the falling rocket debris, this paper proposes a spatio-temporal localization method of debris based on acoustic monitoring. The sonic boom signal is received by multiple monitoring devices, and the effects of time error and equipment layout on positioning accuracy are analyzed to construct a single target and multi-target cooperative positioning model. For single wreckage, the four-sphere intersection principle and the least-squares method with error compensation are adopted to realize the accurate solution of three-dimensional coordinates and sonic boom moment through four monitoring devices, and the localization results are 110.57°longitude, 27.16°latitude, 957.96 m elevation, and 100.753 s. For multi-wreckage scenarios, the classification model of vibration wave signals is proposed, and the classification model of vibration wave signals is proposed by combining the Pearson correlation coefficient and the K-means clustering algorithm. A vibration wave signal classification model is proposed, and a time difference threshold constraint (≤5s) is introduced to optimize the wreckage matching, which effectively solves the problem of overlapping multi-source sonic boom signals. The results of this paper show that the proposed method significantly improves the robustness and accuracy of rocket debris localization in complex scenes, and can provide theoretical support for the rapid confirmation of the drop point.