DOA estimation method and system based on accelerated local weighted K-means clustering

The invention discloses a DOA (direction of arrival) estimation method and system based on accelerated local weighted K-means clustering. The DOA estimation method specifically comprises the following steps: firstly, establishing a DOA estimation system model based on a novel heterogeneous sublink hybrid analog and digital large-scale MIMO (multiple input multiple output) receiving array; receiving signals from a plurality of antenna arrays, pre-processing the signals, grouping the pre-processed signals, and calculating the Cramer-Rao lower bound CRLB of each sub-array group; next, a root-MUSIC method is used for each sub-array group to obtain a candidate solution set, and an accelerated weighted K-means clustering algorithm is used for the initial candidate solution set to obtain a real solution of each group through local search and recursive update; and finally, combining the real solutions obtained by all the sub-array groups, and obtaining final DOA estimation through a weighted average algorithm. Accord