Approximately 280 million people worldwide suffer from depression, with over 700,000 suicides annually attributed to depression. Early diagnosis allows depression patients to receive timely treatment and intervention, alleviating symptoms, reducing the risk of deterioration, and lowering medical costs. However, current clinical diagnostic methods for depression suffer from subjective bias and low accuracy. Therefore, continuous research into more accurate and simple detection methods for depression using artificial intelligence is of great clinical significance. Inspired by existing methods, this paper proposes an innovative automatic depression identification method that combines electroencephalogram (EEG) signals with artificial intelligence technology. Firstly, the phase lag index (PLI) of EEG signals is computed to obtain their functional connectivity networks. Then, the elements within the PLI matrix are select based on the altered Kendall's Rank Correlation Coefficient and convergence of accuracy (AKRC-C). Finally, the selected multidimensional features are input into an RF classifier for automatic classification. This method achieves a classification accuracy of 95.03% for distinguishing depression patients from healthy controls, which is superior to the existing methods. This indicates that the proposed depression detection model in this paper can achieve intelligent and rapid depression detection, providing an efficient, accurate, and diverse solution for clinical depression detection.