An efficient adaptive dynamic hierarchical equilibrium optimizer CGTEO was proposed to address the problems of the equilibrium optimizer (EO) algorithm that was prone to fall into local extremes and poor optimization search accuracy when dealing with complex optimization problems, and its convergence was analyzed both theoretically and experimentally. An adaptive cross-updating mechanism based on sine-cosine coefficients was introduced to enhance the population diversity. A dynamic hierarchical search strategy was incorporated to balance the different needs of sub-populations for exploration and exploitation capabilities. An elite neighborhood learning strategy based on triangular topological units was incorporated to improve the convergence accuracy and effectively avoid local extremes. The global convergence of the CGTEO algorithm was demonstrated through the probability measure method. CGTEO and nine representative comparison algorithms were comprehensively tested and comparatively analyzed by using the CEC2017 test set. The optimization results were evaluated by combining various methods such as optimization searching accuracy, convergence curves, Wilcoxon rank-sum test and violin plots. The experimental results show that the CGTEO algorithm exhibits outstanding performance in optimization precision, convergence capability and stability. The Wilcoxon rank-sum test indicated that the optimization results of the proposed algorithm were statistically significantly superior to the other compared algorithms.