The work aims to study the path optimization of reverse logistics to reduce the enterprise costs and promote the development of reverse logistics for returns. With the focus on the path optimization of reverse logistics, a vehicle routing optimization model aiming at minimizing recovery costs was constructed by fully considering factors such as transportation costs and penalties for violating time windows. To overcome the limitation of the Equilibrium Optimizer (EO) algorithm being prone to local optima, the EO algorithm was combined with the variable neighborhood descent for improvement. The improved EO algorithm was then compared and analyzed against the Simulated Annealing algorithm and the original EO algorithm. At the same time, the original EO algorithm was also compared and analyzed against the variable neighborhood descent. The optimized EO algorithm reduced delivery time by 8.82% compared to the SA algorithm, and decreased the total cost by 4.63%. Compared to the EO algorithm before optimization, the delivery time decreased by 1.40%, and the total cost decreased by 3.55%. The improved EO algorithm has better adaptability and convergence in solving the vehicle path optimization model, which can effectively reduce costs and shorten path and time.