Abstract: To address the issues of premature convergence and low search accuracy encountered by particle swarm optimization (PSO) in complex global optimization and robot path planning tasks, a particle swarm optimization algorithm based on the spin glass model (SG-WCPSO) is proposed. This algorithm integrates the particle spin states, random coupling matrix, and temperature regulation mechanism from the spin glass model into the PSO framework. By leveraging the complex interactions and multi-stability character of the spin glass model, the proposed method significantly enhances the exploration capability of particles in complex energy landscapes. Furthermore, based on the dynamic particle spin states, the algorithm adaptively adjusts its inertia weight and acceleration coefficients, effectively preventing premature convergence and improving global optimization performance. Experimental results on function optimization demonstrate that the search accuracy of SG-WCPSO is improved by up to 145 orders of magnitude compared to PSO and other benchmark algorithms. In path planning simulations, SG-WCPSO shortens the average path length by up to 11.07% and 14.69% on 20×20 and 30×30 grid maps, respectively, compared to standard PSO. Additionally, in 3D path planning experiments, the average fitness value is reduced by 11.10%. These results fully validate the high efficiency and strong robustness of the proposed method in solving mobile robot path planning problems.