Aiming at the problems of traditional control algorithms in maximum power point tracking (MPPT) for thermoelectric generation systems (TEG), such as easy trapping in local optima and overshoot in dynamic response, an enhanced Harris Hawk hybrid optimization algorithm (GEG-HHO) is proposed. By establishing simulation and experimental models of Bi-Sb-S based thermoelectric generators, the tracking performances of Constant Voltage Tracking (CVT), Salp Swarm Algorithm (SSA), standard Harris Hawk Optimization (HHO), and GEG-HHO are compared and analyzed under the conditions of constant hot-end temperature、sudden mutation of hot-end temperature and Sudden Change in Module Quantity. The results show that under both working conditions, GEG-HHO can enter a stable state within 0.052±1s, with an average tracking accuracy of 98.27%, and there is no oscillation or overshoot. Compared with the comparative algorithms, its power oscillation suppression capability is significantly improved, and the dynamic response speed is enhanced.