Abstract: Micro gas turbines, with their flexibility and rapid response capabilities, can serve as flexible dispatch resources in integrated energy systems, filling the energy gap caused by the randomness and intermittency of weather conditions and providing grid support for distributed renewable energy power plants. However, the conventional control methods struggle to maintain good global performance and lack the ability to deal with constrained problems. To solve this problem, a self-adaptive model predictive control (MPC) algorithm is proposed in this paper and a dynamic model of micro gas turbines based on the MATLAB/SIMULINK is built to verify the performance of algorithms. The results show that the average relative errors of the constructed model in standalone mode and grid-connected mode are 0.08% and 0.52%, respectively. The proposed algorithm reduces the average relative error by 36.67% and 88.05%, respectively, compared to proportional integration differentiation (PID) control and single linear model based MPC control when responding to multi-energy complementary system scheduling. With excellent control performance and good compatibility with scheduling algorithms, the new proposed algorithm showcases great development potential.