Micro gas turbine (MGT) is widely used for its high efficiency in meeting the demand of combined cooling

heating

and power in small-scale scenarios (e.g.

off-grid power generation

energy centers and smart microgrids

etc.). This study focuses on the application of natural gas (NG) and hydrogen (H2) hybrid fuel in MGT systems

proposing and validating a modeling approach for MGT that integrates physical mechanisms with data-driven methods. Various learning algorithms are used to predict the key performance parameters of MGT

and bagging regression (BR) and random forest regression (RFR) algorithms are found to have high accuracy. The evaluation indicators R2 and MAPE are both within the range of 0.75 to 0.85. Through the Taguchi-grey method

this paper employs a multi-objective optimization of the fuel mixture ratio and system operating conditions to achieve the synergistic control of power generation efficiency improvement and energy consumption reduction. The system performance is optimized at low temperature (T1＜ 288.15 K) with hydrogen volume fraction (HVF) of 25% or 50%. The study provides theoretical support and an optimization path for the application of hybrid fuel MGT in off-grid power systems.