Abstract: The accumulation of coal gangue not only poses significant environmental hazards but also results in substantial resource wastage, making the enhanced resource utilization of coal gangue imperative. The addition of biomass proves effective in improving coal gangue combustion performance, with co-combustion emerging as a viable approach for simultaneous treatment of coal gangue and biomass. Artificial neural networks (ANNs), recognized for their capacity to model highly complex nonlinear relationships, have demonstrated notable efficacy in predicting combustion experimental outcomes. This study investigates the co-combustion characteristics and gaseous products of walnut shell-blended coal gangue through thermogravimetric analysis (TGA), thermogravimetric-Fourier transform infrared spectroscopy (TG-FTIR), and kinetic analysis. An ANN model was employed to predict thermogravimetric data, establishing a definitive correlation between the industrial analysis parameters of samples and residual mass fractions during co-combustion. Results reveal distinct differences in combustion processes between coal gangue and walnut shells, with the latter exhibiting superior combustion characteristics. The incorporation of walnut shells significantly enhances the combustion performance of coal gangue blends. Kinetic analysis indicates that walnut shell addition facilitates reduction in activation energy. Comprehensive evaluation through TGA, kinetic analysis, and gaseous product characterization suggests an optimal mass blending ratio of coal gangue to walnut shells at approximately 3:7.The ANN model predictions show good agreement with experimental values, demonstrating that the model can be applied to predict the co-combustion characteristics of multi-source solid waste and biomass and to design co-combustion reactors, thus facilitating the reduction of costs in industrial applications.