This paper addresses the characteristics of the combustion system in a decomposition furnace, introducing fuzzy control systems and neural network models. By leveraging the respective advantages of both, the paper seamlessly integrates fuzzy control and neural networks to propose a multi-variable fuzzy neural network-based combustion control optimization algorithm. Through practical industrial applications, the paper demonstrates the combustion efficiency and optimization effects of the system’s multi-objective scheduling during model operation, achieving precise control of combustion in industrial furnaces. The results show that the use of this model can improve the combustion efficiency of industrial furnaces by up to 18.4% compared to before its application. When comparing the concentration ratios of fuel emissions using raw coal as an example, the algorithm reduced emissions by 60.23% and 85.1%, respectively, significantly reducing energy consumption and improving combustion efficiency. Additionally, when the decomposition rates were 65% and 100%, respectively, the unit fuel consumption of total fuel was 3583.67 kJ/kg and 4572.82 kJ/kg, respectively. Furthermore, for every 5% increase in system unit product fuel consumption, the increase was only 19.43 kJ/kg to 41.91 kJ/kg, indicating a decreasing trend in overall fuel consumption. The method proposed in this paper provides efficient and reliable theoretical and technical support for industrial practice.