With the expansion of the scale and the complexity of the functions of colleges and universities, the efficiency and accuracy of the allocation of fire safety facilities resources have become the key to ensure campus safety. Aiming at the problems of aging facilities, uneven distribution of resources and lagging emergency response in traditional fire management mode, this paper puts forward a data-driven optimization framework for resource allocation of fire safety facilities in colleges and universities. Real-time collection of facility operation data through Internet of Things sensors, combined with big data analysis and machine learning algorithms, establishes risk assessment, facility performance attenuation prediction and dynamic resource allocation models to achieve accurate allocation and efficient utilization of resources.