An ontological synthesis of models for interpreting non-stationary dynamics in onboard intelligence systems operating in the emergency computing mode (Urgent Computing - UC) is considered. The ontological system of a dynamic knowledge base is formulated on the basis of theoretical models of the modern catastrophe theory (СT). The analysis and forecast of the evolutionary dynamics of a non-stationary object is implemented in a multiprocessor computing environment. Within the framework of the ontological system, an approach to the interpretation of non-stationary dynamics using fractal geometry and the theory of dynamical systems stability is formulated. The practical application of the developed ontology model is discussed in relation to the interpretation of the interaction of a marine dynamic object (MDO) with the external environment at a given time interval. The dynamic СT model determines the motion of the MDO system to the target attractor and in case of stability loss. Examples of the implementation of the ontological synthesis of non-stationary dynamics in safety systems for navigation and landing of ship-based aircraft are given.