The reference parameters of sprays for i.c. engine are of macro-geometric type, like penetration, cone angle, or fluid dynamic one (velocity, particle size). In this work, the spray is assimilated to a dynamic system and defined through the time value of dynamic variables. The spray behavior is represented in a “phase diagram” that describes its operating points. The development of this methodology is carried out using experimental data of the spray evolution captured by a fast image acquisition system. An 8-hole ECN injector spraying iso-octane was chosen as a case study. It was characterized through the time variation of the cone angles, taken as dynamic discriminating variables of its behavior. Images were captured at high sampling rate and processed according to the theory of “ergodic” systems. Cone angles, derived from images and processed by neural networks algorithms, are represented in the “phase diagram” in order to detect stable behavior and not. Furthermore, they are depicted on the k-ω diagram determining ordered and chaotic structures and defining clearly the energy content as a function of the entropy level. This analysis allows determining anomalous operating conditions and predicting not optimal combustion conditions. This potential can be fruitful as the basis of advanced monitoring and control systems.