Biotechnological processes are very complex systems for modeling, due to the presence of live microorganisms, nonlinear behavior and time-variant features. Many parameters of the process are uncertain and the industrial operating conditions are determined empirically for them. This paper presents a robust dynamic simulator in Matlab/Simulink environment from the work of Scaglia and the study of experimental data for analysis purposes of the fermentation process. The problem with these processes is given by the difficulty of obtaining data online. In this regard, it should be noted that the dominant state variable, given by the substrate concentration in the medium, is very difficult to measure, due to the strong parametric uncertainty and the presence of significant nonlinearities. In addition to this, the measuring of amount of bacteria is typically performed through a culture procedure that requires a significant amount of time. Thus, a state observer is developed from experimental data and the Scaglia's Model, using neural networks, in order to facilitate the estimation of state variables values. This produced satisfactory results for a network with five neurons. The observer could, therefore, estimate in a short time: the amount of substrate, bacteria, CO2 and ethanol in the sample.