Distillation is a process not stationary, non-linear, and multivariable, which presents difficulties in control system design. The use of process simulators facilitates the development of control strategies for processes with these characteristics. Artificial neural networks are systems of artificial intelligence able to collect, store, and use information based on previously provided data. With breakthrough technology, they harve developed new models of identification strategies. Among these new strategies stands out the use of artificial neural networks in the identification and control of nonlinear dynamic processes stand out. The neural predictive control is a control strategy that used a control law based on artificial neural networks to predict future responses of the plant and promote control actions able to maintain process stability. The purpose of this work is to implement a distillation column for the separation of propane and propylene in Aspen Plus simulation environments and Aspen Plus Dynamics, and a multivariable control strategy by developing a neural predictive control system. From the communication between a block diagram created in Simulink and simulation of the distillation column in the Aspen Plus Dynamics, it was possible to implement and evaluate the performance of neural predictive control strategy, which presented satisfactory results in process control. With this, it follows that the methodology presented here serves as an aid in the model identification procedures and implement predictive control strategy, based on the technique of application of artificial neural networks.