An approach to the creation of a digital ecosystem (DES) consisting of interacting services-subsystems (consorts) is presented. Methods for analyzing and prediction of the production situations using machine learning algorithms and inductive knowledge bases are proposed. The possibility of using these methods to predict the state of the digital infrastructure resources for industrial enterprises in real time is demonstrated. It is shown that methods of proactive decision support for situational control help to prevent abnormal situations. Case Studies are presented on the results creation of the association rules search for the problem of the state predicting for an enterprise resources complex.