Abstarct: Nowadays, organizations record their activities and events by information systems such as Enterprise Resource Planning (ERP) systems and Business Process Management Systems (BPMS). Process mining is a field wich handles process discovery, process monitoring and improvement of business processes from event logs. Intelligent management of business processes leads to higher efficiency, higher flexibility and quality assurance. Each event in the event log is specified by a set of properties such as time information. Time control of events of process instances using historical information is a challenge for researchers of this domain. Time control of events includes remaining time prediction of process instances, alignment of arrive time of process instances, synchronization of process instances by controlling lateness and earliness. Investigating state-of-the-art approaches, five approaches are proposed in order to predict remaining cycle time of process instances; four of which are baxsed on path decomposition technique and analytical path formulation which leverages Support Vector Machines. The other is a probabilistic approach taking queues of tasks into consideration. Experiments show that the proposed adaptive approach decreased normalized MAE error by 12.83% in total over 8 datasets compared to multi regression approach. Then, concept drift is also added to assumptions of the problem to make it closer to real-life settings. A statistical approach baxsed on a transition system annotated with Fuzzy SVM probabilities is presented to predict the remaining time of business processes. The proposed dynamic multi-model approach to concept drift adaptation improved the accuracy by 22.56% in total compared to static single model approach over two datasets. Finally, a processing dependency between pairs of dependent process instances is formally defined and extracted from event log. To resolve the issue, time synchronization algorithm is presented to synchronize and control the events of process instances. Experiments on BPI challenge 2012 dataset show 4.3% reduction in overall cycle time and 39% reduction in dependent process instances.