Abstract: The performance of the heavy medium separation system used in traditional coal preparation plant is not so ideal as expected owing to the fact that the related operating parameters of the system cannot get appropriately adjusted. To address the issue, the methods for processing raw coal data and online prediction data are proposed, so as to make it possible to make real-time prediction of clean coal yield and separation density based on raw coal data, and to control heavy medium separation process based on online prediction of production indicators. A feedforward and feedback combined control mechanism is thus realized. Application of the heavy medium separation process intelligent control strategy at Linhuan Coal Preparation Plant has led to great enhancement of intelligence level of the separation process.