Abstract: In the operation of intensity magnetic separation process (IMSP), it's difficult to use any accurate mathematical models to describe the dynamic characteristics such as the strong nonlinearity, severe coupling and time variability between the technical indices, namely the concentrate grade and the tailing grade, and the key controlling variables, i.e. the exciting current, the rinsing water flow and the feed density. Moreover, these two indices cannot be measured continuously. So, the conventional model-based control approach can't be used here. Focusing on this practical challenge, this paper proposes an intelligent setting control approach that consists of a case-based reasoning(CBR) control loop pre-set module, a soft-sensor module based on principal component analysis(PCA) and radial basis function neural network(RBFNN), and an rule-based reasoning dynamic compensator. This intelligent system can automatically adjust the operating points for the IMSP in response to the changes in boundary conditions online. Industrial tests show that the approach proposed can improve the concentrate grade, while reduce the tailing grade.