Abstract: As a high-energy-consuming process in mining and metallurgy，grinding process with good control effect can greatly reduce the production cost of the process and enhance the profits of enterpris⁃ es. However，the grinding process is a strong coupling，nonlinear，multi-input and multi-output process， which is not only affected by many factors，but also susceptible to external interference. Meanwhile，it is unable to establish accurate mathematical model，and the production process often adopts manual adjust⁃ ment. In order to solve the problem of timeliness and accuracy of manual adjustment，the model-free adap⁃ tive control（MFAC）was studied and applied in a semi-autogenous milling and ball milling mine. Based on the input and output data and the good robustness of the system，a model-free adaptive control system with multi-input and multi-output is designed combining on-site actual. The grinding concentration and mill power were controled by adjusting the feed water and feed ore of the semi-autogenous mill，achieving good control effect. Field data simulation and field application show good results.

Key words: model-free adaptive control, semi-autogenous mill, grinding process