Central integration of wind farmmakes it hard to effectively compute Total Transfer Capability (TTC) through traditional way. For this reason,a data mining technique named Differential Evolution Extreme Learning Machine (DE-ELM) is proposed to extract operating rules for the TTC of tie-lines in wind-integrated power systems. Representative operating scenarios are firstly determined by K-medoids clustering under the two-dimensional “wind power-load consumption” feature space. Then knowledge base for TTC operation rule mining is generated by stochastic sampling and repeated power flow. Secondly, to reduce the ultra-high dimensionality of operating features, RELIEF-F algorithm is employed to screen the redundant features and identify the features that are strongly correlated to the TTC. Finally, the TTC operation rules are extracted from the knowledge base by feeding training data into the DE-ELM. Numerical results show that the proposed method can fast estimate TTC with satisfying accuracy and strong generalization. This work is supported by National Natural Science Foundation of China (No. 51437003) and Science and Technology Project of State Grid Sichuan Electric Power Company (No. SGSCJY00JHJS201700009).