With the development of network and control systems, the amount of data acquired and stored in aluminum electrolysis control systems has been growing continuously. How to effectively mine useful information hiding in these data is an attractive research topic that will help to enhance the functions of the control systems. In this paper data warehouse and data mining concepts were introduced into an existing aluminum electrolysis control systems and a data-mining tool called as “aluminum electrolysis data miner” was developed. It consists of three core mining components: the gray association analysis, the cluster analysis based on connected components and the offset coefficient analysis. For different uses and from different profiles, these component models analyze the data stored in a control system thoroughly, thus obtaining knowledge that reflects the status of every pot in a potline, adjusting various operating and control parameters online and providing decision support for operators.