Traditional power grid knowledge mining,just on the grid database mining,the discovery process is a relatively isolated and closed,and then ignore the knowledge innovation,this paper proposes a new Grid knowledge discovery strategy,it will grid database and grid Knowledge Base organic linking,build collaborative grid based on double bases cooperating mechanism knowledge discovery model found,depending on the power grid knowledge base for a scalable and dynamic system evolving,mining,grid knowledge base and the grid at the same time on the grid library database synchronization evolution, change and optimize the traditional grid knowledge discovery process and operation mechanism,the multi-source focus and reduce parity amount to improve grid efficiency of mining. Verified by an example of the model grid short-term load forecasting knowledge discovery process with a comprehensive evaluation of the level of synergy,to achieve prediction,optimization or refinement updated forecast knowledge.