There are more and more production tasks with the distribution grid more and more quick growth. However, the traditional outage planning process is very complex and the plan will be adjusted from time to time. It is very difficult to guarantee the outage plan to be scientific and reasonable, which will result in numerous duplicated outage, power sales lost and low work efficiency. To solve the above problems, this paper studied the key technologies for outage plan optimization in distribution grid. First, this paper proposed a three-layer optimization framework, namely department optimization, coordination optimization and global optimization. In the department optimization layer, an expert knowledge based rules reasoning engine is proposed, which efficiently helps the plan-maker to find and correct the mistakes. In the coordination optimization layer, we addressed a mulit-agent coordination optimization algorithm, which helps different departments to collaboratively work together and then reduced the duplicated outage. In the global optimization layer, a mulit-objective complex constraints programming model is built up and the model is solved by CPLEX, which helps the plan management team to handle the overall plans and make a scientific and reasonable plan. The model and algorithms proposed in this paper have been successfully applied in Pinghu Power Supply Company. After two-years' application, the index ASTC-3 (average scheduled interruption times of customer except limited power supply due to generation shortage of system) has been reduced by 40.8% and the index AIHC-S-3 (average interruption hours of customer due to scheduled except limited power supply due to generation shortage of system) has been reduced by 45.2% and the power sales has been increased by 20.5%. The successfully application shows the effectiveness of the proposed model and algorithms. © 2012 IEEE.