Power system consists of numerous equipment components. We have to draw up appropriate maintenance policy to ensure the safety and reliability of power system. However, even the smallest change of structure of the gird caused by the maintenance of equipment can cause a significant reduction in power system reliability. The maintenance of electric power equipment will reduce the redundancy of the power system. The capability to resist disturbance of our gird can be strongly weakened in this case. Therefore, the probability of huge disaster and the loss of failure will increase as well. In conclusion, the regular pattern of power system reliability under the state of equipment maintenance is worth for us to do some in-depth study. Different from conventional planning reliability, for the work of equipment maintenance usually cost a few days, the evaluation of power system reliability under the equipment maintenance state has its time-varying characteristic. We have to use online realtime data and some prediction algorithm to analyze maintenance state. This paper presents a set of scheme to evaluate the reliability of power system based on maintenance state. the scheme include following aspects: using load forecast algorithm to obtain load and its change during maintenance. Using weather forecast to obtain several factors which has the greatest effect on devices such as temperature, wind velocity and humidity. Amend equipment failure rate that based on Statistics by using weather forecast data. Using Monte-carol simulation method to get possible state of power grid based on maintenance state. Calculate the grid's power flow of every state simulated by last step. Adjust relative load of bus when the result of power flow trend to non-convergence or has any other unsatisfactory situation. Considering all these results and give the reliability index for the maintenance state of power system at last.