The invention discloses a power generation monitoring and load distribution method of a large photovoltaic and wind power generation grid-connected system. The state of the grid-connected points is monitored, and the power generation device of which the online data are abnormal is found if the abnormity is detected; the environmental parameters of the abnormal device are acquired, and an abnormalenvironmental parameter reference database is established; the environmental parameters of each device are acquired in real time and inquiring in the reference database is performed to judge whether the current environmental parameter is the abnormal environmental parameter, and the device is automatically turned off if the current environmental parameter is the abnormal environmental parameter; each power generation device of each sub-net is monitored; and the real-time load is predicted, and the switching state of each grid-connected point is obtained through two steps of neural network training and each power generation device is adjusted in real time so that the sub-nets are enabled to meet the requirements of the grid-connected points. The influence of the environmental factors can beeliminated so that the power supply quality of the power grid can be enhanced; and the power supply utilization rate can be enhanced, and the objectives of energy conservation and meeting the power supply demand can be realized.