Active power distribution network accurate load shedding method based on state identification

The invention discloses a precise load shedding method for an active power distribution network based on state identification. The method comprises the following steps: processing distributed photovoltaic historical data by adopting an isolated forest algorithm; by means of zero correlation between illumination intensity and actual load power, distributed photovoltaic power identification is achieved through a neural network; integrating parameters such as frequency, voltage and demand response quantity to set the power vacancy of the power distribution network; and establishing a precise load shedding model of the active power distribution network by taking the minimum actual load shedding amount as a principle. According to the load shedding method provided by the invention, accurate identification of distributed photovoltaic can be realized, and the distributed photovoltaic nodes are prevented from being cut off, so that the lean level of the load shedding action is effectively improved, and safe and stabl