Distribution unmanned aerial vehicle anomaly detection and processing method based on flight data

The invention discloses a distribution unmanned aerial vehicle anomaly detection and processing method and system based on flight data, which are used for detecting the abnormal condition of parcels delivered by an unmanned aerial vehicle and improving the delivery rate and the accuracy rate. According to the technical scheme, parameter indexes are selected from multiple angles for flight data; performing standardization processing on the obtained parameters of the unmanned aerial vehicle; the method comprises the following steps: inputting unmanned aerial vehicle data into a self-adaptive bidirectional GRU model network, performing forward propagation and self-adaptive back propagation on the input data by a model, respectively extracting characteristic values of forward state information and reverse state information through an attention mechanism, combining the characteristic values of the forward state information and the reverse state information, and outputting final state information; inputting the sta