Identification method of fishing vessel operation type based on LightGBM and BiLSTM

In order to overcome the problem of fishery resources damage caused by fishermen’s illegal operations during the fishing process, the paper proposes a method for identifying the type of fishing vessel operation based on LightGBM-BiLSTM This method studies the Beidou data of different fishing vessel, extracts the 42-dimensional effective spatiotemporal feature vectors of the fishing vessel operation mode, stitches the pre-recognition results and the original data to complete the time series slice and 3D reconstruction, and enters the LightGBM-BiLSTM model to carry out the operation type of the fishing vessel. Recognition. Experiments were conducted with Beidou data from 2.09 million fishing vessel, and the accuracy, precision, and recall rates of identification were 98.42%, 99.04%, and 97.56%, respectively. The experimental results show that this method can accurately identify the type of fishing vessel operation, and can be used for remote fishing vessel operation type identification and supervision, which proves the feasibility of this method.