Method and system for checking abnormal optical network unit in passive optical network

The invention discloses a method and a system for checking abnormal optical network units in a passive optical network, and the method comprises the steps: enabling the network flow data of each terminal device to obtain network flow characteristics through a convolutional neural network according to a network flow data matrix which is formed by arranging the network flow data according to two dimensions of device number and time, and then based on the time information of each terminal device, obtaining a signal model value of the corresponding terminal device in the period of time, performing normalization processing to convert the signal model value into a probability space, weighting the network flow feature map on a sample dimension of the terminal device to obtain a classification feature, and obtaining a classification result through a classifier, wherein the classification result is used for representing whether the optical network unit is abnormal or not. 公开了一种无源光网络中排查异常光网络单元的方法及系统，其通过卷积神经网络将各个终端设备的网络