Research on Fault Location and Detection Technology of Optical Network Based on Long Short-Term Memory Neural Network

As more and more information is carried by the optical network, the fault location of the optical network has become an important issue in current research. Traditional fault diagnosis methods will bring huge operating costs to the fault diagnosis of existing networks. Therefore, this paper studies the fault location technology of optical networks from the perspective of deep learning. First, an optical network fault detection model based on Back Propagation (BP) neural network is created to analyze common alarm data, analyze fault monitoring points and common fault points, and judge whether possible fault points exist. Secondly, in view of the advantages of Long Short-Term Memory (LSTM) neural network that can operate on sequences, we choose it to predict the probability of network nodes and links failure in a certain period of time in the future, and realize fault identification through network parameter analysis, parameter extraction, and time-series feature transformation. The output indicates possible failure points with a probability of 0 to 1, and the failure points with high probability are preferentially processed. The concept of F-measure is also introduced, and the positioning effect is measured by positioning time, MSE and F-measure. The experimental results show that both LSTM and BP neural network can locate the fault of optical transmission network well, but the network model BP + LSTM combined with LSTM has better overall effect, and its positioning time is shorter than that of BP neural network, which meets the requirements of accuracy and real-time fault location.