海底电缆在电能传输和跨洋通信中扮演着重要的角色，而海缆故障识别技术则是确保电力系统和通信系统稳定运行的关键。传统的海缆故障识别方法主要依赖人工经验，存在识别精度差效、识别率低下的问题。为此，本文提出了一种基于卷积神经网络(Convolutional Neural Networks，CNN)长短期记忆网络(Long Short Term Memory，LSTM)级联网络的故障识别方法，以解决这一问题。该方法将故障的海缆磁场数据作为输入，通过CNN对故障进行浅层特征提取，再利用LSTM对深层故障特征进行提取，从而实现对海缆常见故障类型(内芯部分断路、铠装层脱落和内芯劣化)的精准识别。为了验证该方法的优势，本文还基于CNN-LSTM网络模型与单一CNN网络模型和单一LSTM网络模型进行对比。实验结果表明:三种网络模型均具有良好的收敛特性，但是基于CNN-LSTM网络的识别准确率高达 98。75%，比单一CNN网络准确率高 11。86%，比单一LSTM网络准确率高17。75%，该方法在海缆故障识别方面具有一定的应用前景。
Fault Identification Method for Submarine Cables Based on CNN-LSTM
Submarine cables are essential for transmitting electrical energy and facilitating transoceanic communication.However,identifying cable faults is crucial to ensuring the stable operation of power and communication systems.Traditional methods of identifying cable faults rely on manual experience,which can be inaccurate and time-consuming.To address this issue,this paper proposes a fault identification method based on a Convolutional Neural Networks(CNN)Long Short Term Memory(LSTM)cascade network.This method utilizes the magnetic field data of the faulty submarine cable as input,extracting shallow features of the fault through a CNN,and then employing a LSTM network to extract deeper fault features,thereby achieving precise identification of common types of submarine cable faults such as inner core partial disconnection,armor shedding,and inner core degradation.To verify the advantages of this method,the paper also compares the CNN-LSTM network model with single CNN and single LSTM network models.The experimental results demonstrate that all three network models exhibit good convergence characteristics,but the recognition accuracy of the CNN-LSTM network reaches 98.75%,which is 11.86%higher than that of the single CNN model and 17.75%higher than that of the single LSTM model.This method holds promise for application in the field of submarine cable fault identification.
万方数据
维普
