一种基于大数据挖掘模型的故障诊断方法

A Fault Diagnosis Method Based on Big Data Mining Model

针对现有的故障诊断模型存在安全态势感知能力差、实时预警能力差与实时运维保障能力差等问题,提出一种基于大数据挖掘模型的故障诊断方法。该方法将深度学习技术与故障诊断方法相结合,利用卷积神经网络、决策树与长短期记忆网络等模型对故障进行分析,并与传统故障诊断方法相比。结果表明,利用卷积神经网络和长短期记忆网络方法的故障诊断的准确率、回归率、F1值较高,均超过90%,显著提高了故障诊断的准确率与可靠性。

To address the problems of poor security situation awareness,poor real-time early warning capability and poor real-time operation and maintenance guarantee capability of the existing fault diagnosis model,a fault diagnosis method based on big data mining model is proposed.The method combines deep learning technologies with fault diagnosis methods,and uses convolutional neural networks,decision trees and LSTM models to analyze faults,which improves the accuracy and reliability of fault diagnosis compared with conventional fault diagnosis methods.Experimental results indicate that the accuracy,regression rate and F1 value of fault diagnosis using CNN and LSTM methods are fairly high,all exceeding 90%.