基于LSTM-RF的电动钻机绞车齿轮箱故障诊断  

Fault diagnosis of electric drill winch gearbox based on LSTM-RF

作　　者：刘光星[1,2] 马一豪 LIU Guangxing;MA Yihao(School of Electronic Engineering,Xi’an Shiyou University,Xi’an 710065,China;Key Laboratory of Oil and Gas Well Measurement and Control Technology,Xi’an 710065,China)

机构地区：[1]西安石油大学电子工程学院,西安710065 [2]陕西省油气井测控技术重点实验室,西安710065

出　　处：《振动与冲击》2024年第21期156-162,230,共8页Journal of Vibration and Shock

基　　金：陕西省教育厅重点实验室项目(17JS107);西安石油大学研究生创新与实践能力培养计划资助项目(YCS23214251)。

摘　　要：针对提高石油电动钻机绞车齿轮箱故障诊断的准确性和效率,提出了一种基于长短期记忆网络(long short-term memory,LSTM)和随机森林(random forest,RF)融合模型。首先,运用LSTM能够从大规模数据中学习复杂特征,将这些特征作为随机森林的输入。然后,通过随机森林处理非线性和高维数据以及对特征的分类,以实现对齿轮不同故障状态的识别。最后,利用电动钻机绞车齿轮箱运行过程中的实时数据,建立了一个包含多种齿轮故障类型的综合数据集。试验结果表明,LSTM齿轮故障诊断准确率为94.67%,RF齿轮故障诊断准确率为94.34%,支持向量机齿轮故障诊断准确率为82.00%,K近邻齿轮故障诊断准确率88.33%,而融合模型LSTM-RF在齿轮故障诊断准确率方面达到了98.33%,克服了单一模型的局限性,提高了诊断准确性。研究表明了融合模型具有更优的电动钻机绞车齿轮箱故障诊断能力。Here,to improve the correctness and efficiency of fault diagnosis for winch gearbox of petroleum electric drill winch,a fusion model based on long short-term memory(LSTM)and random forest(RF)was proposed.Firstly,LSTM could be used to learn complex features from large-scale data,these features were taken as inputs to RF.Then,non-linear and high-dimensional data were processed with RF,and features were classified to realize recognition of different fault states of gears.Finally,a comprehensive dataset containing multiple types of gear faults was established using real-time data in operation process of electric drill winch gearbox.The experimental results showed that the correct rate of LSTM gear fault diagnosis is 94.67%,RF gear fault diagnosis correct rate is 94.34%,support vector machine(SVM)gear fault diagnosis correct rate is 82.00%,K-nearest neighbor(KNN)gear fault diagnosis correct rate is 88.33%,while the correct rate of the proposed LSTM-RF fusion model gear fault diagnosis reaches 98.33%to overcome the limitation of a single model and improve diagnosis accuracy.The study showed that LSTM-RF fusion model has a better fault diagnosis capability for winch gearbox of electric drill winch.

关 键 词：电动钻机 齿轮箱 故障诊断 长短期记忆网络(LSTM) 随机森林(RF)算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]