基于MLP的船闸控制系统诊断与故障识别研究  

Research on Diagnosis and Fault Identification of Ship Lock Control System Based on MLP

在线阅读下载全文

作  者:蒲浩清 唐念 柏雪 PU Haoqing;TANG Nian;BAI Xue(Yangtze River Three Gorges Navigation Authority,Yichang 443002,China;不详)

机构地区:[1]长江三峡通航管理局,湖北宜昌443002

出  处:《武汉理工大学学报(信息与管理工程版)》2024年第6期958-962,共5页Journal of Wuhan University of Technology:Information & Management Engineering

摘  要:实时监控船闸控制系统(SLCS)各传感器状态数据、准确识别故障情况并采取相应措施,能够提高SLCS的抗风险能力、降低经常性维护导致的经济损失。针对SLCS相关传感器数据特性,采用SHAP模型和文本分析下的正则匹配实现对数据的清洗和处理,基于多层感知机(MLP)构建机器学习模型,实现SLCS的故障识别和风险评估。面向某大型单级船闸系统历史故障数据进行模型应用,结果显示该模型对SLCS故障应用措施识别的准确率达到了98.3%,验证了模型的有效性和机器学习算法在SLCS故障识别中的可行性。研究表明:通过对各传感器状态数据和采取的应对措施进行编码而构建的特征工程可行,通过SHAP模型实现的关键因素遴选设置合理;多层感知机模型可以进一步提高预测的准确率。Monitoring the status data of various sensors related to Ship Lock Control System(SLCS)in real-time,accurately identifying fault conditions,and taking corresponding measures can enhance the risk-resistant capability of the SLCS and reduce economic losses caused by regular maintenance.Based on the data characteristics of SLCS-related sensors,it employed SHAP model and regular matching under text analysis to achieve data cleaning and processing.It further developed a machine learning model based on Multi-Layer Perceptron(MLP)to achieve fault identification and risk assessment of the SLCS.By applying the model to the historical fault data of a large-scale single-stage ship lock,the results showed that the accuracy of the model in identifying SLCS fault application measures reaches 98.3%,which verified the effectiveness of the model and the feasibility of machine learning algorithms in SLCS fault identification.The research results showed that:(1)the feature engineering constructed by encoding the status data of various sensors and the corresponding countermeasures is feasible,and the key factors implemented by SHAP model are selected and set reasonably;(2)the multi-layer perceptron model can further improve the accuracy of prediction.

关 键 词:机器学习 故障识别 多层感知机 船闸控制系统 系统诊断 

分 类 号:TP391.3[自动化与计算机技术—计算机应用技术]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象