Fault Analysis,Diagnosis And Experimental Research Of Subsea Control Module Electro-hydraulic System

Subsea control module(SCM)is the core control equipment of subsea oil and gas production system.It is responsible for the information collection and monitoring of subsea equipment and the execution of control commands,which is very important for the safe production of subsea energy.The failure of the subsea control module will lead to the loss of subsea equipment and serious production accidents,resulting in environmental damage and loss of personnel and property.The subsea control module works in the harsh deep-sea environment for a long time,which is difficult and expensive to maintain.Therefore,the realization of fault diagnosis and state prediction of subsea control module is of great significance to improve the safety of oil and gas production and reduce accident losses.This paper takes the hydraulic and electronic control subsystem of electro-hydraulic compound SCM as the research object,and studies the method of fault diagnosis and state prediction.Aiming at the problem of high SCM fault concealment,the working principle of SCM electro-hydraulic system is analyzed,and the corresponding diagnosis method is proposed.The fault tree models of hydraulic system and electric control system are established respectively.Based on the hydraulic fault tree,the importance of each component fault mode to the system fault is studied by Bayesian network,and the most important component fault mode is obtained.Based on the fault tree of electronic control system,the fault diagnosis process of electronic control system is designed.The SCM sensor monitoring system is established and the correlative parameters corresponding to the fault judgment of each component are obtained.The working characteristics of each component in hydraulic system are quantified and the transfer function of valve control loop is established.By establishing the simulation model of the hydraulic system component fault in the simulation software and analyzing the influence of the fault conditions,including the fault of the hydraulic joint and the fault of the direction control valve on the important monitoring parameters of the hydraulic system,the simulation results are formed into the simulation data set of the SCM hydraulic system.Based on the results of simulation analysis,a dynamic threshold state judgment method of associated valve switch signal is established.Aiming at the time-varying problem of SCM fault,a fault diagnosis method based on Long short-term memory(LSTM)neural network as the core of SCM monitoring data learning is proposed,which includes off-line fault classification diagnosis and on-line state prediction.The single-step processing process of LSTM cell unit was optimized to speed up the calculation and make the model have the ability of online real-time prediction.Taking SCM hydraulic simulation data set as training data,the prediction error and fault classification accuracy of SCM hydraulic simulation data under different time series prediction models,different model over-parameters and different noises are studied,and the optimum model for SCM fault diagnosis and state prediction is obtained.The accuracy of SCM fault simulation and the applicability of the verification state prediction method are verified by experiments.In this paper,an SCM test platform is used to design the on-off valve test and the hydraulic joint fault simulation test,and the accuracy of the fault-free simulation control group and the hydraulic joint fault simulation is verified respectively.In order to verify the applicability of the proposed state prediction method,the actual data of SCM subsea trial operation was selected as the verification set,and the prediction model trained by SCM simulation data set was used for several prediction tests,and the final test results verified the applicability of the state prediction method.