Research On Multi-operating Condition Health State Assessment And Fault Diagnosis Of Oil&Gas Transportation System

The result of state assessment and fault diagnosis for oil & gas transportation system(OGTS)is the important basis for maintenance of OGTS.Generally,the changement of transportation mission will result in the changement of operating condition(OCs)of OGTS,and then running parameters of OGTS will change with OC.Therefore,a full consideration should be taken into OC influence when OGTS is assessed.In this paper,as to the OGTS,state assessment and fault diagnosis of running equipments and noise reduction method for pipeline signals are respectively studied,and the main contents and conclusions are summarized as follows.(1)A noise reduction of signal pre-processing method based on small approximations removal by means(SARM)is proposed for the first time to eliminate the noise which disturbs the signal processing of pipeline acoustic signal and pressure signal.The simulation signals are used to study the characteristics and influencing factors of the noise reduction conducted by the method.Furthermore,according to the simulation signal analysis,superiorities of SARM are verified by comparison studies with wavelet packet transform(WPT)and singular value decomposition(SVD).The denoised signal is smoother and fully conserves the mutations without false mutation or shifts.Finally,pressure and acoustic signal of pipeline leakage is utilized to verify the effectiveness of SARM.(2)The recognition strategy of the output vector in Back Propagation(BP)model is improved to provide a solution for the problem that traditional BP model can't recognize the untrained new OC due to the unknown transportation mission.Then the improved BP model is applied to the OC recognition of the pump in oil transmission station and compressor in gas transmission station.The results validate the improved model's capability of recognizing unknown new OC and its potentiality of application in field.(3)The equipment multiparameter-based OC self-organization health comprehensive evaluation model is put forward to solve the problem that the state assessment is influenced by OC.The model combining fuzzy comprehensive evaluation model and Gaussian mixture model takes the advantages of the two models.Equipment parameters are filtered by FMEA method according to the real-time monitoring condition of the field equipment.What's more,the type of single parameter distribution model(SPDM)is extended and SPDM associated with OC including Gasussian distribution and logistic regression distribution models are established.In addition,the influence of experts' knowledge and experience is greatly reduced due to the expert assessment in fuzzy comprehensive evaluation is replaced by the assessment result of parameter models.Finally,the assessment case of oil pump unit is utilized to verify the effectiveness of the comprehensive evaluation model.(4)In order to reduce the effect that OC acts to fault diagnosis,a diagnosis model based on optimal algorithm by equipment OC self-organization is proposed.The model induces sample set entropy to describe the chaotic degree of sample set and establishes the mapping relationship between OCs and algorithms.Then the optimal algorithm chosen by OC self-organization is used to make the fault diagnosis more accurate.Under experiemental conditions,the pipeline leakage experiment and rotor fault experiment are used to initiatively verify the rationality of definition of sample set entropyand the effectiveness of the proposed model.