Research On Typical Fault Diagnosis And Unstable Condition Monitoring And Evaluation For Piston Engine

As a kind of power machinery,the piston engine is the heart of many mechanical equipments,which has extensive applications in the fields of shipbuilding,power and oilfield.However,poor working conditions,complex mechanical mechanisms and constantly changing operating conditions have made the engine suffer from frequent failures.What's more,the occurrence and development of failures are rapid and disruptive.Malignant failures not only cause serious economic losses,but also casualties.Nowadays,the traditional methods of monitoring and diagnostics which have wide range of applications have been unable to meet increasing practical needs of industrial sites.Therefore,extracting the fault signatures through modem signal processing methods,as well as carrying out research on monitoring and diagnosis approaches of piston engine,then timely diagnosing in the early stages of failures,which is extremely meaningful and valuable to improve the safety,availability,and reliability of the engine.This thesis takes the engine as research object,aiming at improving the typical fault monitoring and diagnosis level and its engineering application effect.Based on the study of fault diagnosis mechanism characteristics,we extract fault-sensitive characteristics through dynamic simulation calculation and signal processing methods,and then,we study the typical fault monitoring and diagnosis methods,meanwhile,the transient condition monitoring and assessment methods are researched as well.Finally,these methods are verified by experiments and engineering cases.The main research contents of the thesis are as follows:Firstly,research on the characteristics of typical failure mechanism based on dynamic modeling simulation and calculation.The 3D solid model and dynamic simulation model are established for abnormal valve gap faults,connecting rod bearing knocking faults and drive shaft faults,respectively.The failure modes are simulated by adjusting the structural parameters and setting simulation parameters.Then,the dynamic responses changing principle of the key components are acquired and the fault mechanism characteristics are captured.Meanwhile,the results are verified by experimental examples and engineering cases.Sencondly,research on the method for signal processing and feature extraction based on the engine signal characteristics.This thesis proposes a vibration signal envelope method that controls the continuous change of the shape of the envelope through a scale parameters,which realizes that the smoothness and compactness of the vibration envelope can be adjusted continuously and provides a new technique for envelope analysis of vibration signals.Meanwhile,the signal envelope method is introduced into the empirical mode decomposition for optimizing envelope during signal selection,which improves its ability of adaptive decomposition.In addition,a precise calculation method for the instantaneous speed is proposed,which lays the foundation for the accurate extraction of instantaneous speed features and accurate re-sampling in angular domain during transient conditions.Thirdly,research on the method for monitoring and diagnosing typical faults of piston engines based on the simulation results of typical failure dynamics simulation and taken advantage of signal processing and feature extraction approaches.For researching it,three fault diagnosis methods are proposed:valve clearance abnormal fault diagnosis method based on precise extraction of initial phase of impact,fault diagnosis method of connecting rod small bearing knocking based on variational mode decomposition and misfire fault diagnosis method based on weighted fusion of multi-parameter evidence.Finally,the system architecture,measuring point arrangement and typical fault diagnosis algorithm for the on-line monitoring and diagnosis system of typical engine faults are designed.Finally,research on the monitoring method for transient conditions of piston engine based on the precise re-sampling method of instantaneous speed angle domain and combined with the characteristics of engine vibration and instantaneous speed under transient conditions.The characteristics of angular vibration and transient speed sensitivity that can characterize the power-on state are extracted,based on which a single cylinder power monitoring method based on the grey evaluation is carried out.In addition,an approach for monitoring the startup performance of the whole machine based on state membership calculation is also proposed.This work provides a new method for monitoring and evaluation of engine transient conditions.