Research And Application On Monitoring And Diagnostic Methods Of Diesel Typical Fault Based On Vibration Impact Features

The impact characteristic is the most typical characteristic of diesel vibration signal relative to other rotating equipment signals:diesel engine has many vibration excitation sources,and the vibration response and impact transmission process are complex.The healthy signal characteristics of the diesel engine may be similar to those of other equipment such as bearing with serious faults.The traditional monitoring and diagnosis methods have general effect on the recognition of diesel engines vibration impact.There is insufficient research on the impact physical meaning and transmission law.The physical meaning of the impact feature contains abundant parts and components behavior and engine operating condition information,and plays a key role in monitoring and diagnosing typical diesel engine faults.(1)In this paper,AVL and ADAMS software are used to simulate diesel engine combustion pressure,cylinder head combustion vibration impact,and valve movement impact characteristics.The characteristics of the pressure in the cylinder,the transmission characteristics of the combustion impact,and the impact characteristics of abnormal valve clearance fault are obtained;(2)Study on the identification and extraction algorithm of impact characteristics of diesel engine vibration signal.An algorithm for enveloping the diesel engine vibration signal based on the joint optimization of multiple algorithms is proposed.The impact phase identification method based on the characteristics of envelope statistics is proposed to extract the impact.The impact source identification method based on the energy distribution characteristics of envelope waveform is further proposed to find the source of vibration,in order to recognize the characteristics of vibration impact such as position,energy and source;(3)Study the diagnosis methods of three typical faults through simulation experiments and engineering projects.Firstly,the diesel cylinder head vibration energy spectrum was defined to fuse state information,and a diesel engine condition identification method based on deep learning of RNN is proposed to determine the status of fault diagnosis.Then a misfire fault diagnosis method based on impulse response characteristics,a collision cylinder fault diagnosis method based on impact source identification,and an abnormal valve clearance fault diagnosis method based on impact characteristics are proposed.The methods have been verified by experiments and engineering projects.