Research On Fault Diagnosis Of Diesel Shaft System Under Unsteady Condition

Diesel is an important power equipment,which is widely used in the fields of diesel locomotives,ships,automobiles and power generation equipment.Shaft system of diesel unit contains many components.As being the main components of shaft system,the performance of the coupling and the absorber will directly affect the stability and reliability of mechanical equipment.Once the shaft system broken down,the mechanical performance of the diesel unit will deteriorate rapidly,which will affect the normal operation of the equipment.If the fault is serious,it will cause significant economic losses,and even endanger personal safety.Therefore,the monitoring of the working condition of diesel shaft system and faults prediction are of great significance.The symptoms of various faults of diesel shaft system are not obvious under the steady condition,and the fault's characteristic signal extraction and diagnosis cannot be carried out timely and accurately.In the unsteady condition,the instantaneous speed and transmission torque of the shaft system change frequently.Compared with the torsional vibration signal under steady condition,the signal under unsteady condition contains more abundant fault information.Therefore,the study of fault diagnosis under unsteady condition makes great sense to improve the accuracy and timeliness of diesel unit fault diagnosis.Based on the torsional vibration calculation of unsteady condition,the order components of torsional vibration and modal parameters are used as characteristic parameters.The fault diagnosis of coupling and absorber are studied by self organizing neural network in this paper.In this paper,through programming,the impulse-response-method and order analysis method are applied to the calculation of torsional vibration under unsteady condition,namely through the convolution integral of unit impulse response function and excitation torque to calculate the torsional vibration response.The calculated results are compared with the measured torsional response and the simulation results of TVCA.The feasibility and accuracy of the method in the calculation of torsional vibration under unsteady conditions are verified.In order to solve the problem of frequency ambiguity when analysing unsteady signal by spectrum analysis,the order analysis method based on equal angle resampling is used in the program to analyze the torsional vibration response of diesel shaft system under unsteady condition.Results show that order analysis method can distinguish order components of torsional vibration signal under the unsteady condition correctly.Compared with other analytical methods,it has significant advantage in the study of unsteady signal of rotating machinery.The equivalent model of shaft system of locomotive diesel generator is established in this paper.The influences of stiffness and damping on the modal parameters and torsional vibration responses are studied by simulation calculation method.The order components of torsional vibration and modal parameters which are sensitive to different faults are confirmed.Eventually,combining order analysis method,modal parameter identification method and self-organizing feature neural network,the faults of coupling and absorber are diagnosed.The result shows that the self-organizing feature neural network can diagnose the faults of coupling and absorber quickly and accurately.In summary,the study on fault diagnosis of diesel shaft system under unsteady condition has certain application value in engineering,because it enriches the analysis method of torsional vibration of shaft system and proposes the fault diagnosis method which based on modal parameter and order analysis of unsteady signal.