Research On Virtual Instrument Based Fault Diagnosis System Of Wind Turbine Rotating Machinery

With the situation of energy shortage becoming increasingly serious and pollutionproblem prominent, wind energy, as a kind of renewable green power, has been the focus ofexploitation for countries all over the world, and wind power technology becomes a hot spottopic for research correspondingly. Owing to the fact that wind turbine has a complicatedstructure and works in terrible conditions, and especially because of the increasing installedcapacity and scale, the occurrence of failure and accident are more frequent, which bringsimmense economic losses. Therefore, it has significant value to assure the safe and reliableoperation of wind turbine and reduce the incidence of failure by developing effectivecondition monitoring and fault diagnosis system.The rotating machinery system of wind turbine is a complex nonlinear system and amass of non-stationary vibration signal will emerge in the process of operation. Studying themonitoring and fault diagnosis technology based on vibration can detect fault in time andavoid safe accident; it’s the key of the wind turbine monitoring and diagnosis technology todesign feasible software and hardware system combining the research methods. In this paper,Local Mean Decomposition (LMD) and improved algorithm based on intrinsic time-scaledecomposition (ITD) are applied as time-frequency methods to the fault diagnosis aiming atrotating mechanical components. Respectively, envelope spectrum characteristics based onLMD and support vector machine (SVM) are used for rolling bearing fault diagnosis、ITDimproved algorithm and correlation dimension are combined for gear fault diagnosis、singularvalue decomposition based on LMD and SVM are used for rotor fault diagnosis. Meanwhile,a Labview based system consisting of these methods are designed to achieve the monitoringand fault diagnosis of wind turbine rotating machinery.The main research contents of this paper are as follows:1. Combining with the operating principle of wind turbine and characteristic ofmechanical structure, the failure mechanism and vibration signal feature of rotatingmachinery are studied. Moreover, common monitoring/diagnosis methods and softwaretechnologies are summarized.2. Pre-existing time-frequency methods are introduced and the adaptive decompositionapproach LMD is analyzed compared with Empirical mode decomposition (EMD). Aiming atthe characteristic that the rolling bearing vibration signal is modulated and the filtrationparameters are difficult to determine, LMD is applied to decompose the vibration signals andthe amplitudes ratios in different characteristic frequencies of envelope spectrum for different components are extracted as input parameters of SVM classifier to train the model andclassify working condition and fault patterns of roller bearings. The analysis resultsdemonstrate the effectiveness of the method.3. Improved algorithm based on ITD is applied to the rotor fault diagnosis and comparedwith EMD to show the advantage of less iterations and faster speed. Vibration signals of rotorare decomposed by improved ITD and recombined as de-noised signals, and then correlationdimension after de-noising are calculated to classify different fault patterns of rotor. Theexperimental results between original signals and de-noised signals show the feasibility of themethod.4. The approach LMD are used based on the multi-component modulated characteristicof gear vibration signal to decompose the signal into several simple component signals. Thecomponents which contain the main fault information are assembled into matrixes and thensingular value decomposition technology is applied to abstract fault features. Afterwards,SVM is utilized to distinguish gear fault types.5. Hardware platform is set up with imitation experimental set、acceleration sensor、signal conditioning instrument and data acquisition card. Furthermore, a Labview basedsystem consisting of these methods are designed to achieve the monitoring and fault diagnosisof wind turbine rotating machinery.