Research On Intermittent Fault Growth State Detection Technology Of The Wind Turbine Key Components Under Variable Working Conditions

Wind energy as a renewable green energy has been generally received high attention from all over the world,especially China's installed capacity has become the world's largest.With the development of wind power technology and equipment for many years,the market competition is becoming more and more fierce.Reducing operating costs and maintaining high equipment health working hours are the goals of enterprises.Therefore,the fault diagnosis technology of wind turbine key components has also become one of the hot topics in the research of wind power system,and it is also an important means of healthy operation of wind turbine.However,the accurate detection of weak intermittent fault is not only the premise of wind turbine fault predictive maintenance,but also a difficult problem in the field of wind turbine fault diagnosis.Taking full advantage of the advantages of fractional Duffing oscillator in periodic signal detection,this paper introduces it into the detection of intermittent fault growth state of key components of wind turbine under variable working conditions.Combined with the theory and method of modern signal processing,a new fractional order Duffing oscillator detection method for weak intermittent fault growth state is proposed.First of all,through the research on the failure mechanism of wind turbine key components under different working conditions,and theoretical reasoning and experimental verification,the analysis of the weak and intermittent fault signals of key components under different working conditions is completed.Combined with chaos system and modern information processing technology,the method of fault modulation mechanism analysis and improved Morlet wavelet transform is proposed to detect the intermittent fault frequency of periodic and aperiodic vibration respectively.On the basis of fully studying the traditional method of distinguishing the phase state of chaotic system,combined with the characteristics of phase diagram,a phase state discrimination method(Poincare section density peak algorithm,PSDPA),that based on the mapping of steady-state phase trajectory,is proposed.The method uses the outliers of the mapping data of Poincare cross section to judge,which is more accurate,can reduce the artificial misjudgment,and improve the reliability of the system phase recognition.It provides theoretical support for weak intermittent faultamplitude detection.Then,the characteristics of weak intermittent fault of Duffing oscillator and key components of wind turbine are discussed and analyzed.A method for identifying the existence of periodic faults based on the output phase state discrimination of 0.95 order Duffing oscillator and the method of detecting weak intermittent fault amplitude with 0.5 order Duffing oscillator are proposed.Through comparison,it is found that PSDPA algorithm is used for detection.The accuracy of the amplitude of the built-in motive power signal amplitude in the critical large-scale periodic state obtained by the phase diagram is much higher than that obtained by the traditional artificial discrimination method,which verifies the accuracy and operability of the algorithm again.Meanwhile,in view of the unclear relationship between the development of intermittent fault and permanent fault,a new quantitative expression of fault growth is employed,so,through the intermittent fault development function(Intermittent Fault Development Function,IFDF)organically links the weak intermittent fault with the permanent fault,reveals the internal relationship between them,solves the traceability problem of the permanent fault,and also characterizes the growth state monitoring of the weak intermittent fault,greatly improves the accuracy of fault early warning,and provides support for the realization of accurate predictive maintenance.Finally,based on the case analysis of the weak intermittent fault growth state of the key components of the variable condition wind turbine detected by the fractional Duffing oscillator,the results are correct and the detection effect is good.The method has strong operability and high accuracy,and provides strong support for the predictive maintenance of the field wind power plant equipment.