Research On Feature Extraction Of Fatigue Crack Of Wind Turbine Blade

Now, wind turbine blade failure has become the accident hidden trouble in wind field.In this paper, we explore the fatigue crack evolution of wind turbine blade under thecombination of random load and alternating load.The method of identification differentinitial cracks has been analyzed. We want to understand dynamic condition and causalrelationship of damage degree, and to identify the damage degree and type of fatigue crack.That is independent intellectual property rights of large-scale wind turbine blade fatiguedamage identification system, so as to solve the problem of real-time monitoring of largewind turbine blade. When fault is minor, the location and extent of damages is identified asearly as possible, and fault warning is demonstrated in advance, to ensure efficient windturbine running safely and reduce maintenance cost greatly.This paper presented the diagnosis model which contains experiment of initial crackand the crack growth. Through the test-bed, different types of crack and fault signal ofcrack phase are collected. The installation location of acoustic emission sensor can beoptimized to effectively monitor state of blade, at the same time the sampling, filtering andfrequency are determined. Crack characteristics of wind turbine blades under cycle loadingis analyzed, The associated mechanism of transient AE guided wave transmission and thefatigue feature of the local stress concentration can be clear, characteristics of the crackdeformation, and the causal relationship between growth rate and fatigue damage havebeen determined to assess of wind turbine blade fatigue state timely.This paper demonstrated the crack initiation and propagation mechanism through theexperimental simulation, and the influence of the dynamic stress. According to the cracktype and status to determine the degree of fatigue, it is the mechanism of adaptive waveletanalysis with AE signal that is built. Firstly, adaptive selection of wavelet basis functioncombined with Shannon entropy method, which eliminates background noise from usefulinformation, and then the wavelet scalogram and reassignment scalogram are applied.Finally the characteristic of different types of cracks is clear by comparing.The fatigue limit of FRP blade is not clear. When the blade cracks and the natural frequency decreases,the influence of the different parts of the crack on natural frequencies is different, and themode will change. Therefore the mechanism of blade fatigue characteristics is verycomplex. To extract crack characteristics with multiresolution is the key to solve thisproblem. This paper applied multi-resolution SVD and reconstruction to get less noisesignal, and multi-resolution reassignment scalogram to get the feature vector, combiningenergy expression method.For the mass behavior from short crack initiation, growth and expansion, long cracks,to breakage, using real-time AE signal will appear a long time span, and control damagestate badly. The fractal theory is adopted to analyze the geometric characteristics of crack,through the fatigue acceleration experiment. That method can eliminate external factors. Itshows that the process of crack initiation to breakage, and the fractal dimension is the newfeature vector of state.