Research On Wind Turbine Gearbox Vibration Signal Feature Extraction And Fault Diagnosis Method

Wind energy source as a clean and renewable energy, in the context of world energy increasingly tense, Wind turbines became one of the important means of resolving the power and energy shortages nowadays. In recent years, wind power industry has developed rapidly in China, because of the work environment of wind turbines is the field of long-term, gearbox as its core components prone to damage, so do fault diagnosis for it has important practical significance. The key of gearbox fault diagnosis is vibration signal processing. This paper studies wind turbine gearbox vibration signal extraction and fault diagnosis method. The main work is arranged as follows:(1) Build experimental platform of wind turbines, on the basis of the analysis of the operating characteristics of wind turbine gearboxes, failure mechanism and the feature of vibration fault, in complex situations, aim the non-linear, non-stationary, feature is difficult to extract and quantify characteristics of gearbox vibration signals, study the methods of extract signal features based on wavelet packet, it can effectively extract and accurate description of the gearbox vibration signal characteristics.(2) Aim at the characteristics of vibration signals when the wind turbine gearbox work, extract the vibration signal feature based on wavelet packet transform method, the extracted feature value would be putted into BP neural network to train and classification, it can achieve intelligent diagnosis to gearbox fault. Experiments show that the method has a good result of feature extraction and fault diagnosis.(3) On the basis of wavelet packet valid extract vibration signal characteristics of wind turbine gearboxes, use FCM fuzzy clustering recognize eigenvectors of gearbox vibration signals. It can be effective identification and judgment the gearbox condition and fault. Experiments show that the method has good result of feature extraction and fault diagnosis.