The Fault Diagnosis System Of Wind Turbines Based On Wavelet Transform And Data Mining Technology

Wind turbine is regard as key equipments for the wind farms. It will some time lead to enormous economically and financial losses and creating negative effects once the wind turbine generator set out of control. The modern wind farms have become bigger a possibly have had many kinds of wind turbines. Normally these imported wind turbines from different companies have their own remote monitoring and fault diagnosis system, but these remote monitoring systems cannot be compatible which will give rise to big Problems to the management of wind farm.The structure of a large wind turbine generator tends to be complicated with the combination of electro-mechanical and electro-hydraulic. The failure signal with background noise interfaces, nonlinear and non-stationary character. The rote of its transformation process is in combination with its natured decreasing sophistication often mixed with failure signal source together. This has created great difficulties for the correct diagnosis failure signal, feature study and capturing its technology. Obtaining information from correct failure signal is the critical technology for the failure signal diagnosis. This paper introduces the status of wind power, and the necessity of fault diagnosis of the wind turbines. as well introduced the wind turbine composition, working principle, and the typical failure. According to the mechanism of wind turbine vibration signals and vibration characteristics. In this paper,modern signal analysis and processing technology is applied for the wind turbine vibration signal analysis, including signal de-noising and singularity monitoring, etc, and extracting features of wavelet de-noising analysis of several de-noising methods were compared.In order to improve the accuracy of the fault diagnosis and predict the fault trend of wind turbines. In this paper, architecture design of wind turbines fault diagnosis and condition monitoring System which based on data-mining technology and B/S model were proposed, as also including the software architecture.