Fault Forecast Model For Wind Turbine Based On BP Neural Network

In recent years, due to worsening environmental problems and fossil resource reserves gradually reduced, the countries all over the world have begun to pay great attention to the development and adJustment direction of consumption of energy, renewable energy source, no pollution is gradually incorporated into national energy consumption system.With the popularization and application of aerodynamic theory more perfect and new materials technology, utilization efficiency of wind energy resources increase rapidly, which makes the wind power generation equipment yields rapid promotion, the use of wind resources has become one of the important means for countries to adJust energy consumption structure.China is a wind resource reserves huge country, in inland and offshore, in theory the development and utilization of wind energy reserves reached more than 10 kW, the development and utilization potential. In recent years, efforts to support the policy of wind resources in China is more and more big, the previous economic development planning of wind power installed capacity are in constant increase, in addition to Tibet, the national various provinces, city, autonomous region have established a large wind farms, small wind generator distributed more widely in the market.According to the calculation of industry statistics, China’s wind power industry unit kilowatt maintenance cost in 30-50 yuan, the average annual rate of return on the proJect is about 10%. With the stand-alone capacity improvement, wind turbine parts price rises quickly also, the equipment life cycle is as long as twenty years, equipment daily maintenance cost has become the main factor of affecting the proJect income. How to improve the quality of products, reduce maintenance costs, has become a maJor research topic of equipment manufacturers and wind farm investorsIn this paper, through the operation characteristics of the wind turbine, combined with BP neural network technology, analysis of the typical fault data of wind turbines, a fault early warning technology of wind turbine based on BP neural network are given. In this paper, by analyzing the structure of wind power generator and the pitch system fault principle, typical fault data of typical device selection of wind driven generator with variable propeller system, through the training for the fault data in BP neural network environment, and the number of faults according to the results of analysis and comparison of different algorithms using BP network, to obtain a suitable for fault early warning function of the wind power generating units to achieve the algorithm of BP neural network.