Research on Intelligent Fault Diagnosis of Wind Power Generation System Based on Data Fusion

With the consume of traditional petrifaction energy origin such as coal, matelote and physical gas and the increasingly serious question of entire warming, the penetration ratio of wind power in the energy economy continues to enhance. Wind farms are generally built-in areas with strong winds, tough working environments and a high probability of equipment failure. Faults on large grid-connected wind turbines will seriously influence the safety and stability of conventional strength grids. In addition, unplanned maintenance after a breakdown of wind turbines needs a lot of manpower and corporeal resources, which greatly decrease the efficiency of wind strength production and enhance production costs. Therefore, the key to solving the above problems is to quickly and efficiently identify fan faults, which in turn enables accurate troubleshooting. In the article, the malfunction diagnosis of intelligent wind power system based on data fusion is discussed, and it is found that the GBoost algorithm has high accuracy in detecting sensor gain error, sensor offset error and sensor standard error when the Gaussian white-to-noise ratio exceeds 45 dB. In addition, DBN has different diagnostic effects for different faults with different Gaussian noises, at 45 dB and 35 dB, each type of error varies slightly, and the dotted line varies; at 25 dB, each type of error has a large difference. The difference is large, indicating that at 25 dB, this type of error is more sensitive; comparing the state estimation effect makes DLSTM have good adaptability to time series, and also shows that DLSTM considers the system to be reliable enough, and can be obtained by data fusion of the parameters of each system. What is the state of its system, and then take corresponding measures.