Wind turbine power generation performance evaluation under faults condition

Since faults in wind turbine transmission chains would reduce the efficiency of generated power, accurate power generation performance (PGP) evaluation is necessary for the economic use of wind energy and reasonable arrangement of maintenance plan. However, the harsh environment and illegal operations lead the PGP difficult to evaluate. In this paper, a robust statistical method which combines Gaussian statistics (GS) method and least trimmed squares (LTS) to evaluate the PGP. Firstly, the data set is preprocessed by the GS method to eliminate the abnormal data caused by illegal operations; taking the mass center (MC) in each small wind scope as a representation of PGP, the LTS method is then used to fit the wind-power curve (WPC); finally, the decline ratio of PGP with faults can be obtained by comparing the change ratio of WPC area between fault and normal WTs. Taking the generator bearing fault as an example, a real data set from a large wind farm is used to evaluate the PGP under generator bearing faults. It is found that the PGP would decline over 3%. The evaluation result can be used to give a guideline on an economic maintenance plan.