The health status of gearbox directly affects the power generation of wind turbine. In order to achieve early warning of gearbox fault status in engineering practice

a K-means clustering algorithm based on improved lion swarm optimization was proposed. The supervision mechanism and the sine and cosine optimization algorithm considering nonlinear weights are introduced into the lion swarm algorithm

and then the optimized lion swarm algorithm is used to iterate the lion king position. By selecting the optimal solution as the clustering center of the K-means algorithm

the problem of strong dependence of conventional clustering algorithms on the selection of initial clustering centers is solved. The UCI data are selected for comparative verification of the algorithm

and the results show that

the K-means clustering algorithm based on the improved lion swarm optimization has achieved a better improvement in classification accuracy and stability. This algorithm is then applied to comparative test of gearbox vibration acceleration effective value for four wind turbines of the same type in a wind farm. It is found that the distribution of classification centers determined by this algorithm is consistent with the actual operating status of the gearbox

and agrees well with the vibration energy distribution corresponding to different states of the gearbox specified in the standard

indicating that the algorithm can realize early fault warning of wind turbine gearbox.