Gear fault trend prediction based on FGM(1, 1) model

Abstract

Gears have the highest rate of occurrence of failures in the wind turbine components. Therefore, strengthening the judgment of the gear running state in the future and forecasting its developing trend is of great importance. The required data of grey forecasting model is less and the prediction precision is high. Besides the method is simple, we can more accurately describe the inherent law of practical problems. However, it has been found that the actual phenomena tend to be irregular in practice. People usually use the fractional order to replace the integer order. This paper adopts the accumulated generating operation to weaken the original sequence randomness, which makes the disturbances of solutions of grey forecasting model smaller, and fractional order FGM(1,1) model is established to predict the trend of gear fault. The results of GM(1,1) model, Grey Neural Network model and fractional order FGM(1,1) model are compared respectively, and we analyze the differences of fitting precision between the three kinds of models mentioned above. The results illustrate that fractional order FGM(1,1) model has a high prediction capability compared with GM(1,1) model and Grey Neural Network model.