Bearing Fault Detection and Separation of Wind Turbine Based on Artificial Immune System

In this paper, a novel bearing fault diagnosis algorithm based on artificial immune system is proposed for wind turbine. Historical data of different cases is firstly analyzed and features including mean and variance are got, which are then transformed into binary strings. These binary strings are regarded as self-strings, based on which detectors are generated using negative selection algorithm for fault detection and positive selection algorithm for fault separation. For test data, features and their corresponding binary strings are calculated, which are then compared with proposed detectors according to the improved Hamming matching rule. From the comparison results, it can be got whether the bearing fails or not (namely fault detection) and if any, which fault belongs to (namely fault separation). To verify the effectiveness, simulations are conducted finally.