Search

The fluid power accumulator is a high-risk component in the safety shutdown system of most modern wind turbines. Recent studies of failures in wind turbines have shown a high failure rate of such accumulators. This paper proposes an estimation method for detecting one of the most common faults for accumulators, namely gas leakage. The method utilizes an Extended Kalman Filter for joint state and parameter estimation with special attention to limiting the use of sensors to those commonly used in wind turbines. The precision of the method is investigated on an experimental setup which allows for operation of the accumulator similar to the conditions in a turbine. The results show that gas leakage is indeed detectable during start-up of the turbine and robust behavior is achieved in a multi-fault environment where both gas and external fluid leakage occur simultaneously. The estimation precision is shown to be sensitive to initial conditions for the gas temperature and volume.