Fault Reconstruction using a Takagi-Sugeno Sliding Mode Observer for the Wind Turbine Benchmark

In this paper, a Takagi-Sugeno sliding mode observer is designed to detect, isolate and reconstruct three different sensor faults and an actuator fault in a specified wind turbine benchmark problem. An extension of the discontinuous observer term using a weighted combination of gains responsible for establishing the sliding motion is proposed. The weights are equivalent to the membership functions of the Takagi-Sugeno (TS) model of the wind turbine. This approach enables new design possibilities for sliding mode observer if the matrices of each local model of the TS model differ significantly. The effectiveness of the design approach is demonstrated by simulation results for fault scenarios defined in the wind turbine benchmark.