New Method for Monitoring of Turbine Blade Tip Using Microwave Sensor and k-Nearest Neighbor Classification Algorithm

In this paper, a K band microwave sensor is simulated to monitoring of turbine blade and is optimized in CST software and And is embedded in the turbine shell and if any change in the tip of the blade or displacement at the tip clearance, the scattering parameters of this sensor is changed and the scattering parameter obtained from the sensor mounted on the crust is defined as the turbine blade fingerprint. In this paper, the measurements indices based on scattering parameters of the near field of microwave sensor as a blade tip failure detector system as well as k-NN classification algorithm for interpreting measurable scattering parameters to determine the failure amount as a new method for monitoring of turbine blade is presented. The advantage of this method is online monitoring of turbine blades with fully extracting the measuring indices due to the scattering parameters of a sample blade and It has been shown that the k-NN classification method has an acceptable accuracy in identifying and determining the amount of tip clearance and the deformation of the blade because the error rate can be reached below 1.8% in this way.