Wind Turbine Health Monitoring Application
The wind turbine health monitoring application is developed for the Robust Distributed Wind Power Engineering project. The original application is implemented in a subset of the Real-Time Specification for Java (RTSJ). This application implements a crack detection algorithm for turbine blades based on vibro-acoustic modulation. It consists of three major components:
- Prober: Imposing a clean audio tone of several kHz in frequency on a turbine blade via a transducer mounted on the blade.
- Recorder: Recording audio from the blade via a transducer mounted at a different point.
- Crack detector: Analyzing the spectrum of the induced tone.
Real-time Requirement of Major Components
- The prober and recorder run with the same priority (49) and the same periodicity (50 milliseconds);
- The detector is implemented by using periodic real-time thread in RTSJ; In RTdroid, it is implemented as real-time receiver, and triggering by recorder.
The priority of detector is always lower than prober and recorder.
To validate the predictability of RTDroid, we measure the periodic components, prober and recorder. Namely, We collect the time intervals between every release. The plots only include the releases that actually performance application logic.
The experimental results can be found in following links: 1) Scatter Plot for turbine application in RTDroid, 2) Scatter Plot for turbine application in RTcdfSJ. 3) Cumulative Distribution of these two components for RTSJ and RTDroid implementation.
The results of turbine monitor application show similar performance on both of RTDroid and RTSJ implementation. Additional, we also plot the CDF for these prober and recorder of both implementation in one CDF plot.
The implementations of Turbine application under RTSJ and RTDroid are hosting in a shared public repository:
The instruction of compiling and running the application is detailed here.
The implementations of Turbine application in Android can be download here.