A robust processing technique for trend prognosis On Mechanical Coupling for Common Remotely Operated Weapon system (CROWS)

A robust processing technique for diagnosing and predicting a recipient fault of static components of mechanical systems is presented. In particular, we demonstrate the technique (methodology) to the loosened bolts on the elevation control motor (ECM) of the Common Remotely Operated Weapon Station (CROWS) in order to prevent a disastrous fault or failure in the gear mechanism that loose bolts could cause. Our method, called inverse normalized energy, uses the "symlet" wavelet to de-noise non-stationary vibration signals from a tri-axial accelerometer mounted on the ECM. We calculated the inverse of the ratio of the normalized energy of the "baseline," in which all the bolts were tightened, to cases in which the bolts were loosened to different levels of torque. The normalized energy signals were calculated from the output of Fast Fourier Transform (FFT) spectral components at the frequency band of the residual signals - the difference between the "raw" experimental data and the de-noised data. We conducted a series of controlled experiments where we deliberately loosened the ECM bolts to demonstrate the system's diagnostic and prognostic capability. Based on the experimental data and results from our method, we showed that we can detect a fault and the trend of the loosening bolts on the weapon station. If faults are detected early enough, appropriate measures can be to taken to enhance the reliability of the weapon station. This robust technique may also be well suited for detecting the fault of components of mechanical systems in manufacturing and heavy fixed machinery.