Fault Prognosis System on Face-Mask Body Machine with Adabelief-Backpropagation Neural Network

Ultrasonic welding workload on vertical roller components face-mask body machine in making masks has a highvibration of up to 20kHz. This high vibration causes the locking bolt loosen serrations, thus causing greater failure function, such as wear and tear teeth. If this function fails, it will cause downtime costs waiting process for cleat component be remanufactured. The damage prognosis system based condition implements Classification types along with recommendations maintenance activities that need carried out Face-Mask Body Machine. type using There is Belief-Backpropagation Neural Network (BPNN), method looking weight settings neural network error rate obtained previous iteration. optimized AdaBelief, which can adapt step size "confidence" gradient get Convergence rates generalization abilities better so these problems known from signal previously was parsed wavelet packet decomposition into frequency bands obtain data low or frequency. From results performance testing, modeling accuracy 98.4%, declared good feasible use slack detection fixing bolts.