Fault Identification in Honda Scoopy 110CC Continuous Variable Transmission Using Backpropagation Neural Networks

Intensive research in the field of signal processing has driven remarkable advancements in communication technology, particularly in the realm of voice recognition. Voice recognition concepts find application across various domains, with one such application being sound recognition within the context of Continuous Variable Transmission (CVT) for 110cc motor scooters.This study aims to identify potential issues in CVT systems by employing artificial neural networks using Learning Predictive Coding (LPC), Mel Frequency Cepstral Coefficient (MFCC), the Artificial Neural Network (ANN) Backpropagation method to classify distinct sounds emanating from Honda Scoopy 110cc motor scooters. The dataset used comprises 100 CVT engine sound recordings, equally distributed between 50 samples of normal engine sounds and 50 samples of damaged engine sounds.The research findings reveal the highest level of accuracy achieved with order 16 and 16 hidden neurons, resulting in a testing accuracy of 81.3%, a validation accuracy of 100.00%, and a testing accuracy of 90%. This data strongly supports the effectiveness of the backpropagation artificial neural network method for precise CVT issue identification.