An optimized ensemble framework for machinery fault detection in IoT environments.

Fault detection in IoT-enabled machinery involves identifying defects in the operation of industrial equipment to foil breakdowns that ensure reliability. It typically relies on analysing data from IoT sensors that monitor key parameters such as temperature, vibration, pressure, and speed in machinery. Fault detection systems in industrial settings often face challenges due to the inconsistent distribution of critical sensor data, affecting monitoring systems reliability. Thus, this research aims to design an Optimized Robust PCA-based Ensemble framework for Machine Learning model to find the abnormalities in the IoT-enabled machinery and ensure robust performance by analysing the distribution patterns of critical sensors. This frameworkA is used to extract the essential samples based on voltage, speed, temperature and vibrations. The ensemble learning model includes K-Nearest Neighbour (KNN) and Adaboost, which are optimized by Bayesian optimization. The importance of optimization technique used to fine-tune the parameters of ensemble learning model. The improved ensembled learning model detects the fault in the IoT enabled machinery can be ensured by the performance metrics such as accuracy, precision, and detection rate and false negative rate. The effectiveness of the proposed model is evaluated by varying operational conditions and viability is analysed using an extensive dataset is IoT enabled machinery operations. Thus, the experimentation of the proposed model significantly improves fault detection reliability, ensuring operational efficiency and reducing downtime in industrial infrastructures.