Induction motors are the primary driving power in modern manufacturing, and fault diagnosis is critical to product safety, component quality, and maintenance cost management. The fundamental principle behind fault diagnosis is establishing the problem's kind, size, location and time of discovery based on the system's available data. Fault diagnosis has evolved into a critical component of contemporary process automation. It lays the groundwork for fault tolerance, dependability, and security, essential design aspects in complex engineering systems. This study provides solutions to these problems by introducing a Convolutional Neural Network with several tasks (CNN) with data coimbaining for defect identification and regionalization in rolling element bearings. Predictive maintenance of mechanical equipment utilizing multi-source sensing information from the Internet of Things (IoT) with fusion data processing technologies may greatly increase machine service life and save labour costs for identifying mechanical issues, making it a very important topic. IoT-CNN data fusion may balance the pace of convergence of two classification tasks, allowing training procedures to perform fault diagnosis and localization concurrently, identifying a defect utilizing symptoms, knowledge application, and test results in analysis. The findings indicate discovery and exploration developing a better smart integration model that integrates advantages of multiple fused models is challenging and has certain values for accelerating the improvement of mechanical defect detection and forecast.