Abstract
Loose screws and mechanical vibrations are common causes of equipment failure in industrial and mechanical systems. Early detection of such faults can prevent serious damage, reduce maintenance costs, and improve system reliability. This project presents an Intelligent Loose Screw Detection System using an ESP32 microcontroller, two accelerometer sensors, and two vibration sensors, integrated with Machine Learning (ML) techniques for accurate fault detection.
The system continuously monitors vibration patterns and motion variations from different points of a mechanical structure using dual sensors. The accelerometers capture orientation and dynamic movement, while vibration sensors detect abnormal oscillations caused by loose components. The collected sensor data is processed by the ESP32 and transmitted for analysis.
A Machine Learning model (such as Random Forest) is trained using labeled datasets representing normal and faulty conditions. The model learns patterns associated with loose screws and predicts anomalies in real-time. When abnormal vibration patterns are detected, the system generates alerts, enabling early intervention.
This intelligent approach improves detection accuracy compared to traditional threshold-based systems and minimizes false alarms. The system is cost-effective, easy to deploy, and suitable for applications in industrial machinery, automotive systems, structural health monitoring, and robotics.
The proposed solution enhances predictive maintenance by combining IoT and ML, ensuring safer and more reliable operation of mechanical systems.

Introduction
In modern industrial and mechanical systems, maintaining structural integrity is essential for safe and efficient operation. One of the most common yet often overlooked issues is the loosening of screws and fasteners due to continuous vibrations, mechanical stress, and environmental conditions. Loose screws can lead to abnormal vibrations, reduced performance, equipment failure, and even safety hazards if not detected at an early stage.
Traditional maintenance methods rely on periodic manual inspection, which is time-consuming, less efficient, and may fail to identify early-stage faults. To overcome these limitations, there is a growing need for smart and automated fault detection systems that can continuously monitor machine conditions in real time.
This project introduces an Intelligent Loose Screw Detection System using an ESP32 microcontroller, combined with two accelerometer sensors and two vibration sensors. These sensors are strategically placed to monitor different parts of the system, allowing accurate detection of abnormal motion and vibration patterns. The accelerometers measure tilt, acceleration, and orientation changes, while vibration sensors capture oscillations caused by mechanical instability.
To enhance detection accuracy, Machine Learning (ML) techniques are integrated into the system. Instead of relying only on fixed threshold values, the ML model learns from sensor data under normal and faulty conditions. This enables the system to identify subtle changes in vibration behavior and detect loose screws more reliably.
The ESP32 collects sensor data and can transmit it to a cloud platform or local system for analysis. When a potential fault is detected, the system generates alerts, allowing timely maintenance and preventing further damage.
This approach combines IoT and Artificial Intelligence to create a smart, efficient, and cost-effective solution for predictive maintenance. The system can be widely applied in industries such as manufacturing, automotive, robotics, and infrastructure monitoring.
Objectives
• To design a system using ESP32, accelerometers, and vibration sensors to monitor machine vibrations.
• To collect and analyze real-time data from multiple sensors.
• To detect abnormal vibration patterns caused by loose screws.
• To develop a Machine Learning model for accurate fault detection.
• To reduce false alarms compared to traditional threshold-based systems.
• To provide early alerts for preventive maintenance.
• To create a low-cost and efficient smart monitoring system.