Predictive Maintenance in Electronic Assembly Lines Using AI and Edge Analytics

Abstract

The modern factory setting that consists of electronic assembly lines necessitates advanced maintenance approaches that will guarantee the highest level of productivity and the reduction of any unwanted unplanned downtimes. The focus of the present paper is to introduce a broad framework of predictive maintenance within electronic assembly lines with the use of artificial intelligence (AI) algorithms and edge analytics. The proposed system incorporates real-time data collection with sensors, machine learning models to predict failures, and edge computing infrastructure to allow making an immediate decision. The framework makes use of several AI models such as Random Forest, Support Vector Machines, and Convolutional Neural Networks to process equipment health indicators as well as anticipate the occurrence of failures before they happen. Edge analytics facilitates local processing of data and therefore minimizes the bandwidth and latency requirements through data privacy and security. Testing on a representative electronic assembly line shows that the proposed system has 94.3% success in predicting failure and predicting one on average of a 72-hour horizon. The implementation has led to a reduction in the number of unplanned downtime by 37, 37 percent lessening in the cost of maintenance, and a 15 percent enhancement in the efficiency of overall equipment (OEE). The computed edge architecture can handle sensor data with an average latency of 12 milliseconds, which allows responding to important equipment conditions and in real-time. The system is effective to detect various failure modes such as bearing wear, overheating of motors, wear on conveyor belt and misalignment of the components. Integration with current manufacturing execution systems offers easy integration of workflow and automatic scheduling of maintenance. The modular design guarantees scalability of various electronic assembly designs but at the same time makes it economical when it comes to small to medium scale operations.