DQ-DeepLearn: Data Quality Driven Deep Learning Approach for Enhanced Predictive Maintenance in Smart Manufacturing

Abstract

In the realm of smart manufacturing, predictive maintenance plays a pivotal role in ensuring equipment reliability, minimizing downtime, optimizing costs, and reducing product failure rate by detecting faulty products in the early stage. However, the efficacy of predictive maintenance hinges on the quality of training data employed for predictive modeling. Inadequate data quality can lead to erroneous predictions and, consequently, ineffective maintenance strategies. This research paper introduces a solution that leverages deep learning and data quality management to enhance predictive maintenance in smart manufacturing. Our proposed methodology encompasses two major components such as data quality management and faulty product detection. Data quality management includes data preprocessing, feature reduction, and data balancing, whereas faulty product detection is done using deep learning techniques. By combining these elements, a predictive model capable of accurately forecasting faulty products in early stages to reduce economic losses is developed. The proposed approach effectively addresses data quality issues and is tested on the SECOM dataset which indicates that it surpasses traditional models with an accuracy of 96.4% and perfect recall. Ultimately, this research contributes to the advancement of predictive maintenance and deep learning in the context of smart manufacturing, benefiting both industrial practitioners and researchers alike.