Machine Learning and DMAIC to Improve the Production Process in the Industrial Sector: Systematic Literature Review (#886)

Authors

Peña Hernández, Marco Antonio

Leiva Piedra, Jorge Luis

Tenorio Ortiz, Yenny Anali

Abstract

The growing need to optimize processes in the industrial sector has driven the adoption of new approaches based on data analysis, systematization, and interpretation. This systematic literature review aims to analyze how machine learning and the DMAIC method have been developed to improve production processes in the industrial sector. To this end, the PICOC method was used to identify search keywords and questions for subsequent analysis. The search was conducted in Scopus and Web of Science, and inclusion and exclusion criteria were applied. The PRISMA method was used to systematize the screening system, and 42 articles were selected. The results show a decrease from 16,066 to 119 DPOM and an accuracy of 98.8% in the coefficient of determination (R²) by the K-Nearest Neighbors (KNN) model. Advantages such as process standardization, fault prediction, and quality improvement were identified. Despite this, organizational, technical, and social challenges were also reported, such as organizational resistance to change, lack of trained personnel, and the need to ensure data security. It was concluded that the integration of Machine Learning with DMAIC is an effective strategy for continuous improvement, provided that it is accompanied by technologies such as IoT and sensors, as well as the proper use of tools such as FMEA, VSM, and control charts.