Machine learning applied to forecasting the manufacturing time of new products prototypes and ETO products: An exploratory study

Abstract

Industry 4.0 is a transformative initiative that integrates various technologies and reshapes industrial processes, production methods, and business models. However, forecasting future events within this paradigm shift presents significant challenges. Predicting the cycle time for new product development (NPD) in a dynamic and competitive environment, especially in a highly globalized market driven by innovation, is crucial. Previous research has shown that prototype manufacturing lead times are key parameters for predicting NPD times-to-market. This study investigates the predictive capabilities of artificial intelligence algorithms in estimating manufacturing lead times under varying part geometries and materials at an aerospace and medical equipment company. By leveraging predictive analysis and machine learning techniques, specifically Artificial Neural Networks (ANN), Support Vector Machine (SVM), and Random Forest (RF) algorithms, the proposed methodology demonstrates its efficacy and variations. The results show that top-performing models achieve an accuracy rate exceeding 87 % and an average absolute error of less than one day, which have significant practical benefits for project production planners. They can utilize the most popular AI frameworks on easier-to-use programming platforms to estimate the time required to manufacture their prototypes, predict their new product development (NPD) cycle times, and negotiate lead times for in-house and third-party manufacturing more effectively, thereby improving project planning and delivery.