No abstract

No abstract

No abstract

This study introduce the application of machine learning algorithms for supporting the manufacturing quality control of a complex process as an alternative for the destructive testing methodologies. The choice of this application eld was motivated by the lack of a robust engineering technique to assess the production quality in real time, this arise the need of using advanced smart manufacturing solution as AI in order to save the extremely high cost of destructive tests. In concrete, this paper investigates the performance of machine learning techniques including Ridge regression, Linear Regression, Light Gradient Boosting Machine, Lasso Regression and more, for predicting the at glass tempering quality within the building glass industry. In the rst part, we applied the selected machine learning models to a dataset collected manually and made up by the more relevant process parameters of the heating and the quenching process. Evaluating the results of the applied models, based on several performance indicators such as Mean Absolute Error, Mean Squared Error, R Squared, declared that Ridge Regression was the most accurate model. The second part consist of developing a digitalized device connected with the manufacturing process in order to provide predictions in real time. This device operates as an errorproo ng system that send a reverse signal to the machine in case the prediction shows a non-compliant quality of the current processed product. This study can be expanded to predict the optimal process parameters to use when the predicted values does not meet the desired quality, and can advantageously replace the trial and error approach that is generally adopted for de ning those parameters. The contribution of our work relies on the introduction of a clear methodology (from idea to industrialization) for the design and deployment of an industrial-grad predictive solution within a new eld which is the glass manufacturing.

This research investigates the applicability of Deep Reinforcement Learning (DRL) to control the heating process parameters of tempered glass in industrial electric furnace. In most cases, these heating process parameters, also called recipe, are given by a trial and error procedure according to the expert process experience. In order to optimize the time and the cost associated to this recipe choice, we developed an offline decision system which consists of a deep reinforcement learning framework, using Deep Q-Network (DQN) algorithm, and a self-prediction artificial neural network model. This decision system is used to define the main heating parameters (the glass transfer speed and the zone temperature) based on the desired outlet temperature of the glass, and it has the capacity to improve its performance without further human assistance. The results show that our DQN algorithm converges to the optimal policy, and our decision system provides good recipe for the heating process with deviation not exceeding process limits. To our knowledge, it is the first demonstrated usage of deep reinforcement learning for heating process of tempered glass specifically and tempering process in general. This work also provides the basis for dealing with the problem of energy consumption during the tempering process in electric furnace.