Conception of a new quality control method based on neural networks

The prediction of failures in a factory is now an important area of industry that helps to reduce time and cost of non-quality from the data generated from the sensors installed on production lines, this data is used to detect anomalies and predict defects before they occur.  The purpose of this article is to model an intelligent production line capable of predicting various types of non-conforming products.  For that, we will utilize the neural network methodology within the specific context of a production line specialized in juice manufacturing.  Firstly, we introduce the production line under study, along with its distinct manufacturing phases.  Secondly, we evaluate the performance indicators of this line, enabling us to gain an overview of its efficiency and overall performance.  Subsequently, we present common industrial solutions that are frequently implemented to address the issues identified during our analysis.  At this stage, we propose a predictive model based on neural network methodology.  This model will possess the capability to detect and identify defective products and potential hazards within a production line before they occur.  Throughout this study, we compare between three models of neural networks: LSTM model using Stochastic gradient descend (SGD), Feed forward model using ADAM Optimization and Feed forward model using Levenberg–Marquardt back propagation (LMBP), in order to determine the most optimal method in terms of achieved results.  Finally, we demonstrate the effectiveness, performance, and accuracy of the results through the testing phase of the neural networks.