Abstract
Nowadays, the creation of smart manufacturing systems has high importance. Neural networks have been widely applied to solve complex manufacturing challenges. The paper is devoted to the study of neural networks with reinforcement learning as PPO (Proximal Policy Optimization), DQN (Deep Q-network) for state diagnosis of industrial equipment within the GEMMA (Guide d’Etude des Modes de Marche er d’Arret) model. The GEMMA French approach is established on the SFC (Sequential Function Charts) language and includes standards for controlling technical processes. An application of neural networks in area D of the GEMMA model is introduced. Modelling and experimental results were conducted based on synthetic and experimental datasets. The implementation of the architecture considered allows us to achieve reliable results for industrial data.