Reconfiguration process for neuronal classification models: Application to a quality monitoring problem

Abstract

HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Multicriteria decision-making method for scheduling problem based on smart batches and their quality prediction capability

“…To meet these new requirements of high quality and flexibility, companies can either invest in state-of-the-art technologies and production tools or adapt their control architectures. Concerning the first option, we can consider the NN-based setup to limit non-quality as proposed by Noyel et al (2016). The model depends on a multitude of both internal and external factors, such as actual workstation setups, production range of the next product to be processed, and environmental factors such as air humidity or temperature.…”

“…Consequently, because it is very likely that the entire batch must be reworked, its process time is doubled and lengthened by another lacquering and drying process. Therefore, in previous studies, a prediction model of non-quality production risk was developed based on the use of a neural network (NN) (Noyel et al, 2016). Starting from this NN and its subsequent improvements (Thomas et al, 2018), we propose to use it as a means of triggering and reordering as well as for criterion selection involved in the analytic hierarchy process (AHP) decision-making process.…”

Relearning procedure to adapt pollutant prediction neural model: Choice of relearning algorithm

“…However, relearning is time consuming and must be performed only if needed. Noyel et al [7] have proposed to use a control chart [28] in order to determine when relearning is needed.…”

“…The backpropagation algorithm is given by (7). It is an iterative algorithm which evaluates the gradient of the criterion to minimize given by (5).…”

“…So, to reduce the computing power need, relearning step must be performed only when it is necessary. To do that, a control chart may be used [7]. Moreover, the relearning itself must use least possible computing time.…”