Control Chart Patterns (CCPs) Forecasting Using Probabilistic Deep Learning.

Control chart, a powerful tool of statistical process control, monitors efficiently the production quality. In practice, it is realized that each pattern in a control chart usually corresponds to a functional/failed state of the production process. Most existing works in the literature focus on the identification of control chart patterns (CCPs). In this paper, we propose a prognostic model for CCPs forecasting. The model is constructed based on the probabilistic long-short-term memory (LSTM) network to capture the dependencies and uncertainties of the control chart data. The performance of the proposed model is then validated through a numerical study. The prediction results can be used as a basis for the predictive maintenance decision-making and helping to avoid unexpected failures of the production process.