Efficient stochastic framework for availability improvement of stone door frame manufacturing plants using artificial neural networks and regression analysis

Abstract

The main objective of this study is to introduce an efficient stochastic framework to improve the availability of the stone door frame manufacturing plants along with the reliability, maintainability, and dependability (RAMD) investigation and prediction of steady state availability of the plant using regression analysis (RA) and artificial neural networks (ANNs). The plant has five subsystems connected in series configuration. The RAMD methodology is employed to identify critical components that significantly impact the system’s overall performance. For this purpose, a mathematical model is developed using Markov birth–death process and Chapman-Kolmogorov differential difference equations derived for steady state availability evaluation. The incorporation of exponential distribution for failure and repair rates, coupled with the Markovian technique, yields insights into the intricate variations within the system. Several goodness-of-fit metrics, such as R², MAE, RMSE, and collinearity diagnostics, are used to evaluate the performance of the proposed model. Results show that in this application, ANN performs better than regression analysis. The findings showcase the efficacy of the proposed stochastic framework in achieving remarkable improvements in availability. Numerical outcomes, meticulously presented in structured tables and figures, provide tangible evidence of the framework’s success. The novelty of the study lies in the strategic combination of these methodologies to achieve enhanced insights into availability improvement. By enhancing availability, the proposed framework directly influences production efficiency and overall plant performance. The findings of present work are valuable insights for industrial practitioners seeking resilient operational strategies.