Neural Network-Assisted Cost-effective Attribute Control Chart for Weibull Data under Uncertainty

Accepted May 2026

Abstract

The np-control chart is widely used in industry to monitor the number of defective items. Existing np-control charts designed under classical statistics work when the data precise. The current np-control charts under uncertainty assume a fixed level of uncertainty. To address these limitations, this paper presents the design of an np-control under generalized interval statistics for the Weibull distribution, considering both data and uncertainty as random and correlated within a neural network framework. In addition, the proposed chart is designed by minimizing a cost model. The optimization of the proposed control chart is performed within the neural network framework. Extensive simulation results are presented, and the application of the proposed control chart is demonstrated using LED manufacturing data. The results show that the proposed control chart is more efficient than the existing chart in terms of cost. It also shows that the proposed chart is flexible under uncertainty and more effective in detecting shifts in the process compared to the existing control chart.