Online monitoring and control of drilling stage can improve drilling quality and efficiency in laser drilling. The rapid development of machine learning technologies has facilitated online process monitoring toward a data-driven paradigm. However, the decision-making process of most of data-driven methods is binary decision, which would lead to excessive decision deviation and further cause large decision cost and low drilling quality. In this paper, a model revaluation method, called fusion weighted sequential three-way decision, is proposed. To solve the decision deviation caused by machine learning methods, the decision made by data-driven model is reevaluated based on the idea of sequential three-way decision and granular computing. And considering the characteristics of streaming decision and time-series in industrial process monitoring, a new decision-making mechanism, called forward aided three-way decision, is proposed. Unlike the traditional three-way decision, the prior information and previous decision results are used to assist the current decision of data-driven model in a weighted way instead of waiting for further information in this paper. Furthermore, a real case study of online monitoring and control of drilling stage have been conducted. Experimental results demonstrate that the proposed method can flexibly combine with different machine learning method. And the decision deviation can be greatly reduced. The research provides a new perspective for online monitoring and robust control of industrial process.