A multi-model ensemble approach to process optimization considering model uncertainty

ABSTRACT

Traditional approaches in constructing response surface models typically ignore model uncertainty. If the relationship between the input factors and output characteristics of a process is very complex, traditional model building approaches may have limited effectiveness. In this paper, we propose a multi-model ensemble and then implement this ensemble model to optimize the process performance. To form a multi-model ensemble, we need to determine the weights of the different models, that is, values indicating relative importance among the models. To determine the weights, a hybrid weighting method is proposed, in which both global and local weighting methods are taken into account. Based on the hybrid weights of different models, a multi-model ensemble is built and optimized. An example is illustrated to verify the effectiveness of the proposed approach. The results show that the proposed model can achieve more accurate predictive capability and that a better process improvement is reached.

KEYWORDS:

• Multi-model ensemble
• model uncertainty
• hybrid weight
• process optimization

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China [71571102], which is gratefully acknowledged.

Notes on contributors

Ke-Ning Liu

Ke-ning Liu is an assistant professor in the Department of Marketing at the Ludong University. Her research activities include supply chain management, asymmetric information, and quality management.