A multi-objective robust optimization scheme for reducing optimization performance deterioration caused by fluctuation of decision parameters in chemical processes. (2nd November 2018)

• Record Type: Journal Article
• Authors: Zhiqiang, Liao; Taifu, Li; Peng, Chen; Shilun, Zuo
• Is Part Of: Computers & chemical engineering. Volume 119(2018)
• Page Start: 1
• Page End: 12
• Publication Date: 2018-11-02
• Subjects: Decision parameters fluctuation -- Robustness evaluation metric -- Chemical process -- Multi-objective robust optimization; Chemical engineering -- Data processing -- Periodicals; 660.0285
• DOI: 10.1016/j.compchemeng.2018.08.037
• Languages: English
• ISSNs: 0098-1354

Abstract

Highlights: The designed evaluation metric can elaborate the relationship between robustness and fluctuation. The proposed method can assess the performance robustness. We built the model of chemical process based on decision parameters and output of performance. The robust optimization method can reduce the optimization performance deterioration. The multi-objective robust optimization superior to single-objective weighted robust optimization. Abstract: The fluctuation of decision parameters will deviate from the optimal decision, which will have significant impact on the optimization performance of chemical processes. To reduce optimization performance deterioration caused by fluctuation of decision parameters in chemical processes, a multi-objective robust optimization scheme is developed to assess performance robustness. In addition, based on the model that maps decision parameters to objective performance through neural network, a new robustness evaluation metric is created as the fitness value of the multi-objective evolutionary algorithm (for improving the strength Pareto evolutionary algorithm (SPEAII)) to elaborate the relationship between robustness and fluctuation. The efficacy of the proposed method is verified with HCN production process application by comparing with genetic algorithm (GA) and weighted single-objective robust optimization.