Resource allocation in multiple energy-integrated biorefinery using neuroevolution and mathematical optimization

Wai Mun Chan, Dinh Van Khoa Le, Zhiyuan Chen, Jully Tan, Irene Mei Leng Chew

Research output: Contribution to journalArticleResearchpeer-review

7 Citations (Scopus)


A multiproduct lignocellulosic biorefinery converts various types of biomass into value-added products or energy through different conversion pathways. However, its operation is susceptible to the changing nature of biomass properties, biomass feedstock supply, ambient temperature, and product demands. Therefore, a new optimal resource allocation scheme must be devised instantly upon detecting any fluctuations in the biorefinery to avoid oversupply or undersupply issues. Previous literature on biorefinery resource allocation uses a mainly nonlinear programming approach that assumes steady state for all parameters during simulation; this may result in a delay of response time due to the time taken during the optimization stage. In this paper, a resource allocation system based on deep neural network (DNN) is proposed for the biorefinery. The input nodes of the DNN are the parameters that undergo fluctuations while the output nodes are the flowrate allocation of biomass to different chemical and energy conversion pathways. The connection weights and topology of the DNN are optimized using the neuro-differential evolution (NDE) algorithm. The optimization results of the DNN yields an average optimality of 97.7% and reduces the response time by 99.5% as compared to the conventional nonlinear solver. The proposed DNN-NDE framework accounts for both responsiveness and cost performance during the synthesis of a smart resource allocation system. Graphical abstract: [Figure not available: see fulltext.]

Original languageEnglish
Pages (from-to)383-416
Number of pages34
JournalProcess Integration and Optimization for Sustainability
Issue number3
Publication statusPublished - Sept 2021


  • Artificial intelligence
  • Biomass
  • Integrated energy
  • Neural network
  • Smart manufacturing
  • Sustainable engineering

Cite this