Synergic optimization of pollution prevention and resource recovery of secondary lead smelting industry based on two-stage BPNLP network model

• Published in: Journal of cleaner production 2021-02, Vol.284, p.124717, Article 124717
• Main Authors: Li, Yanping, Zhang, Xin, Yang, Yi, Guo, Xiyun, Zhi, Jing, Zhao, YaZhou, Guo, Jianxin
• Format: Article
• Language: English
• Subjects: Pollution prevention; Resource recovery efficiency; Secondary lead smelting; Synergic optimization; Two-stage BPNLP network Model
• ISSN: 0959-6526; 1879-1786
• DOI: 10.1016/j.jclepro.2020.124717

Environmental pollution and low resource utilization efficiency caused by pyrometallurgy have seriously restricted the sustainable development of secondary lead smelting industry. A large number of studies focus on the research of hydrometallurgy to replace pyrometallurgy aiming at pollution prevention. However, due to economic cost, quality of secondary lead product and secondary pollution, hydrometallurgy hasn’t been widely applied by now. At present, most of the researches of pyrometallurgy focus on its environmental pollution assessment, or merely focus on the recovery efficiency of secondary lead. There are fewer researches on the synergic optimization methodology of pollution prevention and resource recovery for secondary lead smelting process. BP neural network is a type of model widely used in industrial processes optimization. While the gradient method adopted by the BP neural network often falls into a local minimum. If the fitting function is too complicated, there may be multiple fitting local minima, which will cause great obstacles to practical applications. Therefore, according to the characteristics of secondary lead smelting process, this paper uses a BP neural network-based nonlinear optimization combined with an approximate global optimization method to obtain the optimal smelting parameters. The innovative feature of this paper is that after the first stage of the neural network preprocessing has formed the fitting of the smelting process, the smelting parameters optimization process is carried out based on the fitted neural network model. According to the characteristics of the entire two-stage process, as well as the corresponding conditional constraints and the smoothing process of the BP neural network output function, we have formed a two-stage BPNLP network model with formally unified constraint by adopting the approximate global optimal search method. Compared with the average level of pollution prevention and lead recovery efficiency of pyrometallurgy technology in China, the pollution load Sulphur, lead, arsenic and cadmium in smelting flue gas per unit secondary lead product has been reduced by 78.40%, 52.00%, 72.63% and 16.00% respectively, and the recovery efficiency of lead increased by 8.85%. Thus, the two-stage BPNLP model can achieve the goal of synergic optimization of pollution prevention and resource recovery of pyrometallurgy technology of secondary lead smelting industry which can also provide effective methodologic