Using an enhanced ant lion optimizer to improve artificial neural networks for the prediction of food-grade sodium alginate treatment effects for ready-mixed concrete plant wastewater

• Published in: Bulletin of engineering geology and the environment 2023-07, Vol.82 (7), Article 287
• Main Authors: Zhou, Zhong, Ding, Haohui, Zhang, Junjie, Yang, Hao
• Format: Article
• Language: English
• Subjects: Earth and Environmental Science; Earth Sciences; Foundations; Geoecology/Natural Processes; Geoengineering; Geotechnical Engineering & Applied Earth Sciences; Hydraulics; Nature Conservation; Original Paper
• ISSN: 1435-9529; 1435-9537
• DOI: 10.1007/s10064-023-03286-1

The harmless treatment of high-pollution geotechnical wastewater is of considerable importance to protect the surrounding ecological environment and expand the potential reuse utilization value of wastewater. Therefore, accurate prediction of wastewater treatment effect is an important guideline for precise control of wastewater treatment procedures. In this paper, a new prediction model, namely, LSALO-ANN, is proposed. The lightweight self-adaption ant lion optimizer (LSALO) was used to optimize the prediction model of artificial neural network (ANN). Taking the Dongqi tunnel supporting a ready-mixed concrete plant in Guangxi Province as a research case, food-grade sodium alginate is innovatively used to green treat this type of wastewater, and the turbidity removal rate of this kind of geotechnical wastewater can reach 98.77% under optimal conditions. Relevant experimental data is used as the data set to realize the establishment and validation of the prediction model. In addition, the LSALO algorithm is compared with particle swarm optimization (PSO) and genetic algorithm (GA). Prediction model test results show that the LSALO-ANN prediction model has a mean relative error (MRE) of 8.37% and goodness of fit ( R 2 ) of 0.9786 for the test set data, which is superior to 11.91% and 0.9625 of PSO-ANN and 12.41% and 0.9596 of GA-ANN considering prediction accuracy and generalization capability, respectively. The results prove that the LSALO algorithm effectively improved the reliability of prediction data output by ANN and further provide an effective method for real-time adjustment of wastewater treatment process parameters.