AUTHOR=Wang Shengdi , Chen Junzhi , Zhang Yonggang , Qiao Xiaowei , Zhang Zhiping , Wang Xin 

TITLE=Determination of similar material proportions based on orthogonal experiments and neural network optimization in the goaf area

JOURNAL=Frontiers in Earth Science

VOLUME=Volume 13 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/earth-science/articles/10.3389/feart.2025.1582941

DOI=10.3389/feart.2025.1582941

ISSN=2296-6463

ABSTRACT=The stability of the surrounding rock in the goaf of the mine is poor, which can easily cause collapse disasters in the mining area. This paper used orthogonal experiments and multi factor optimization methods to study the optimal mixing ratio of similar materials for limestone surrounding rock in a goaf of a certain iron mine in Yunnan, and introduced a new material combination (quartz, cement, gypsum, barite, and glycerol), and utilized orthogonal experiments alongside machine learning techniques for predictive analysis. Systematic testing of similar material samples with 25 different mixing ratios yielded extensive data on various physical and mechanical parameters, which were then utilized to reveal the complex interactions among various influencing factors.The cement-gypsum-ratio significantly influenced the uniaxial compressive strength and elastic modulus, while the barite content influenced the density, and the glycerol content impacted the internal friction angle. Furthermore, this study proposed a novel machine learning-based prediction model that utilizes a PSO-BP neural network to regress and predict experimental data. Compared with traditional BP neural network, the results show that the PSO-BP model has a higher prediction correlation coefficient R2 than the traditional BP model, while the root mean square error (RMSE) and mean absolute error (MAE) are lower than the traditional BP model, indicating that the PSO-BP model has better and more stable prediction performance.So PSO-BP neural network model can more accurately predict the optimal mixing ratio for similar materials. The effectiveness of this mixing ratio was verified through practical engineering case studies. This study provides new theoretical foundations and technical support for the stability analysis of surrounding rock in goafs, demonstrating significant engineering application value.