AI-assisted computational design with synchronous grouting materials in the framework of industrial solid waste

The use of industrial solid waste as supplementary cementitious materials (SCMs) for cement to improve the properties of cementitious materials has received much attention. In this paper, an AI-assisted computational design method is proposed, which can utilize SCMs to prepare high-performance synchronous grouting materials. Initially, the Itransformer model for predicting the performance of grouting materials was developed, and the advanced Ivy algorithm was used to optimize the hyperparameters of the model. Based on the model prediction results, 1-d compressive strength (CS), fluidity and carbon emission were used as optimization objectives, and the material proportions were inverted by the multi-objective slime mould algorithm (MOSMA). The validity of the proposed design method is verified in terms of macroscopic tests and microscopic mechanisms by taking an example of a metro line in China. The results show that: (1) The Ivy-itransformer model has good prediction accuracy with R 2 of 0.938–0.996 for the prediction of grouting material performances; (2) The proposed method can improve the 1 d CS of synchronous grouting materials by 6.52 % and the flowability by 11.94 %, and also reduce the carbon emissions by 16.41 %; (3) Artificial intelligence algorithms can quickly determine the optimal amount of industrial solid waste, thus realizing a dual improvement in the macroscopic properties and microstructure of the grouting material. This paper innovatively used a data-driven model to optimize the material design process and achieve fast and stable design of high-performance grouting materials. The proposed method provides a powerful tool to promote the high-quality and green development of underground engineering. • Combining DL and MOO for performance optimization of grouting materials. • The macro performances of the grouting material are investigated. • The micro properties are analyzed using SEM, rheometer and MIP. • Applicability of the proposed method are verified via a field engineering case.