Multi-objective optimisation of shield synchronous grouting materials: a synergistic architecture integrating intelligent algorithms and convolutional neural network

This study addresses the multi-objective optimisation problem involved in designing shield synchronous grouting material ratios by proposing a collaborative architecture that integrates an improved multi-objective grey wolf optimisation algorithm and a one-dimensional convolutional neural network.A high-precision surrogate model constructed by the one-dimensional convolutional neural network accurately predicts material strength, achieving a test set R 2 of 0.961 and a root mean square error of 3.12 MPa.The improved multi-objective grey wolf optimisation algorithm is then applied to simultaneously optimise both material strength and cost.Experimental results indicate that the proposed method outperforms comparison algorithms across multiple performance indicators, including inverted generational distance (0.038), hypervolume (0.752), and spacing (0.015).These outcomes confirm the effectiveness of the architecture in enhancing optimisation efficiency and solution set quality, offering a practical and intelligent approach to grouting material design.