GA-BPNN-Based Multi-objective Optimization Framework for Synchronous Grouting in Shield Tunneling

Abstract

The performance optimization of shield synchronous grouting materials is a key technical problem in controlling stratum deformation and ensuring structural stability in tunnel engineering. Traditional proportioning design methods rely on empirical trial and error and have bottlenecks such as poor multi-objective synergy and weak dynamic adaptability. This study proposes a collaborative optimization strategy of genetic algorithm (GA) and BP neural network (BPNN), which achieves a multi-objective dynamic balance of fluidity, strength and timeliness of grouting materials by fusing global search and local optimization mechanisms. Based on 126 sets of experimental data on grouting materials, a multi-modal data set was constructed, covering the proportion parameters and corresponding performance indicators of the cement-fly ash-bentonite system. Combined with dynamic boundary conditions such as propulsion speed and grouting pressure in shield construction, a GA-BP collaborative optimization framework was designed. Using a genetic algorithm and BP neural network series coupling strategy, based on twelve sets of experimental data, a multi-objective optimization study of synchronous grouting materials was conducted, and the optimal solution was obtained through fifty iterations of simulation. Experiments show that the optimal ratio scheme (water-binder ratio 0.38, bentonite content 12%) generated by the synergistic strategy in clay formation makes the slurry fluidity reach 248mm, 17.6% higher than the empirical ratio. The 3d compressive strength is increased by 22.4% to 1.85 MPa, and the initial setting time is shortened to 5.8 h. Through multi-objective Pareto solution set analysis, the solution space coverage of the collaborative strategy is increased to 89.2%, which is 29.7% higher than that of the single genetic algorithm, and the number of convergence iterations is reduced by 41.3%. In the field verification, the optimized scheme controls the segment staggering amount within 2.3 mm, which is reduced by 36.1% compared with the traditional method, and the standard deviation of surface settlement is reduced from 4.5 mm to 2.1 mm. Given the sudden working conditions of gravel formation, the adaptive adjustment response time of the model is shortened to 7.5 min, the slurry utilization rate is increased to 92.4%, and the single-ring grouting cost is saved by 13.8 yuan. The research confirms that the collaborative strategy effectively solves the problems of performance imbalance and engineering adaptability in multi-objective optimization of grouting materials and provides a new technical path for intelligent construction of shield tunnels.