Projection-Pursuit-Based Weighting for Assessing the Performance Enhancement of Isolation Piles in an Existing Tunnel–Deep Foundation Pit System

With the saturation of above-ground spatial resources, megacities are increasingly shifting their development focus to underground space. However, new underground construction inevitably disturbs adjacent structures. To ensure underground engineering safety, isolation piles are often employed in practice to mitigate such disturbances. It is therefore necessary to conduct a full-lifecycle assessment of construction disturbances and the performance enhancement effects of isolation piles. This study proposes a projection-pursuit-based weighting method for multi-indicator performance data derived from the existing tunnel–deep foundation pit system under construction disturbance, and uses it to assess the enhancement effect of isolation piles on system performance. Refined spatiotemporal coupled numerical simulations for scenarios with and without isolation piles are conducted using Plaxis 2D, and a multi-indicator performance dataset is established in accordance with relevant engineering codes. On this basis, a projection-pursuit-based weighting scheme is adopted to integrate multiple performance indicators into a composite performance measure. To solve the associated projection-direction optimization problem, a hybrid genetic algorithm incorporating elitist preservation, adaptive operator control, and real-coded SBX–Gaussian operators is employed. The results show that the installation of isolation piles markedly improves the comprehensive performance of the existing tunnel–deep foundation pit system throughout the construction process. Compared with a basic GA, the adopted hybrid design provides a more stable and effective solution strategy for the present projection-pursuit-based weighting task. The proposed method offers a practical data-driven tool for comparing mitigation scenarios and supporting system-level interpretation of construction disturbance in underground engineering. The present results are obtained for a case-derived problem under a 2D numerical modeling framework, with field validation mainly available for the excavation stage.