A-PSO-Based TEM inversion for efficient seepage detection in earth–rock dams

Accurate identification of seepage hazards in earth – rock dams is essential for ensuring hydraulic safety. However, conventional transient electromagnetic (TEM) inversion algorithms often suffer from low computational efficiency, solution non-uniqueness, and insufficient resolution for thin low-resistivity layers. To address these limitations, this study proposes an Anchor-guided Particle Swarm Optimisation (A-PSO) method that incorporates resistivity – depth anchors extracted from smoke-ring calculations as physics-informed trend constraints. These anchors reshape the inversion search space by narrowing feasible resistivity intervals and enforcing layer-wise trend continuity, thereby improving global convergence and inversion stability. Numerical experiments show that A-PSO reduces the initial RMS error by 60–92% and achieves target misfits with only 1/18–1/22 of the iterations required by conventional PSO, while more robustly recovering thin interlayers under over-layered discretisation. From an engineering perspective, A-PSO provides clearer and more continuous low-resistivity anomalies and exhibits markedly better agreement with borehole verification. Application to a real earth – rock dam successfully identifies two deep high-permeability zones associated with potential seepage pathways, demonstrating that A-PSO is an efficient and reliable inversion tool for rapid seepage localisation and risk assessment in complex dam foundations.