To address the difficulty of achieving coordinated coverage across multiple altitude layers with conventional ground base stations in 5G low-altitude networked UAV communications and the resulting low-altitude signal coverage gaps, a multi-objective base station layout optimization method based on a piecewise air-to-ground channel model and an improved NSGA-II (INSGA-II) is proposed. Using 22.5 m and 300 m as height thresholds, the propagation environment is divided into near-ground, low-altitude, and near-high-altitude regions. Probability-based path loss models incorporating line-of-sight (LOS) and non-line-of-sight (NLOS) propagation are established in accordance with 3GPP TR 38.901 and TR 36.777. The deployment problem is formulated as a dual-objective optimization model in which both the worst-layer coverage gap and the normalized number of base stations are minimized, subject to constraints on the signal-to-noise ratio (SNR), propagation delay, and minimum inter-site distance. Based on the minimax criterion, joint optimization is performed across three altitude layers at 200 m, 400 m, and 600 m. Three improvement mechanisms are incorporated into NSGA-II, namely heuristic initialization integrating three strategies, balanced dual-path bit-flip mutation, and local search based on Pareto representative points. The mean absolute errors between the analytical model and the shooting and bouncing rays (SBR) ray-tracing results are shown not to exceed 3.3 dB at any of the three altitude layers. With only six base stations, coverage rates of 100% at both the 200 m and 400 m layers and 96.14% at the 600 m layer are achieved by the recommended deployment scheme. Compared with NSGA-II, MOEA/D, and MOPSO, the fastest convergence, the highest hypervolume, and the best trade-off between network deployment cost and communication performance are achieved by INSGA-II. It is further demonstrated that the scale of base station deployment can be reduced while communication quality across multiple altitude layers is maintained. The findings are expected to provide a useful reference for coordinated multi-altitude planning and the engineering deployment of base stations in urban communication networks for low-altitude networked UAVs.