This paper addresses the energy consumption optimization problem in Multi-UAV collaborative field data collection scenarios. We propose an Energy-Efficient Path Planning Model for Data Collection in Cooperative Multi-UAV Systems (EEPPM-DCCMUS) that jointly optimizes UAV deployment strategy, flight trajectory planning, velocity control, and device transmission power allocation to achieve three objectives: maximizing the minimum transmission rate, minimizing device communication energy consumption, and minimizing total system energy consumption. To solve the EEPPM-DCCMUS problem, we develop an Improved Multi-objective Artificial Hummingbird Algorithm (IMOAHA) that innovatively integrates K-means clustering initialization, Lévy flight-based mutation foraging strategy, and ant colony optimization foraging strategy. Simulation results demonstrate that IMOAHA significantly outperforms the baseline MOAHA algorithm in terms of Pareto front convergence and distribution. Specifically, IMOAHA achieves a minimum transmission rate of 12,754,066.74 bps (comparable to MOAHA), reduces total energy consumption to 12,138.94 J (a 58.9% improvement), decreases total flight distance to 365.05 m (a 53.7% reduction), and lowers energy consumption fluctuation by 81.3%. Compared with existing algorithms including MOAVOA and MOCryStAl, IMOAHA reduces total energy consumption by 25.8%–70.6% and total flight distance by 50.2%–66.3%. This study provides an innovative solution for wide-area coverage and dynamic load balancing of UAV clusters in complex agricultural environments, with significant theoretical and practical implications for promoting large-scale UAV applications in precision agriculture.