Aiming at the energy efficiency optimization difficulties caused by the multi-component coupling and hierarchical structure of industrial robot electrical drive systems, this paper proposes a multilevel genetic algorithm (MGA) co-optimization method. First, an improved dq-axis motor model integrating iron loss, saturation effect and temperature influence is established to define a multi-constraint optimization problem with the objective of minimizing the total energy consumption of the system (covering the motor, inverter and transmission loss). Second, a hierarchy-dependent genetic coding scheme is designed to express the variable structure design space through hierarchical description with prefix tagging method, and the adapted genetic operators are developed. In ZDT1/3/4 tests, the MGA improves the hypervolume (HV) by 3.3%~6.2% compared with the conventional GA, increases the independent solution ratio by 4.5%~7.1%, and reduces the generation distance (GD) and inverse generation distance (IGD) by up to 72% (e.g., the IGD of ZDT4 is reduced from 0.0304 to 0.0084). In the drive system layout optimization, the convergence speed of MGA is improved by a factor of 2.7 over GA with objective function values of 4.394 and 4.311, respectively. Based on the multi-electrical aircraft load management experiments, the system achieves 98.66% energy efficiency under healthy working conditions, 35kW load shedding by priority optimization when the main generator fails, and a 21-fold improvement in the computational efficiency of the hierarchical control strategy (23.41 seconds vs. 8.35 minutes for a single layer).
