Abstract: In broiler brooding houses, the accumulation of surface manure on litter significantly increases the risk of diseases such as coccidiosis and colibacillosis, and also deteriorates indoor air quality due to the release of ammonia, hydrogen sulfide, and methane. However, existing double-crank shoveling-throwing mechanisms for surface manure cleaning suffered from low operational efficiency, excessive vibration, and poor overall performance, mainly because the key structural parameters were determined empirically without systematic multi-objective optimization. To address these problems, this study aimed to develop a multi-objective optimization method for the double-crank shoveling-throwing mechanism based on an improved NSGA-II algorithm, thereby achieving synergistic improvements in shoveling efficiency, energy consumption, and operational stability. A kinematic and dynamic model of the double-crank shoveling-throwing mechanism was first established to reveal the influence of structural parameters on the shoveling trajectory and force characteristics. A coupled ADAMS-EDEM simulation model was then built to simulate the interaction between the mechanism and manure particles. Using the optimal Latin hypercube sampling method, 60 sets of design variable samples were generated. A Gaussian process regression (GPR) surrogate model was constructed to map the relationship between six design variables and the shovel throwing quality Q. The relative error between the surrogate model prediction and the simulation result was 3.85%, indicating high prediction accuracy. To overcome the limitations of the standard NSGA-II algorithm—namely, significant dimensional differences among the three objectives, highly nonlinear parameter-performance mapping, and discontinuous parameter space—an improved NSGA-II algorithm was proposed. Three targeted improvements were introduced: (1) an adaptive normalization mechanism to eliminate dimensional effects; (2) a hybrid optimization framework combining global search (NSGA-II) and local refinement (sequential quadratic programming, SQP) to improve convergence accuracy; and (3) an improved crowding distance and solution selection mechanism to enhance the distribution uniformity of the Pareto front. The improved algorithm was compared with standard NSGA-II, NSGA-III, MOPSO, and MOEA/D using three performance indicators: Inverted Generational Distance (IGD), Spacing, and Hypervolume (HV). The comparative results showed that the improved NSGA-II algorithm significantly outperformed the other algorithms. Specifically, the IGD value decreased by 71% compared with standard NSGA-II, 60% compared with NSGA-III, 68% compared with MOPSO, and 64% compared with MOEA/D. The Spacing value decreased by 26% compared with standard NSGA-II, and the HV value increased by 16%. These results indicated that the improved algorithm achieved better convergence, more uniform distribution, and higher coverage of the true Pareto front. Using the entropy-weighted TOPSIS method, the optimal compromise solution was selected from the Pareto set, with recommended parameters: l₁=70.4 mm, l₂=88.5 mm, l₃=100.2 mm, l₄=30.3 mm, b=99.7 mm, β=37.4°. A prototype double-crank shoveling-throwing device was manufactured based on the optimized parameters, and field experiments were conducted in a brooding house in Zhouzhi County, Xi’an, Shaanxi Province, in December 2025. The experimental conditions were as follows: litter layer with a thickness of 50–60 mm, chicken manure layer thickness of 10-20 mm, chicken manure layer density of 0.6 g/cm³, litter layer density of 0.237 g/cm³, manure moisture content of 30%–35%, and three repetitions. The results demonstrated that the optimized mechanism improved shoveling efficiency by 138.24%, reduced driving torque by 35.12%, and decreased the angular acceleration peak at the shovel end by 35.60%. In addition, the surface manure residual rate decreased from 18.45% to 8.13%, a reduction of 55.99%, indicating significantly improved cleaning quality. This study provides a systematic multi-objective optimization framework for the double-crank shoveling-throwing mechanism, integrating ADAMS-EDEM coupling simulation, GPR surrogate modeling, and an improved NSGA-II algorithm. The proposed framework effectively addresses the challenges of dimensional differences, high nonlinearity, and high computational cost in engineering optimization problems. The improved NSGA-II algorithm demonstrates superior convergence and distribution performance compared with mainstream multi-objective algorithms. The optimized mechanism achieves synergistic improvements in shoveling efficiency, energy consumption, and operational stability, offering a complete technical pathway and replicable paradigm for the performance enhancement of hinge-type multi-bar agricultural mechanisms, particularly for manure cleaning equipment in the poultry industry.