Addressing the issues of strong subjectivity， poor dynamic adaptability， and lack of self-learning capabilities in traditional battlefield equipment mobile repair task allocation models. It analyzes the task allocation problem under scenarios involving multiple locations， multiple tasks， and multiple teams. A decision optimization objective function is established based on equipment combat effectiveness restoration levels and repair cost consumption. A battlefield equipment mobile repair task allocation decision model is then constructed with repair time as a hard constraint condition. To address the multi-dimensional decision-making problem， we propose a multi-objective particle swarm optimization （MOPSO） algorithm based on an improved Levy-flight （LF） model. This algorithm effectively avoids local convergence and reduces the dimensionality of the optimization problem， demonstrating unique advantages in terms of convergence， dynamic adaptability， diversity， and robustness in dynamic multi-objective scenarios. Experimental results show that this model can be used to solve the task allocation problem in emergency battlefield equipment repair operations， and that the solution algorithm is stable to a certain degree.