Aiming at the existing target allocation problem of air combat groups, a target allocation model based on target value estimation is proposed. It comprehensively considers the relative motion information of projectiles and targets and the inherent information of targets. By constructing the combat intention function, the model’s attention to high-value targets is increased on the basis of maintaining the original overall allocation capability, and the influence of command information on the model is enhanced. Secondly, the Dream Optimization Algorithm (DOA) is introduced. Through the efficient search ability of the Dream Optimization Algorithm in different dimensional solution Spaces, the problem that traditional heuristic algorithms tend to converge prematurely and have insufficient stability when facing this discrete multi-constraint problem is solved. Simulation experiments show that under the algorithm ranking test with the same computational complexity, the performance of the dream optimization algorithm is significantly better than that of the genetic algorithm and the particle swarm optimization algorithm.