Aiming at disadvantages of traditional weapon target assignment (WTA) methods which treat targets as mutually independent entities, an associated target model being the same with anti-tank missile target assignment was proposed. A combined algorithm based on self-adaptive genetic algorithm for solving weapon target assignment problem was presented. A new coding way of original swarm was established, which reduced the solution space. By introducing single ancestor chiasmatypy with double-single ancestor chiasmatypy, the inchoate convergent character was avoided with searching accelerated. Simulation results demonstrate that the algorithm can converge to the ideal steady state with high speed and provide practical solutions.