[en] The targets and the use of weapons are complicated on attacking ground target, Besides, the weapon-target assignment (WTA) research is extremely difficult and lacking. For a timely and rational WTA scheme can not only help to seize fleeting war opportunities, but also optimize the use of weapon resources to maximize the battlefield benefits with minimum costs, this article builds a mathematical model and uses genetic algorithm (GA) to find the optimal WTA result. But the algorithm is difficult to prepare better initial population because of the random assignment of variables which results in undesirable optimization results. To overcome the challenge in initializing GA, this article puts forward a variable probability assignment technique. This technique first uses feasible solutions to determine the probability distribution of the variable values, then, uses probability assignment mode to replace the stochastic uniform assignment mode to initialize variables, and produces the initial population individual close to the optimal individual with certainty. The simulation calculation shows that the improved GA can effectively solve the WTA problem of the large scale ground targets attacking with better performance. (paper)