Pairwise combinatorial coverage testing which is a significant method for test data generation in functional test fields was discussed. As the method is difficult of solution, a improved method of test data generation for the solution space tree based on intelligent planning was proposed according to the relationship of different factors in the system.The method generates all the available data into a solution space tree, then searches a path from a root node to a leaf node to produce a test data.Experimental results show that this method can test adequately and cover more effect factors of software system by using less test data and can achieve the combinatorial corerage of two factors for test data. As compared with traditional methods, it can obtain the simplest and more complete data sets.In conclusion, it improves the efficiency and veracity of generating test data when this method is taken to a bear space camera software test.