Thank you, CAB, thank you in his busy schedule to help me solve the problem.
My idea is: the whole data in a total of11 columns, changes in the column is sixth column and the eleventh column, let old data in the sixth column of the number of random changes, then the random changes after the number is multiplied by tenth row number, finally get the eleventh row number, finally let eleventh row number all together get the last line of the last digit, sixth columns of the sum of the number of total should and the original final line second the same number, finally is the new get the last line of the last digit should be larger than the original data of the last line of the last digit ( this is larger than the range can be set a digital ), my English is not very good, if I express is not very clear, you can run down my program, I think may be of help to you.
CAB, actually these in my program has been reflected, the only problem is that my program in the output format problems, you can also modify my program, join your output function, estimation can reach my goal.