By using the optimal value of lower-level stochastic programming as a response to feedback to a class of upper bi-level stochastic peogramming problem,the requirement for providing sole optimal solution may be relaxed to lower-level stochastic programming demand.This paper firstly discusses the convergence for the optimal value of lower-level stochastic programming approximation,then feedbacks the optimal value of lower-level stochastic programming to the upper level and finally obtains the upper-level semi-convergence of the optimal solution set sequence of upper stochastic programming approximation.