摘要： A class of stochastic generalized linear complementarity problems with finitely many realizations is studied. Based on Expected value formulation and smoothing symmetric perturbed Fischer function,the stochastic generalized linear complementarity problems are reformulated as a system of smoothing equations. Then, a smoothing Newton method with nonmonotone line search strategy is presented to solve the new formulation. Moreover, it's proved that this nonmonotone smoothing algorithm is globally and local quadratically convergent under suitable assumptions.