本文通过对有效和非有效决策单元施加限制或引入正则限制条件下提出了三种改进的博弈交叉效率评价方法,与传统博弈交叉效率评价方法相比,改进方法不仅利于在相同的权重空间中对不同决策单元进行比较,而且在收敛速度上有明显改善,在一定程度上可消除传统博弈交叉效率结果的不惟一问题.结果表明对所有决策单元施加限制可提高对决策单元的区分度,从而改善结果的不惟一性,施加全局限制可提高迭代算法的收敛性.利用改进模型对中国9个城市的轨道交通企业运营效率进行了评价,评价结果也表明改进方法在区分决策单元上有更好的表现,收敛速度占绝对优势,得到了更为可信的评价结果. In this paper, we propose three improved evaluation methods for game cross efficiency by imposing restrictions on valid and ineffective decision-making units or introducing regular constraints. Compared with the traditional game cross-efficiency evaluation methods, the improved method not only benefits in the same weight space The different decision-making units are compared, and there is a clear improvement in the convergence speed, which can eliminate the non-unique problem of the cross-efficiency results of the traditional game to a certain extent.The results show that the restriction on all the decision-making units can improve the differentiation degree of decision-making unit, The result is not unique and the application of global constraints can improve the convergence of the iterative algorithm.Using the improved model to evaluate the operational efficiency of rail transit enterprises in nine Chinese cities, the evaluation results also show that the improved method has better performance in distinguishing decision-making units Performance, convergence rate accounted for the absolute advantage, has been more credible evaluation results.