讨论了适于估计起迄点出行分布矩阵 ( OD矩阵 )的交通检测点的合理分布问题 .根据检测点应当满足的规则 ,建立了关于检测点分布的非线性规划模型 .在已知极点间转移概率的前提下 ,将检测点的分布问题描述成一个平均报酬 Markov决策过程 ,并通过转化为一个等价的整数线性规划问题来求解 .最后实例结果表明该模型是有效的、合理的 This paper discusses the reasonable distribution of traffic detection points which are suitable for estimating the starting and ending point distribution matrix (OD matrix) .A nonlinear programming model about the detection point distribution is established according to the rules that the detection points should satisfy. Probability, the problem of distribution of detection points is described as an average reward Markov decision process, which is solved by transforming into an equivalent integer linear programming problem.Finally, the result of the example shows that the model is effective and reasonable