随机观测误差对水环境评价的影响

水环境评价既存在模糊性，又不可避免随机性。以往评价大多只考虑了实测指标的物理权重的影响，而很少考虑指标实测过程中不可避免的随机观测误差的影响。本文先应用了最大熵原理来确定在给定约束条件(即已知信息)下，实测指标最小偏差的先验概率分布。在此基础上，应用蒙特卡罗法构造算例，进一步讨论了单一模糊模型和基于最大熵原理的相对隶属度模糊优化评价模型Ⅰ与Ⅱ，分为方差不均一和方差均一两种情况，分别研究了随机观测误差对上述水环境评价模型的影响。结果表明：随机观测误差对水环境评价的影响不容忽视，有时甚至直接改变评价等级；当观测精度有差异，尤其差异较大时，这种影响也随之加大，此时，评价结果只取决于某一(几)项随机观测误差最小的指标。

Influence of stochastic observation error on water environment evaluation

The principle of maximum entropy and Monte Carlo method are applied to study the influence of stochastic observation error on water environment evaluation. First, the least biased prior probability distribution of observed index under the given constraint condition is derived using the principle of maximum entropy. Then the influence of stochastic observation error on different model for environment evaluation in two cases, namely uniform variance and non-uniform variance, is investigated using Monte Carlo method. The result shows that the influence of stochastic observation error on water environment evaluation cannot be neglected for all models in all cases. It even affects the evaluation grade sometimes. Furthermore, if the observation error is remarkable the influence will increase correspondingly. In that case the evaluation rsult depends on some indexes or one of the indexes with minimum observation error.