中文摘要：

针对风险评估中现有指标定权方法主观性较强的问题，从风险权重的概念出发，以港口自然灾害风险的危险性指标为例，通过设定损失分级和对应的指标损失程度范围，提出了“损失期望”概念；基于可获得的历史资料数据，引入防范因子进行指标的分样本客观定权，提高了权重计算方法的科学性和准确性。利用聚类一判别分析证明了该方法合理性与适用性，并将权重计算结果应用于贝叶斯网络建模过程，实现了港口自然灾害风险的量化评估。

英文摘要：

Focusing on the problem of the stronger subjectivity of current method deciding index weight in the process of risk assessment, and starting from the concept of risk weight, this paper takes port＇s natural disaster risk index as an example, presents concept of＂ loss expectation＂ and corresponding index loss ranges. Based on obtain- able historical imformation data, the paper introduces precautionary factors to carry out the objective weighting of the indices according to different samples, enhances scientificalness and accuracy of weight calculating method. The rationality and applicability of the method are verified by cluster and discrimination analysis, the calculated weights are applied to Bayes network modelling process, and the quantized assessment of the port＇s natural disaster risk is implemented.