目的 针对海底管道长期腐蚀损伤的随机性、多态性等问题，研究不同服役年限的海底管道腐蚀损伤分布特性及演化规律。方法 提出了一种基于混合分布模型的海底管道腐蚀特性分析方法，该方法以老龄海底管道腐蚀损伤实测数据为基础，通过对不同服役年限下的海底管道腐蚀损伤特征进行统计分析，确定出海底管道腐蚀损伤的最佳分布模型，建立基于混合分布模型海底管道腐蚀损伤随机过程模型。进一步使用ARIMA方法对模型参数进行持续修正，并对管道腐蚀损伤进行预测。结果 不同服役时间下的腐蚀损伤最佳分布不同，Weibull、Gumbel及Gamma分布模型均具有较好的拟合优度，且各分布模型随服役时间的增长呈现出不同的变化趋势，相应分布模型参数呈现出动态变化。结合ARIMA模型修正方法，充分利用实测数据能够不断降低模型的不确定性。结论 海底管道的腐蚀损伤具有明显的多态性和随机性特征，很难采用单一分布模型对管道腐蚀损伤数据进行普遍性描述，既有的腐蚀损伤模型存在一定的不确定性及局限性，而基于混合分布模型的分析方法更能准确地反映海底管道长期腐蚀损伤的实际分布规律。

Aiming at the randomness and polymorphism of long-term corrosion damage of submarine pipeline, the distribution characteristics and evolution law of corrosion damage of submarine pipeline with different service years are studied. An analysis method of corrosion characteristics of submarine pipeline based on mixture distribution model is proposed. Based on the measured data of corrosion damage of aging submarine pipeline, through the statistical analysis of corrosion damage characteristics of submarine pipeline under different service years, the optimal distribution model of corrosion damage of submarine pipeline is determined, and the random process model of corrosion damage of submarine pipeline based on mixture distribution model is established. Further, Arima method is established to continuously modify the model parameters and predict the pipeline corrosion damage. The results show that the optimal distribution of corrosion damage is different under different service time. Weibull, Gumbel and Gamma distribution models have good goodness of fit, and each distribution model shows different change trends with the increase of service time, and the corresponding distribution model parameters show dynamic changes; Combined with ARIMA model correction method, making full use of the measured data can continuously reduce the uncertainty of the model. The corrosion damage of submarine pipeline has obvious characteristics of polymorphism and randomness. It is difficult to use a single distribution model to generally describe the pipeline corrosion damage data. The application of the existing corrosion damage model has certain uncertainty and limitations. The analysis method based on the mixture distribution model can more accurately reflect the actual distribution law of long-term corrosion damage of submarine pipeline.