基于径向基网络的含缺陷管道安全系数修正

安全系数是管道完整性评价方法的重要参数之一,传统评价方法仅考虑了管道所处地区等级对其影响,未能全面评价管道的真实安全状态。提取管道相关风险因素,比较各因素间相对重要程度并建立表征不同风险状态的分值体系,根据所得风险分值对安全系数进行修正,使得修正后的安全系数能够反映含缺陷管段实际风险状态;结合径向基网络方法的自学习特点,将各风险因素作为输入,修正后的安全系数作为输出,建立基于径向基网络的管道安全系数计算模型。由模型训练得出的修正安全系数与测试样本吻合度较好,验证了模型的准确性。结果表明:修正安全系数在考虑地区等级的基础上同时引入了其他重要风险因素的影响,为管道完整性评价中的安全系数取值提供了新的思路。

Modification of safety factor of defective pipeline based on radial basis function network

Safety factor is one of the important parameters for pipeline integrity assessment, but only the impact of the safety level of the place where the pipeline is located is considered in the traditional assessment method, so the actual safety status of pipeline cannot be assessed comprehensively. Here, the relevant risk factors of pipeline were extracted, the relative importance of factors was compared, a scoring system was established to feature the different risk states, and thereby, the safety factor was modified according to the risk score obtained, so that the modified safety factor could reflect the actual risk state of the defective pipeline section. Combined with the self-learning characteristics of the radial basis function network method, a calculation model of pipeline safety factor based on radical basis function network was established with risk factors as the input and the modified safety factor as the output. The calculated safety factor from the model training is in good agreement with the test sample value, which verifies the accuracy of this model. The results show that the modified safety factor introduces the influence of other important risk factors on the basis of the regional level, which provides a new idea for determining the value of safety factor in the pipeline integrity assessment.