With the development of oil pipeline transportation domains, the layout of regional oil pipeline transportation network is doubly influenced by regional development and its own evolution, more than that, it is a dynamic concept. In order to get more reasonable evaluation results, it is necessary to improve the traditional evaluation method by regarding the network layout behavior and regional space expansion behavior as interactive objects. In this paper, on the basis of analyzing regional development characteristics and pipeline transportation network layout demands from the spatiotemporal perspective, the co-evolution mechanism between them is dissected. Based on this, the multi-dimensional layout evaluation index system is constructed from four dimensions, and a layout evaluation model of oil pipeline transportation network based on multi-channel convolution and the Hidden-Markov model with co-evolution mechanism (i.e., CEM-MCNN-HMM) is proposed, which serves as a framework of co-evolutionary behavior recognition to use convolution kernel of Afferent sizes to extract feature information of different granularity from data in Afferent channels, effectively obtain the property features, behavior features, and interactive features of the behavior objects, and then convert the behavior recognition problem into a classification problem. The Hidden-Markov model is used to excavate the status dependent relations with a certain span of time and main, the classification results to improve the robustness of the model Finally, taking real data set as training data, performance testing of the proposed model is carried out from three aspects: the rationality test of using fractal dimension metrics and resetting fractal measured unit, and the model evaluation criteria based on confusion matrix analysis. The result shows that the performance of CEM-MCNN-HMM is best among all models and can improve the judgment level of the transport network layout.

中文翻译：

---

基于多通道卷积和协同进化机制的H-Markov模型的石油管网评价

随着石油管道运输领域的发展，区域石油管道运输网络布局受区域发展及其自身演变的双重影响，不仅是一个动态的概念。为了得到更合理的评价结果​​，需要改进传统的评价方法，将网络布局行为和区域空间扩展行为作为交互对象。本文在从时空角度分析区域发展特征和管网布局需求的基础上，剖析它们之间的协同演化机制。在此基础上，从四个维度构建多维布局评价指标体系，提出了一种基于多通道卷积和具有协同进化机制的Hidden-Markov模型（即CEM-MCNN-HMM）的输油管道运输网络布局评价模型，作为协同进化行为识别的框架利用传入大小的卷积核从传入通道的数据中提取不同粒度的特征信息，有效获取行为对象的属性特征、行为特征、交互特征，进而将行为识别问题转化为分类问题。Hidden-Markov模型用于挖掘具有一定时间跨度的状态依赖关系，主要的分类结果提高模型的鲁棒性，最后以真实数据集作为训练数据，所提出模型的性能测试从三个方面进行：使用分形维数度量和重置分形度量单位的合理性测试，以及基于混淆矩阵分析的模型评估标准。结果表明，CEM-MCNN-HMM的性能在所有模型中是最好的，可以提高交通网络布局的判断水平。