The corrosion evolution of oil and gas transmission pipelines is highly complicated, and sufficient data on influencing factors are often difficult to obtain in actual operation. Additionally, traditional empirical models produce significant errors in long-term predictions. To more comprehensively characterize the dynamic characteristics associated with memory effects and measurement randomness in pipeline corrosion, this paper proposes a non-Markov Wiener process prediction model considering both measurement errors and historical dependency. Model parameters are estimated and updated using maximum likelihood estimation and Bayesian inference. Based on weak convergence theory and the definition of first-passage failure time, an approximate analytical solution for the distribution of corrosion depth is derived, enabling predictive assessment of internal corrosion progression. Finally, monitoring data from the inner wall of a natural gas pipeline in the Chongqing Gas Mine are used to verify the effectiveness of the proposed method.

The digital twin concept completes the mapping and interaction between physical space and digital space, showing great potential for development in the industrial field. With considering the low detection efficiency of natural gas dehydration performance parameters and the inability to optimize gas station process parameters online, this paper applies the digital twin concept in the chemical industry to establish an overall framework of the digital twin system for triethylene glycol(TEG) dehydration. On one hand, the geometric model of the twin system is constructed by integrating physical devices. On the other hand, the flow model dehydration system technology is established based on the real-time driving of physical data. Finally, the twin model of dehydration is established by designing virtual-real mapping model, completing the mapping of physical space and digital space, which enables the parallel operation of the physical device and the virtual device. Through the proposed digital twin system, real-time prediction of natural gas water dew point and other dehydration performance parameters can be achieved. To achieve the goal of low power consumption, the optimization of dehydration process parameters is realized by combining optimization algorithms with the twin model, thereby improving economic efficiency.