Evaluation of disaster-bearing capacity for natural gas pipeline under third-party damage based on optimized probabilistic neural network

Third-party damage has become a significant factor leading to pipeline failures and leaks, resulting in severe ecological damage and human casualties. Traditional safety risk analysis methods in the past have focused on severing the causal chain of accidents, while neglecting the unique characteristics of different pipeline segments and environmental factors on their ability to bearing disasters. This paper presents a framework for evaluating the disaster-bearing capacity of natural gas pipeline third-party damage based on probabilistic neural networks. A comprehensive disaster-bearing capacity index system for evaluating third-party damage to natural gas pipelines has been constructed by analyzing the causal factors and potential consequences of third-party damage risk incidents. Next, the cluster analysis and indicator weighting information is used for achieving the comprehensive disaster-bearing capacity assessment. Finally, the data samples are constructed with the index levels and comprehensive disaster-bearing capacity levels of each evaluation unit. An optimization algorithm, Particle Swarm Optimization (PSO), is utilized to improve the parameter selection process of the Probabilistic Neural Network (PNN) model, in order to evaluate the disaster-bearing level of each unit. A typical long-distance natural gas pipeline in Zhejiang Province was selected as the case study, and a total of 200 research units were obtained through pipeline analysis. Furthermore, 14 indicator data were obtained by leveraging multiple sources of data surrounding the pipeline. With the utilization of cluster analysis, it was determined that Gaussian mixed model clustering method ultimately identified a total of 5 different levels of disaster-bearing capacity. The optimal smoothing factor can be accurate from 0.5 to 0.588 by using the PSO-PNN model, and the final accuracy rate is improved from 88% to 98%, which is obviously the introduction of PSO optimization algorithm in the PNN model can improve the accuracy of the model. This represents a significant improvement in accuracy when compared to traditional PNN models. The proposed evaluation method can provide scientific basis for the pre-vention and control of third-party damage in oil and gas pipelines and pipeline planning.

Automated summary

This paper presents a framework for evaluating the disaster-bearing capacity of natural gas pipeline third-party damage based on probabilistic neural networks. With the utilization of cluster analysis, it was determined that Gaussian mixed model clustering method ultimately identified a total of 5 different levels of disaster-bearing capacity. The proposed evaluation method can provide scientific basis for the prevention and control of third-party damage in oil and gas pipelines and pipeline planning.