Research On Abnormal Data Recognition And Intelligent Monitoring Of Underground Oil And Gas Pipeline Network

Underground oil and gas transportation pipeline is an important oil and gas transportation device,which plays an indispensable role in the national economy and construction.However,as the oil and gas transported in the pipeline belongs to class a dangerous goods,which is flammable and explosive,and the underground environment is relatively complex,the pipeline is vulnerable to damage and destruction,which may cause serious resource loss and environmental pollution,and even bring heavy casualties.At present,the academic research on the management of underground oil and gas pipeline network is mostly limited to the establishment of its safety assessment system and "incident early warning" type monitoring system,which can only passively reduce the loss of buried pipeline in case of accident,and does not fully tap the potential information contained in the operation data of oil and gas pipeline.Therefore,how to use big data analysis and processing technology to establish an intelligent online monitoring and evaluation system for each key node of the underground oil and gas pipeline network to realize the early warning of risk is a problem that needs in-depth study.The purpose of this study is to move forward the risk disposal port of the underground oil and gas pipeline network.Combined with the characteristics of Shewhart Control Chart theory and Probabilistic Neural Network(PNN),the BP Neural Network is integrated to establish the risk judgment model.Finally,the intelligent monitoring of the underground oil and gas pipeline network is realized by mining the potential rules of the oil and gas pipeline network data.The research work of this paper is mainly reflected in the following four aspects:Firstly,according to the traditional management mode of "prevention disposal" and "incident warning" of underground oil and gas network,the abnormal data are screened by using Shewhart Control Chart,and the risk disposal port is moved forward by detecting the gradual evolution of the hazard sources by combining the characteristics of real-time processing data of Probabilistic Neural Network;Secondly,the Error Feedback Probabilistic Neural network(EF-PNN)is proposed.The weight matrix of input layer and mode layer is optimized to improve the Probabilistic Neural Network.At the same time,the spatial complexity is reduced and the computational accuracy is improved.The results of open source data sets show that EF-PNN has a significant improvement in both computational efficiency and accuracy;Thirdly,a new classification criterion is proposed by combining EF-PNN and BP network.Among them,EF-PNN is used as the decision-maker of pre abnormal data division,and multiple BP networks are trained by effectively dividing the training data,which improves the operation efficiency and recognition accuracy of the corresponding data processed by the same BP network;Finally,the system method is applied to process the data of oil and gas pipeline network in the actual industry,and the results further illustrate the effectiveness of the improved model in the actual industry.Through the combination of management ideas and machine learning methods,this study provides new methods and new ideas for the intelligent monitoring of underground oil and gas pipeline network facilities safety,provides new research methods for the field of intelligent online early warning theory,and provides new ways for improving the data processing ability of machine learning algorithm.