A Method Research Based On Bayesian Network For Offshore Oil Platform Quality Traceability

Offshore oil platform is a kind of working and living offshore steel structure building for the petroleum exploitation, natural gas and other resources.Offshore oil platform type of production is a single piece production (production in small batch is extremely rare),so the difference between different offshore oil platform is very big and the customization is also high.Its production characteristics are small batch production, large differences between different projects, enterprise high degree of customization features, and its production process is often used "to build, and secondly to create "and so on. As the national strategy of "Marine Power" is proposed and the development of the Marine energy, more and more people pay attention to the large Marine engineering equipment such as offshore oil platform, and put forward higher requirements on the quality of the product at the same time.Based on offshore oil platform construction as business background, in view of the quality project construction process tracing difficult situation, through researching the particularity in offshore oil platform project conditions, the writer puts forward the core idea to use the process chain of quality traceability management mode analysis, and uses the Bayesian Network to establish traceability model quality of offshore oil platform.Finally by welding process instance, this paper prove the usefullless of the method.The main research contents are as follows in this paper:(1) According to the topic of the research background and objective, further study of the quality tracking related aspects, especially in the field of Marine engineering equipment quality tracking of the status quo, and other methods of quality tracing are discussed to make a profile. On the basis of the offshore oil platform characteristics and quality traceability status, the framework of the study on overall train of thought, research methods and thesis of this paper are given finally;(2) In terms of quality traceability management aspect, first of all, the characteristics of offshore oil platform is illustrated in this paper, then the concept of traceability is introduced ;And on the basis of the concept of traceability, two kinds of quality traceability management model are discussed thoroughly, which is based on tracking BOM quality traceability management and based on the activity and the quality of the product traceability management;The writer analyzes the advantages and disadvantages of these two kinds of model,and the concept of process chain is proposed in this paper , which is the groundwork to improve these models in this paper. In terms of the quality tracing analysis theory, the writer illustrates the popular mass retrospective analysis methods, such as Fault Analysis Tree (FAT),Neural Networks, Entropy theory, Bayesian Network (BN) models, and analyzes the advantages of each method and disadvantages.(3) For questions and inverse probability and "Hume problem", the writer explores the history and development of the Bayesian theory and the application naive Bayes theory in artificial intelligence (AI); According to the theory of Bayesian Network based on Bayesian theory and graph theory, the writer illustrates the expression parameterized network definition and Bayesian Networks; To deal with the complicated characteristics of offshore oil platform construction, the writer put forward a method to analysis the quality influence factors, which according to the horizontal and vertical to the construction process. The longitudinal analysis based on the idea of offshore oil platform construction process of layered; Horizontal analysis based on partition of professional - work - process, first of all, the project is divided into eight major, then subdivided into operations for each specialty, finally divided into process that can be completed directly; According to the characteristics of the offshore oil platform project of production and quality management of BOM, the writer put forwards the probability of BOM,namely P-BOM; By using the P-BOM as the node model of Bayesian Networks, it makes possible that qualitative analysis to quantitative calculation between offshore oil platform in different parts of the project and establishes the offshore oil platform project quality tracking model on the basis of the P-BOM; In terms of construction of Bayesian Network, the EM algorithm of parameter study is introduced; climbing method and SEM algorithm are introduced in Structure learning which based on BIC score; Others, the concept of clique is illustrated in Bayesian network inference and illustrates the quality traceability reasoning mechanism based on the clique tree propagation algorithm.(4) In this paper, the writer introduces two Bayesian network simulation platform: Hugin Expert and Matlab BNT and the contrast analysis was carried out on the two platforms.Through using the Hugin Expert, On the basis of model we had built, the writer build the Bayesian Network for the welding process, which is the most important process in the building of offshore oil platform. In terms of the quality of retrospective reasoning ability and quality prediction ability of the model, through the analysis of specific examples to demonstrate the usefulness of the proposed model.Finally, the method proposed in this paper and other methods are compared and analyzed,elaborated the model significance in the large discrete manufacturing such as offshore oil platform; Aiming at the quality management at present of the four modernizations demand(intelligent, modular, hierarchical, fine) ,this paper gives advice on application quality traceability model based on Bayesian Network.