Investigation And Application Of LSSVM In Safety Assessment For Railway Loading System Of Oil Depots

Railway loading space of oil depots is a special place for arriving and delivering combustible oil, and coordinates crude oil processing, product oil supplying and transporting. Once an accident happened in railway loading space of oil depots, the results will be serious. With the development of petrochemical industry, the safety assessment for the railway loading system of oil depots has already become indispensible to safety in production and safety management of oil depots. And choosing assessment methods which is the key of the work, influences the assessment results directly, furthermore, has significant meanings to improve the quality of the safety assessments for railway loading space of oil depots and ensure safety in production.In this paper, firstly, introduced common assessment methods systematically, concluded all the characteristics of themselves and their limitations in applications. Meanwhile, introduced a new machine learning method named SVM (Support Vector Machines), and fully dissertated the advantages of SVM in safety assessment by comparing with other traditional methods.Secondly, on the basis of principles of scientific, systemic, maneuverability, quantitative etc, in the sight of"Human-Machine-Environment-Management", fully considered the characteristics that there are more nonlinear factors influence the loading system of oil depots for safety in production, then, established the index system of safety assessment for railway loading system of oil depots , in which showed many important factors in the safe production process of the railway loading system of oil depots.Finally, combined the index system which had been established with the data that was collected from the oil depots affiliated with Jilin Branched of PetroChina Company Limited, built up a Least Square Support Vector Machine (LSSVM) model for safety assessment which is suited to assess the railway loading system of oil depots, then, completed the model training and test with MATLAB by its powerful computing and visualization functions.Guided by the theory of safety system engineering, applied LSSVM in system safety analysis and assessment, which can provide more effective ways and methods for safety management and scientific decision-making in railway loading system of oil depots and scientific, rational, qualitative and quantitative safety assessment in safety assessment agencies.