Research On Virtual-real Fusion Modeling And Intelligent Operation Optimization In Residual Hydrogenation Process

Residual hydrogenation technology is one of the most important,reliable and effective processing methods for heavy oil reuse in large oil refineries and petrochemical enterprises.However,the residue hydrogenation process has problems such as complicated process,low production transparency,and difficulty in operation optimization.Therefore,it is urgent to integrate mechanism analysis and industrial production data to build a virtual model of residual oil hydrogenation based on virtual-real fusion,that is,a digital twin model,and then introduce intelligent optimization methods to efficiently solve the expensive optimization problems involved in operational optimization.Based on the in-depth study of the residual oil hydrogenation process,reaction mechanism and operation optimization needs,this paper carried out the research on the virtual-real fusion modeling and intelligent operation optimization of the residual oil hydrogenation process.The main innovative research work is as follows:(1)Aiming at the problem that the mechanism of the actual residual oil hydrogenation process is complex and difficult to describe accurately,combined with the mechanism analysis,a digital twin model of residual oil hydrogenation based on HYSYS was built,and a digital twin model parameter of the residual oil hydrogenation process based on Na STA was proposed.The correction method adopts the neighborhood adaptive state transition algorithm including the variable local neighborhood strategy and the out-of-bounds candidate solution processing mechanism,and corrects the parameters of the digital twin model based on the virtual and real fusion of industrial production data,so as to realize the high precision of the digital twin model to the actual physical model.(2)Aiming at the problems that the digital twin model of residual oil hydrogenation is difficult to converge at some working points in the optimization process,which leads to time-consuming calculation.In addition,the industrial sampling period of the actual physical model of residual hydrogenation is long and the frequency is different,resulting in incomplete data and scarcity of valid data.Therefore,a new method is proposed.A proxy model construction method based on virtual reality fusion technology.The industrial sampling points of the actual physical model are complemented by the residual oil hydrogenation digital twin model,and the adaptive dynamic sampling Kriging proxy model construction method is designed to further solve the problem of small sample modeling and realize the efficient proxy for the residual oil hydrogenation process.(3)Aiming at the large scale and expensive calculation of the operation optimization of residual oil hydrogenation process,a research on parallel intelligent operation optimization of residual oil hydrogenation integrating digital twin model and surrogate model was proposed.The parallel state transition optimization algorithm is designed,and the model calling mechanism and task allocation acceleration strategy are introduced to efficiently solve the operation optimization problem of the residual hydrogenation process,and realize the rapid optimization under the condition of high precision of the industrial model.Combined with the methods mentioned above,a parallel intelligent operation optimization system for residual oil hydrogenation process was developed.The operation results show that the proposed method can effectively improve the optimization efficiency of operational optimization of residue hydrogenation process,and provide a practical and efficient solution for the actual residual hydrogenation process to achieve energy saving,emission reduction,cost reduction and efficiency increase.