Data Modeling And Energy Saving Optimization Of High Sulfur Natural Gas Sweetening

With the enhancement of environmental awareness,natural gas as a clean energy has been widely valued,and its demand is increasing year by year.But more than 25% of natural gas in China can not be used directly because of its high acid component(H2S and CO2).It must be treated by sweetening process and this process has the characteristics of low desulfurization selectivity and high energy consumption,which is not conducive to the sustainable development of petroleum enterprises.Therefore,it is an urgent problem to improve the selectivity of desulfurization and reduce energy consumption in the high sulfur gas sweetening process so as to improve the competitiveness and economic benefits of enterprises.Fortunately,it not only makes the distributed control system of industrial production field collect and store mass production data,but also provides technical support for the data driven industrial process model and multi-objective optimization with the rapid development of information technology and the maturity of big data mining technology.Therefore,it is the key issue to study the problem of data modeling and intelligent optimization of operation parameters for high sulfur gas sweetening process,including the following contents.(1)The Mi UKFNN algorithm is proposed to solve the problem of modeling efficiency and stability of high sulfur gas sweetening process.The Mi UKFNN uses the best nonlinear estimation property of UKF algorithm to overcome the influence of noise data on the accuracy of BPNN model.Moreover,the minimum sigma set algorithm is used instead of the symmetric sigma set algorithm to effectively reduce the number of sigma points in order to improve modeling efficiency and shorten modeling time.The Mi UKFNN algorithm is applied to industrial data and the results show that the Mi UKFNN algorithm not only greatly improves the operation efficiency of the model,but also has a higher model accuracy compared with the BPNN and UKFNN.Finally,model analysis is carried out by Leverage approach(LA)to prove the validity and reliability of the model.(2)The Sc Mi UKFNN algorithm is proposed,which not only improve the precision and generalization of the industrial model but also effectively guarantee the efficiency of modeling.Based on the Mi UKFNN algorithm,the Sc Mi UKFNN algorithm introduces Sc UT to replace the UT algorithm in order to improve the prediction accuracy of the mean and covariance of the state variable in the UKF algorithm.The results of simulation experiment and industrial application show that that the Sc Mi UKFNN model not only has the maximum R2 value and the smallest model error compared with the BPNN,UKFNN and Mi UKFNN algorithms,but also can effectively ensure the operation efficiency of the model.In a word,the Sc Mi UKFNN algorithm can provide high-precision,strong generalization and reliable model for multi-objective optimization of industrial process.(3)The NSGA-II-RVC algorithm is proposed to improve the convergence and diversity of multi-objective optimization algorithm and solve the industrial optimization problem with constraints.Firstly,the user preference vector,the selection strategy that violates the constraint function and the adaptive reference vector of the irregular PF are used in the framework of the NSGA-II algorithm to choose the elite individuals instead of the crowded sorting algorithm.Meanwhile,the angular distance penalty function is used to balance the convergence ability of the NSGA-II-RVC algorithm and maintain the ability of population diversity.Finally,it can obtain the new population by using genetic operator.The results of simulation experiment show that convergence of the NSGA-II-RVC algorithm and the diversity of the obtained solutions are all superior to the NSGA-II and NSGA-III.The NSGA-II-RVC algorithm is further applied to the industrial production process.The optimal operating parameters obtained by optimizing the Sc Mi UKFNN model not only improve the desulfurization selectivity by 0.58%,but also increase the production efficiency by 1.3%.