Study And Realization Of MDEA-PZ-H₂O Software Model Based On Multi-kernel SVR

In liquefaction process of natural gas,the containing of CO₂ in raw gas will cause corrosion of equipment and pipeline,which will affect the quality of natural gas badly.The MDEA-PZ-H₂O decarburization technology not only owns the characteristics of high solubility in CO₂ and low reaction heat,but also has several advantages,such as low regeneration energy cost,good stability,no danger of degradation,low volatility and no corrosion effect on carbon steel equipment.Therefore,it has been widely used in liquefaction process of natural gas.As an important algorithm for machine learning theory,Support Vector Regression?SVR?has been widely used in many fields.Instead of adhering to the search for an optimal single kernel implementation regression algorithm.The introduction of multi-kernel function makes SVR try to find a new fusion kernel that is closer to the characteristics of the sample data set,which can improve the accuracy and generalization of the model.In this paper,the multi-kernel SVR theory is introduced from the perspective of the MDEA-PZ-H₂O decarburization technology model,which has a certain guiding significance for the optimization of process parameters.Firstly,based on the basic principles of chemical thermodynamic and dynamic theories,analyzing the characteristics and application scope of the CO₂ solubility in MDEA-PZ-H₂O model.Combining with the project of natural gas liquefaction project,the paper proposed the thermodynamic model of CO₂ solubility in MDEA-PZ-H₂O.Secondly,combining with the mainly used kernel function and SVR theory,four kinds of decarburization process prediction model based on mononuclear SVR are established.Then,studying the multi-kernel function theory and establish an MDEA-PZ-H₂O Software Model Based on multi-kernel SVR.Using the actual production data and creating a simulation environment to experiment to shows that the multi-kernel SVR model is more accurate than the single core SVR model,and the model can meet the requirements of industrial production.Finally,the MDEA-PZ-H₂O decarbonization process model software system based on multi-kernel SVR was developed by Object oriented language,to optimize the process parameters and provide scientific guidance and theoretical basis for determining the parameters of natural gas decarburization process.