CN-121997691-A - Method, device, medium and equipment for predicting Reed vapor pressure of gasoline

CN121997691ACN 121997691 ACN121997691 ACN 121997691ACN-121997691-A

Abstract

The invention discloses a method, a device, a medium and equipment for predicting the Reid vapor pressure of gasoline. The method comprises the steps of obtaining a gasoline sample to be predicted, determining gasoline fingerprints of the gasoline sample, wherein the gasoline fingerprints comprise molecular components of the gasoline sample, inputting the gasoline fingerprints into a gasoline Reid vapor pressure prediction model to obtain the gasoline Reid vapor pressure of the gasoline sample to be predicted, training a neural network model by using training data, wherein the training data comprise the gasoline fingerprints of the gasoline sample of a preset type and the gasoline Reid vapor pressure corresponding to the gasoline fingerprints, and the gasoline Reid vapor pressure is obtained by processing the gasoline fingerprints of the gasoline sample of the preset type by using a gasoline Reid vapor pressure mechanism model. The method solves the technical problems of repeated training of the model and no universality, and simultaneously further improves the prediction accuracy of the Reed vapor pressure of the gasoline and the efficiency.

Inventors

CAI GUANGQING
LI CHUNPENG
XING DINGFENG
Xue Dao
GUAN JINGJUN
HU YIJIONG
WANG HONGLI

Assignees

中国石油天然气股份有限公司

Dates

Publication Date: 20260508
Application Date: 20241106

Claims (14)

1. A method for predicting the reed vapor pressure of gasoline, comprising: Obtaining a gasoline sample to be predicted, and determining a gasoline fingerprint of the gasoline sample, wherein the gasoline fingerprint comprises molecular components of the gasoline sample; The method comprises the steps of inputting gasoline fingerprints into a gasoline Reid vapor pressure prediction model to obtain the gasoline Reid vapor pressure of a gasoline sample to be predicted, training a neural network model by using training data to obtain the gasoline Reid vapor pressure prediction model, obtaining the training data by using an established gasoline Reid vapor pressure mechanism model, wherein the training data comprise gasoline fingerprints of a preset type of gasoline sample and the gasoline Reid vapor pressure corresponding to the gasoline fingerprints, and the gasoline Reid vapor pressure mechanism model is an experimental detection process for simulating the gasoline Reid vapor pressure and is a calculation model obtained by using a gas-liquid phase balance principle in the experimental detection process.
2. The method of claim 1, obtaining a predictive model of the reed vapor pressure of gasoline by; Acquiring training data by using an established gasoline Reid vapor pressure mechanism model; Inputting the training set into a neural network model, and adjusting the super parameters of the neural network model according to the loss function value of the neural network model, wherein the loss function value is not more than a preset value, so as to obtain a trained neural network model; Inputting the test set into the trained neural network model, comparing the output result with the Reid vapor pressure of the test set data to obtain a prediction rate, and if the prediction rate is smaller than the expected prediction rate, continuing training the trained neural network model until the prediction rate is not smaller than the expected prediction rate, and obtaining a prediction model of the Reid vapor pressure of the gasoline.
3. The method of claim 1, wherein obtaining training data using an established model of a reed vapor pressure mechanism of gasoline comprises: Generating constraint conditions of virtual oil products according to molecular component data of known types of gasoline samples, and randomly generating preset types of gasoline samples according to the constraint conditions; and processing the gasoline fingerprint of the preset type gasoline sample by using the gasoline Reid vapor pressure mechanism model to obtain the gasoline Reid vapor pressure of the preset type gasoline sample, wherein the gasoline fingerprint of the preset type gasoline sample and the corresponding gasoline Reid vapor pressure form training data.
4. A method according to claim 3, wherein generating constraints from molecular composition data of known types of gasoline samples, and randomly generating preset types of gasoline samples from the constraints, comprises: Obtaining the molecular types of the molecular components and the ranges of the molecular weights of various types according to the molecular types and the molecular weights of various types in the molecular components of the known type gasoline sample; And randomly selecting molecular types and molecular mass fractions of all types within the molecular types and molecular mass ranges of all types of molecular components, and carrying out normalization treatment to obtain a gasoline sample of a preset type.
5. The method of claim 4, wherein the gasoline fingerprint of the gasoline sample of the preset type is processed by using a gasoline Reid vapor pressure mechanism model to obtain the gasoline Reid vapor pressure of the gasoline sample of the preset type, and the specific process is as follows; Determining initial calculation conditions of a flash evaporation model according to a Reid vapor pressure test experiment of the gasoline of a Reid vapor pressure mechanism model of the gasoline aiming at each preset type of gasoline sample, wherein the calculation conditions comprise an operation temperature, a first gasification fraction and an operation pressure; obtaining a gasification fraction difference value according to the first gasification fraction and the second gasification fraction; If the gasification fraction difference value is smaller than a preset precision threshold value, taking the current operating pressure as the Reid vapor pressure of the gasoline sample; And if the gasification fraction difference value is not smaller than the preset precision threshold value, taking the value of the second gasification fraction as the value of the first gasification fraction, and re-executing the step of obtaining the second gasification fraction and the current operating pressure until the gasification fraction difference value is smaller than the preset precision threshold value, so as to obtain the Reed vapor pressure of the gasoline sample, and further obtain the Reed vapor pressure of the gasoline sample of the preset type.
6. The method of claim 5, wherein the processing of the molecular components of the gasoline sample under the calculated conditions using a flash model that yields a second gasification fraction, the current operating pressure, at a flash equilibrium state comprises: under the calculation condition, treating the molecular components of the gasoline sample by utilizing a flash evaporation model to obtain the gas phase content and the liquid phase content of the gasoline sample; Comparing the gas phase content with the liquid phase content to obtain a gas-liquid difference value, and comparing the gas-liquid difference value with a preset balance threshold value to obtain a state of a flash evaporation model; if the state of the flash evaporation model is a flash evaporation equilibrium state, calculating the molecular components of the gasoline sample by the flash evaporation model to obtain a second gasification fraction; If the state of the flash model is the non-flash equilibrium state, adjusting the operation pressure value by using a preset algorithm, continuously executing the step of judging whether the flash equilibrium state is reached or not until the state of the flash model is the flash equilibrium state, and calculating to obtain the second gasification fraction.
7. The method of claim 6, wherein the processing of the molecular components of the gasoline sample under the calculated conditions using a flash model to obtain the gas phase content and the liquid phase content of the gasoline sample comprises: aiming at each molecular component of the gasoline sample, treating one molecular component of the gasoline sample by utilizing a flash evaporation model under the calculation condition to obtain a gas-phase fugacity coefficient and a liquid-phase fugacity coefficient of one molecular component of the gasoline sample; obtaining a phase equilibrium constant of a molecular component of the gasoline sample according to the gas phase fugacity coefficient and the liquid phase fugacity coefficient; And calculating the gas phase content and the liquid phase content of one molecular component of the gasoline sample according to the phase equilibrium constant, the first gasification fraction and the one molecular component, so as to obtain the gas phase content and the liquid phase content of the gasoline sample.
8. The method of claim 7, wherein comparing the gas phase content to the liquid phase content yields a gas-liquid difference value, and comparing the gas-liquid difference value to a preset equilibrium threshold yields a state of a flash model, comprising: comparing the gas phase content with the liquid phase content to obtain a gas-liquid difference; And comparing the gas-liquid difference value with a preset balance threshold value, if the gas-liquid difference value is smaller than the preset balance threshold value, obtaining the state of the flash model as a flash evaporation balance state, and if the gas-liquid difference value is not smaller than the preset balance threshold value, obtaining the state of the flash model as a non-flash evaporation balance state.
9. The method of claim 8, wherein calculating the second gasification fraction for the molecular component of the gasoline sample by the flash model if the state of the flash model is a flash equilibrium state comprises: if the state of the flash model is a flash equilibrium state, calculating the molecular components of the gasoline sample by utilizing a Soxhlet- kuang equation to obtain the density of a mixed gas phase, calculating the molecular components of the gasoline sample by utilizing a Rake equation to obtain the density of a mixed liquid phase, and obtaining a second gasification fraction according to the density of the mixed gas phase and the density of the mixed liquid phase.
10. The method of claim 6, wherein adjusting the operating pressure value using a preset algorithm comprises: and adjusting the operation pressure value by utilizing a Newton iteration method.
11. The method of claim 2, wherein adjusting the hyper-parameters of the neural network model based on the loss function values of the neural network model comprises: And adjusting at least one of the number of layers of the hidden layer, the number of nodes of the hidden layer, the learning rate and the iteration number of the neural network according to the loss function value of the neural network model.
12. A vapor pressure predicting apparatus for gasoline, comprising: The data acquisition module is used for acquiring a gasoline sample to be predicted, and determining a gasoline fingerprint of the gasoline sample, wherein the gasoline fingerprint comprises molecular components of the gasoline sample; The prediction module is used for inputting the gasoline fingerprint into a gasoline Reid vapor pressure prediction model to obtain the gasoline Reid vapor pressure of the gasoline sample to be predicted, the gasoline Reid vapor pressure prediction model is obtained by training a neural network model by using training data, the training data are obtained by using an established gasoline Reid vapor pressure mechanism model, the training data comprise gasoline fingerprints of the preset type of gasoline sample and the gasoline Reid vapor pressure corresponding to the gasoline fingerprints, the gasoline Reid vapor pressure mechanism model is an experimental detection process for simulating the gasoline Reid vapor pressure, and a calculation model is obtained by using a gas-liquid phase balance principle in the experimental detection process.
13. A computer storage medium having stored therein computer executable instructions which when executed by a processor implement the method of predicting the reed vapor pressure of gasoline as claimed in any one of claims 1 to 11.
14. The terminal equipment for predicting the Reed vapor pressure of the gasoline is characterized by comprising a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes the method for predicting the Reed vapor pressure of the gasoline according to any one of claims 1-11 when executing the program.

Description

Method, device, medium and equipment for predicting Reed vapor pressure of gasoline Technical Field The invention relates to the technical field of crude oil processing, in particular to a method, a device, a medium and equipment for predicting the Reed vapor pressure of gasoline. Background The Reid vapor pressure (Reid Vapor Pressure, RVP) of gasoline is an important index of the evaporation performance of gasoline, and can reflect the air resistance trend of the oil product in the engine during the starting process of the engine and the running process of the engine at high temperature or high latitude. When the Reid vapor pressure of the gasoline is too low, the fact that the oil is difficult to gasify can lead to difficult starting of the automobile, uneven mixing of the oil and the air and incomplete combustion, and increase of pollutants such as discharged CO, hydrocarbons and the like. When the Reid vapor pressure of the gasoline is too high, the gasoline evaporation performance is too strong, the gasoline is easy to gasify, so that the evaporation emission of the gasoline is increased, namely the VOC (Volatile Organic Compounds ) emission is increased, and meanwhile, air resistance is formed in a pipeline to interrupt oil supply. In the related technology for predicting the Reid vapor pressure of gasoline, the IR spectrum, the gas chromatograph, the distillation curve and the Reid Vapor Pressure (RVP) of various gasoline samples are related by a correlation method to obtain a prediction model of the Reid vapor pressure of the gasoline, and the prediction model of the Reid vapor pressure of the gasoline is used for prediction to obtain the Reid vapor pressure of the oil to be predicted. Disclosure of Invention And predicting the Reid vapor pressure of the oil product by using a prediction model of the Reid vapor pressure of the gasoline obtained by the correlation method. The method is characterized in that a prediction model of the gasoline Reid vapor pressure is obtained by using a correlation method, the method is realized based on an established database for storing infrared spectra, gas chromatograms, distillation curves and the gasoline Reid Vapor Pressure (RVP) of various gasoline samples, and the database comprising the infrared spectra, the gas chromatograms, the distillation curves and the gasoline Reid Vapor Pressure (RVP) of the gasoline samples needs to be reconstructed for the gasoline samples which are not stored in the database, so that the obtained prediction model of the gasoline Reid vapor pressure has no universality, has higher maintenance cost and has the problems of low efficiency and inaccuracy of predicting the gasoline Reid vapor pressure of an oil product. In view of the foregoing, the present invention has been developed to provide a method, apparatus, medium, and device for predicting the reed vapor pressure of gasoline that overcome, or at least partially solve, the foregoing problems. The embodiment of the invention provides a method for predicting the Reid vapor pressure of gasoline, which comprises the following steps: Obtaining a gasoline sample to be predicted, and determining a gasoline fingerprint of the gasoline sample, wherein the gasoline fingerprint comprises molecular components of the gasoline sample; The method comprises the steps of inputting gasoline fingerprints into a gasoline Reid vapor pressure prediction model to obtain the gasoline Reid vapor pressure of a gasoline sample to be predicted, training a neural network model by using training data to obtain the gasoline Reid vapor pressure prediction model, obtaining the training data by using an established gasoline Reid vapor pressure mechanism model, wherein the training data comprise gasoline fingerprints of a preset type of gasoline sample and the gasoline Reid vapor pressure corresponding to the gasoline fingerprints, and the gasoline Reid vapor pressure mechanism model is an experimental detection process for simulating the gasoline Reid vapor pressure and is a calculation model obtained by using a gas-liquid phase balance principle in the experimental detection process. In a further alternative embodiment, the predictive model of the reed vapor pressure of the gasoline is obtained by; Acquiring training data by using an established gasoline Reid vapor pressure mechanism model; Inputting the training set into a neural network model, and adjusting super parameters of the neural network model according to the loss function value of the neural network model, wherein the loss function value is not more than a preset value, so as to obtain a trained neural network model; Inputting the test set into the trained neural network model, comparing the output result with the Reid vapor pressure of the test set data to obtain a prediction rate, and if the prediction rate is smaller than the expected prediction rate, continuing training the trained neural network model until the prediction rate is not smaller