A method, device, storage medium and system for predicting the octane rating of reformed gasoline

Abstract

The application discloses a kind of reforming gasoline octane number prediction method, device, storage medium and system.The first lumped molecular volume concentration is obtained by carrying out single hydrocarbon analysis and classification mapping to gasoline training sample, and the octane number prediction optimization mixed model of mechanism model is included by the penalty term of first loss function, and the octane number of to be predicted gasoline sample is predicted, to combine the advantage of molecular composition model and data model, the prediction accuracy of octane number is improved, and the model extension is also improved.

Description

Technical Field

[0001] This invention relates to the field of reformed gasoline octane number prediction technology, and in particular to a method, apparatus, computer-readable storage medium and system for predicting reformed gasoline octane number. Background Technology

[0002] Catalytic reforming is a crucial process in modern oil refining and chemical industries. It plays a vital role in converting naphtha into high-octane gasoline blending components, petroleum aromatics (BTX), and producing inexpensive hydrogen as a byproduct. It serves as an important bridge connecting crude oil processing and chemical production, and is indispensable for modern integrated oil refining and chemical plants. It is a key technology of focus in the petrochemical industry, and an important indicator to consider when optimizing the operation of this unit is the octane number of the reformed gasoline.

[0003] In existing technologies, the common method for predicting the octane number of catalytic cracking gasoline is to use a data model: First, the production operation data of the unit is collected and preprocessed; then, feature selection is performed using an algorithm, and the feature subset is divided into a training set and a test set; finally, the parameters of the trained model are used to obtain the optimal octane number prediction model.

[0004] However, existing technologies still have the following drawbacks: overfitting can lead to problems such as inability to guarantee accuracy when predicting samples outside the training set (poor extension).

[0005] Therefore, there is a current need for a method, apparatus, computer-readable storage medium, and system for predicting the octane number of reformed gasoline, in order to overcome the aforementioned deficiencies in the prior art. Summary of the Invention

[0006] This invention provides a method, apparatus, computer-readable storage medium, and system for predicting the octane number of reformed gasoline, thereby improving the accuracy of octane number prediction.

[0007] An embodiment of the present invention provides a method for predicting the octane number of reformed gasoline. The prediction method includes: acquiring molecular data of a gasoline sample to be predicted, and obtaining the lumped molecular volume concentration of the gasoline sample to be predicted based on the molecular data; using the lumped molecular volume concentration as input to a preset octane number prediction optimization hybrid model to obtain the octane number of the gasoline sample to be predicted; the penalty term of the first loss function of the octane number prediction optimization hybrid model includes a mechanistic model.

[0008] As an improvement to the above scheme, before acquiring the molecular data of the gasoline sample to be predicted and obtaining the lumped molecular volume concentration of the gasoline sample to be predicted based on the molecular data, the prediction method further includes: acquiring training sample data and training an optimized hybrid model for octane number prediction based on the training sample data, a preset mechanistic model, and a preset hybrid model.

[0009] As an improvement to the above scheme, training sample data is obtained, and an octane number prediction optimization hybrid model is trained based on the training sample data, a preset mechanistic model, and a preset hybrid model. Specifically, this includes: obtaining training sample data of gasoline training samples; inputting the training sample data into a preset mechanistic model to calculate a first octane number; training a preset hybrid model based on the training sample data, the first octane number, and a preset optimization algorithm to obtain a first training hybrid model; and performing hyperparameter optimization on the first training hybrid model to obtain an octane number prediction optimization hybrid model.

[0010] As an improvement to the above scheme, a first training hybrid model is obtained by training a preset hybrid model based on the training sample data, the first octane number, and a preset optimization algorithm. Specifically, this includes: obtaining the first total molecular volume concentration from the training sample data; obtaining a first loss function based on the loss function of the mechanism model and the preset hybrid model, and optimizing the first loss function based on the preset Adam optimization algorithm to obtain an optimized hybrid model; and training the optimized hybrid model based on the first total molecular volume concentration and the first octane number to obtain the first training hybrid model.

[0011] As an improvement to the above scheme, the mechanism model is as follows: In the formula, i is the molecule index, v i Let β be the volume fraction of molecule i. i ON contributes parameters to the octane number of molecule i i I is the octane number of molecule i. P These are the interaction parameters between alkane molecules and other types of molecules.

[0012] As an improvement to the above scheme, the first loss function is: In the formula, Y is the actual measured octane number, and Y is the calculated value from the mixed model. phy These are values ​​calculated for the mechanistic model.

[0013] As an improvement to the above scheme, the hybrid model is: ON hdp =f(D, ON) phy ); where D is the volume concentration vector of 57 lumped molecules; ON phy These are values ​​calculated for the mechanistic model.

[0014] As an improvement to the above scheme, obtaining training sample data of gasoline training samples specifically includes: performing monomeric hydrocarbon analysis on gasoline training samples to obtain molecular data of gasoline training samples; performing classification mapping on the molecular data to obtain lumped molecules and corresponding lumped molecular volume concentrations; and storing the lumped molecular volume concentrations as training sample data of gasoline training samples.

[0015] Another embodiment of the present invention provides a reformed gasoline octane number prediction device. The prediction device includes a data acquisition unit and a model prediction unit. The data acquisition unit is used to acquire molecular data of the gasoline sample to be predicted and obtain the lumped molecular volume concentration of the gasoline sample to be predicted based on the molecular data. The model prediction unit is used to use the lumped molecular volume concentration as input to a preset octane number prediction optimization hybrid model to obtain the octane number of the gasoline sample to be predicted. The penalty term of the first loss function of the octane number prediction optimization hybrid model includes a mechanistic model.

[0016] As an improvement to the above scheme, the prediction device further includes a model training unit, which is used to acquire training sample data and train an optimized hybrid model for octane number prediction based on the training sample data, a preset mechanism model, and a preset hybrid model.

[0017] As an improvement to the above scheme, the model training unit is also used to acquire training sample data of gasoline training samples; input the training sample data into a preset mechanism model to calculate the first octane number; train a preset hybrid model based on the training sample data, the first octane number and a preset optimization algorithm to obtain a first training hybrid model; and perform hyperparameter optimization on the first training hybrid model to obtain an octane number prediction optimized hybrid model.

[0018] Another embodiment of the present invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the reformed gasoline octane number prediction method as described above.

[0019] Another embodiment of the present invention provides a reformed gasoline octane number prediction system, the prediction system including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the reformed gasoline octane number prediction method as described above.

[0020] Compared with existing technologies, this technical solution has the following beneficial effects:

[0021] This invention provides a method, apparatus, computer-readable storage medium, and system for predicting the octane number of reformed gasoline. By performing monomeric hydrocarbon analysis and classification mapping on gasoline training samples to obtain a first lumped molecular volume concentration, and by optimizing a hybrid model for octane number prediction using a penalty term in a first loss function that includes a mechanistic model, the octane number of the gasoline sample to be predicted is predicted. This combines the advantages of molecular composition models and data models, improving the accuracy of octane number prediction and enhancing the model's extrapolation. Attached Figure Description

[0022] Figure 1 This is a schematic flowchart of a method for predicting the octane number of reformed gasoline according to an embodiment of the present invention;

[0023] Figure 2 This is a schematic diagram of a reformed gasoline octane number prediction device provided in an embodiment of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Specific Implementation Example 1

[0026] The present invention first describes a method for predicting the octane number of reformed gasoline. Figure 1 This is a schematic flowchart of a method for predicting the octane number of reformed gasoline provided in an embodiment of the present invention.

[0027] like Figure 1 As shown, the prediction method includes:

[0028] S1: Obtain the molecular data of the gasoline sample to be predicted, and obtain the lumped molecular volume concentration of the gasoline sample to be predicted based on the molecular data.

[0029] After obtaining the gasoline sample to be predicted, the sample is subjected to monomer hydrocarbon analysis to obtain its specific molecular composition. The molecular data obtained from the analysis of the gasoline sample to be predicted is generally about 150+. Subsequently, the molecular composition results obtained from the analysis are classified and mapped into five categories (a total of 57 types) of lumped molecules: alkanes, cycloalkanes, aromatics, alkenes / cycloalkenes, and oxygen-containing compounds, and the lumped molecular volume concentration of each type of lumped molecule is calculated.

[0030] S2: Use the lumped molecular volume concentration as input to a preset octane number prediction optimization mixing model to obtain the octane number of the gasoline sample to be predicted.

[0031] In practical applications, a corresponding prediction model is constructed to predict the octane number of gasoline samples. By inputting the aforementioned lumped molecular volume concentration, the octane number of the gasoline sample to be predicted can be directly obtained.

[0032] In order to incorporate physical constraints as a penalty term in the loss function of the data model, thereby introducing relevant physical laws into the data modeling process and achieving combined prediction of molecular composition model and data model, thus improving prediction accuracy and model extension, this embodiment of the invention pre-constructs and trains a hybrid model for octane number prediction that combines molecular composition model and data model. In this embodiment of the invention, the penalty term of the first loss function of the hybrid model for octane number prediction includes the mechanism model.

[0033] Specifically, firstly, a pre-defined mechanistic model calculation formula is added as a penalty term to the loss function of a pre-defined hybrid model to obtain a first loss function. Then, the first loss function is optimized using a pre-defined Adam optimization algorithm to construct an optimized hybrid model that combines a molecular composition model and a data model. Next, the volume concentration (vol%) of the aforementioned 57 aggregated molecules is used as input to the pre-defined mechanistic model to calculate a first octane number. Then, the optimized hybrid model is trained based on the first aggregated molecule volume concentration and the first octane number to obtain a first training hybrid model. Finally, hyperparameter optimization is performed on the first training hybrid model to obtain an octane number prediction optimized hybrid model.

[0034] Specifically, hyperparameter optimization involves optimizing some parameters (number of network layers, number of nodes, etc.) in the first training hybrid model to obtain the optimal parameter combination, thereby obtaining the hybrid model with the best performance (described in this paper as "octane number prediction optimization hybrid model").

[0035] In one embodiment, before acquiring molecular data of the gasoline sample to be predicted and obtaining the lumped molecular volume concentration of the gasoline sample to be predicted based on the molecular data, the prediction method further includes: acquiring training sample data and training an optimized hybrid model for octane number prediction based on the training sample data, a preset mechanistic model, and a preset hybrid model.

[0036] In one embodiment, acquiring training sample data and training an optimized hybrid model for octane number prediction based on the training sample data, a preset mechanistic model, and a preset hybrid model specifically includes: acquiring training sample data of gasoline training samples; inputting the training sample data into a preset mechanistic model to calculate a first octane number; training a preset hybrid model based on the training sample data, the first octane number, and a preset optimization algorithm to obtain a first training hybrid model; and performing hyperparameter optimization on the first training hybrid model to obtain an optimized hybrid model for octane number prediction.

[0037] In one embodiment, obtaining training sample data for gasoline training samples specifically includes: performing monomeric hydrocarbon analysis on the gasoline training samples to obtain molecular data of the gasoline training samples; performing classification mapping on the molecular data to obtain lumped molecules and corresponding lumped molecular volume concentrations; and storing the lumped molecular volume concentrations as training sample data for the gasoline training samples.

[0038] In one embodiment, a preset hybrid model is trained based on the training sample data, the first octane number, and a preset optimization algorithm to obtain a first training hybrid model. Specifically, this includes: obtaining a first set of total molecular volume concentrations from the training sample data; obtaining a first loss function based on the loss function of the mechanistic model and the preset hybrid model, and optimizing the first loss function based on a preset Adam optimization algorithm to obtain an optimized hybrid model; and training the optimized hybrid model based on the first set of total molecular volume concentrations and the first octane number to obtain the first training hybrid model.

[0039] In one embodiment, the mechanism model is as follows:

[0040]

[0041] In the formula, i is the molecule index, v i Let β be the volume fraction of molecule i. i ON contributes parameters to the octane number of molecule i i I is the octane number of molecule i. P These are the interaction parameters between alkane molecules and other types of molecules.

[0042] In one embodiment, the first loss function is:

[0043]

[0044] In the formula, Y is the actual measured octane number, and Y is the calculated value from the mixed model. phy These are values ​​calculated for the mechanistic model.

[0045] In one embodiment, the hybrid model is:

[0046] ON hdp =f(D, ON) phy );

[0047] In the formula, D is the volume concentration vector of the 57 lumped molecules; ON phy These are values ​​calculated for the mechanistic model.

[0048] Compared to the data model, the octane number prediction optimization hybrid model constructed in this embodiment of the invention requires less training data. Furthermore, due to the introduction of physical constraints, the hybrid model exhibits better extensibility. Compared to the molecular composition model, it can extract more information from the data, improving prediction accuracy. In practical applications, the octane number prediction method based on a physics-guided neural network hybrid model described in this embodiment incorporates a mechanistic model into the input features for neural network training, while simultaneously adding mechanistic constraints to the loss function. This ensures that the trained model not only conforms to the experience and common sense of relevant industry experts but also utilizes data information to improve the model's prediction accuracy.

[0049] This invention describes a method for predicting the octane number of reformed gasoline. By performing monomer hydrocarbon analysis and classification mapping on gasoline training samples to obtain the first lumped molecular volume concentration, and by optimizing the hybrid model for octane number prediction using a penalty term of the first loss function that includes the mechanistic model, the octane number of the gasoline sample to be predicted is predicted. This method combines the advantages of molecular composition models and data models, thereby improving the accuracy of octane number prediction and enhancing the model's extrapolation. Specific Implementation Example 2

[0051] In addition to the methods described above, this invention also discloses a reformed gasoline octane number prediction device. Figure 2 This is a schematic diagram of a reformed gasoline octane number prediction device provided in an embodiment of the present invention.

[0052] like Figure 2 As shown, the prediction device includes a data acquisition unit 11 and a model prediction unit 12.

[0053] The data acquisition unit 11 is used to acquire molecular data of the gasoline sample to be predicted, and to obtain the lumped molecular volume concentration of the gasoline sample to be predicted based on the molecular data.

[0054] The model prediction unit 12 is used to take the lumped molecular volume concentration as the input of a preset octane number prediction optimization hybrid model to obtain the octane number of the gasoline sample to be predicted; the penalty term of the first loss function of the octane number prediction optimization hybrid model includes a mechanism model.

[0055] In one embodiment, the prediction device further includes a model training unit, which is used to acquire training sample data and train an optimized hybrid model for octane number prediction based on the training sample data, a preset mechanistic model, and a preset hybrid model.

[0056] In one embodiment, the model training unit is further configured to acquire training sample data of gasoline training samples; input the training sample data into a preset mechanistic model to calculate a first octane number; train a preset hybrid model based on the training sample data, the first octane number, and a preset optimization algorithm to obtain a first training hybrid model; and perform hyperparameter optimization on the first training hybrid model to obtain an octane number prediction optimized hybrid model.

[0057] In one embodiment, the model training unit is further configured to: perform monomeric hydrocarbon analysis on a gasoline training sample to obtain molecular data of the gasoline training sample; perform classification mapping on the molecular data to obtain lumped molecules and corresponding lumped molecular volume concentrations; and store the lumped molecular volume concentrations as training sample data of the gasoline training sample.

[0058] In one embodiment, the model training unit is further configured to: obtain the first set of total molecular volume concentration from the training sample data; obtain a first loss function based on the mechanism model and the loss function of the preset hybrid model, and optimize the first loss function according to the preset Adam optimization algorithm to obtain an optimized hybrid model; train the optimized hybrid model based on the first set of total molecular volume concentration and the first octane number to obtain a first training hybrid model.

[0059] If the unit integrated into the prediction device is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. Another embodiment of the present invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the reformed gasoline octane number prediction method as described above.

[0060] The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0061] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between units indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0062] This invention describes a reformed gasoline octane number prediction device and a computer-readable storage medium. By performing monomer hydrocarbon analysis and classification mapping on gasoline training samples to obtain the first lumped molecular volume concentration, and by optimizing the hybrid model for octane number prediction through the penalty term of the first loss function including the mechanistic model, the octane number of the gasoline sample to be predicted is predicted. This combines the advantages of molecular composition models and data models. The prediction device and computer-readable storage medium improve the accuracy of octane number prediction and also enhance the model's extrapolation. Specific Implementation Example 3

[0064] In addition to the methods and apparatus described above, embodiments of the present invention also describe a reformed gasoline octane number prediction system.

[0065] The prediction system includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the reformed gasoline octane number prediction method as described above.

[0066] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the device, connecting various parts of the device via various interfaces and lines.

[0067] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0068] This invention describes a reformed gasoline octane number prediction system. By performing monomer hydrocarbon analysis and classification mapping on gasoline training samples to obtain the first lumped molecular volume concentration, and by optimizing the hybrid model for octane number prediction using a penalty term of the first loss function that includes the mechanistic model, the system predicts the octane number of the gasoline sample to be predicted. This combines the advantages of molecular composition models and data models, thereby improving the accuracy of octane number prediction and enhancing the model's extrapolation.

[0069] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for predicting the octane number of reformed gasoline, characterized in that, The prediction method includes: Obtain molecular data of the gasoline sample to be predicted, and obtain the lumped molecular volume concentration of the gasoline sample to be predicted based on the molecular data. The lumped molecular volume concentration is used as the input to a preset octane number prediction optimization hybrid model to obtain the octane number of the gasoline sample to be predicted; the penalty term of the first loss function of the octane number prediction optimization hybrid model includes a mechanistic model; The mechanism model is as follows: ； In the formula, i For molecular indexing, v i For molecules i volume fraction, β i For molecules i Octane number contribution parameter, ON i For molecules i Octane number, I P These are the interaction parameters between alkane molecules and other types of molecules; The first loss function is: ； In the formula, Y is the actual measured octane number, and Y is the calculated value from the mixed model. Values ​​calculated for the mechanistic model; The hybrid model is as follows: ； In the formula, D This represents the volume concentration vector of 57 lumped molecules; These are values ​​calculated for the mechanistic model.

2. The method for predicting the octane number of reformed gasoline according to claim 1, characterized in that, Before acquiring the molecular data of the gasoline sample to be predicted and obtaining the lumped molecular volume concentration of the gasoline sample to be predicted based on the molecular data, the prediction method further includes: Acquire training sample data, and train an optimized hybrid model for octane number prediction based on the training sample data, a preset mechanism model, and a preset hybrid model.

3. The method for predicting the octane number of reformed gasoline according to claim 2, characterized in that, Acquire training sample data, and based on the training sample data, a preset mechanistic model, and a preset hybrid model, train an optimized hybrid model for octane number prediction, specifically including: Obtain training sample data for gasoline training samples; The training sample data is input into a preset mechanism model to calculate the first octane number; The preset hybrid model will be trained based on the training sample data, the first octane value, and the preset optimization algorithm to obtain the first training hybrid model. Hyperparameter optimization is performed on the first training hybrid model to obtain an optimized hybrid model for octane number prediction.

4. The method for predicting the octane number of reformed gasoline according to claim 3, characterized in that, The preset hybrid model will be trained based on the training sample data, the first octane number, and a preset optimization algorithm to obtain a first training hybrid model, specifically including: Obtain the total molecular volume concentration of the first set from the training sample data; Based on the aforementioned mechanism model and the loss function of the preset hybrid model, a first loss function is obtained, and the first loss function is optimized according to the preset Adam optimization algorithm to obtain an optimized hybrid model. The optimized hybrid model is trained based on the total molecular volume concentration of the first set and the first octane number to obtain the first training hybrid model.

5. A device for predicting the octane number of reformed gasoline, characterized in that, The method for predicting the octane number of reformed gasoline as described in any one of claims 1 to 4, wherein the prediction device includes a data acquisition unit and a model prediction unit, wherein, The data acquisition unit is used to acquire molecular data of the gasoline sample to be predicted, and to obtain the lumped molecular volume concentration of the gasoline sample to be predicted based on the molecular data. The model prediction unit is used to take the lumped molecular volume concentration as the input of a preset octane number prediction optimization hybrid model to obtain the octane number of the gasoline sample to be predicted; the penalty term of the first loss function of the octane number prediction optimization hybrid model includes a mechanistic model.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device on which the computer-readable storage medium is located to perform the reformed gasoline octane number prediction method as described in any one of claims 1 to 4.

7. A reformed gasoline octane number prediction system, characterized in that, The prediction system includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the reformed gasoline octane number prediction method as described in any one of claims 1 to 4.

Images