A method, system, and storage medium for predicting hydrocracking products.

Abstract

This invention provides a method, system, and storage medium for predicting hydrocracking products. The method for predicting hydrocracking products includes the following steps: acquiring sample data of various feedstock properties during the hydrocracking process, and generating molecular aggregate composition data of hydrocracking products using a process mechanism model as the actual output value; constructing a product prediction model; constructing a weight determination model for each weight based on a mini-batch gradient descent algorithm; expanding the sample data according to the data format of the polynomial linear regression equation, and performing adaptive step-size training on the weight determination model to determine the weights of each power, and obtaining a trained product prediction model; and inputting the various feedstock property data of the sample to be tested into the trained product prediction model to predict the molecular aggregate composition of the corresponding product.

Description

Technical Field

[0001] This invention relates to the field of petrochemical technology, and in particular to a method for predicting hydrocracking products, a system for predicting hydrocracking products, and a computer-readable storage medium. Background Technology

[0002] In the field of petrochemical technology, hydrocracking is a common process used to break down large molecules in hydrocracking feedstocks into hydrocarbons with lower molecular weight and lower boiling points. Due to the complexity of petroleum molecular composition, those skilled in the art generally use step-size adaptive machine learning methods to construct computational models for hydrocracking processes to predict the aggregate molecular composition of feedstock oils.

[0003] Currently, commonly used adaptive algorithms include the momentum method and the RMSprop algorithm. The momentum method, based on gradient descent, adds a variable to accumulate and sum the gradients of each iteration, while simultaneously using a decay rate to decay the accumulated sum. However, when the step size is large, this method may miss the optimal solution or even lead to oscillations and divergence. The RMSprop algorithm uses the root mean square to accumulate the gradient in the denominator, resulting in a smaller step size, which may also lead to excessively slow iteration speeds and severely reduce computational efficiency.

[0004] In order to overcome the above-mentioned defects in the existing technology, there is an urgent need in the field for a prediction technology of hydrocracking products, which can adaptively adjust the step size of the iteration to quickly and accurately obtain the molecular aggregate composition of the hydrocracking reaction. Summary of the Invention

[0005] The following provides a brief overview of one or more aspects to offer a basic understanding of them. This overview is not an exhaustive summary of all conceived aspects, nor is it intended to identify key or decisive elements of all aspects, nor to define the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed descriptions that follow.

[0006] To overcome the aforementioned deficiencies in the prior art, the present invention provides a method for predicting hydrocracking products, a system for predicting hydrocracking products, and a computer-readable storage medium for adaptively adjusting the step size of the iteration to quickly and accurately obtain the molecular aggregate composition of the hydrocracking reaction.

[0007] Specifically, the method for predicting hydrocracking products according to the first aspect of the present invention includes the following steps: obtaining sample data of various feedstock properties during the hydrocracking process, and generating molecular aggregate composition data of the hydrocracking products through a process mechanism model as the actual output value; constructing a polynomial linear regression equation. The product prediction model, where y is the output data of the product prediction model, and x... i x j These are the input data for the product prediction model, where n is the dimension of the input data, and θ0, θ... i θ ij These are the weights of each power in the product prediction model; based on the mini-batch gradient descent algorithm, a weight determination model for each of these weights is constructed:

[0008]

[0009] Where α is the step size of each iteration, m represents the number of samples selected in each iteration, and x k For the input value of the k-th sample, Let y be the predicted output value for the k-th sample. (k) The true output value of the k-th sample is obtained; the sample data is expanded according to the data format of the polynomial linear regression equation, and the weight determination model is trained with adaptive step size to determine the weight of each power and obtain the product prediction model after training; and the multiple raw material oil attribute data of the sample to be tested are input into the product prediction model after training to predict the molecular aggregate composition of the corresponding product.

[0010] Optionally, in some embodiments of the present invention, the feedstock oil property data includes at least one of density, boiling point, sulfur content, and nitrogen content.

[0011] Furthermore, in some embodiments of the present invention, the step of expanding the sample data according to the data format of the polynomial linear regression equation includes: preprocessing an original sample set X0 constructed from the original sample data of multiple training samples to remove outliers and missing values; normalizing the original sample data in the preprocessed original sample set X0 to obtain a first sample set X1 that conforms to a standard normal distribution; and expanding the first sample set X1 by introducing cross terms to obtain a second sample set X2 that conforms to the input data format of the polynomial linear regression equation.

[0012] Furthermore, in some embodiments of the present invention, the step of normalizing the original sample data in the preprocessed original sample set X0 includes:

[0013] X=(X0-μ) / σ

[0014] Where X0 is the input original sample data, X is the normalized first sample data, μ is the mean, and σ is the variance.

[0015] Furthermore, in some embodiments of the present invention, the matrix form of the first sample set X1 is represented as follows:

[0016]

[0017] Among them, X i The first sample data is from the first sample set X1. The step of expanding the first sample set X1 by introducing cross terms to obtain a second sample set X2 that conforms to the input data format of the polynomial linear regression equation includes: multiplying each of the first sample data in the first sample set X1 (excluding the first feature value) pairwise to obtain the expanded second sample set X2.

[0018] Where, X′ i This refers to the second sample data in the second sample set X2.

[0019] Furthermore, in some embodiments of the present invention, the matrix form of the polynomial linear regression equation is expressed as:

[0020]

[0021]

[0022]

[0023] Where p is the second sample data X′ in the second sample set X2. i The number of samples, where n is the number of samples in the first sample set X1.

[0024] Further, in some embodiments of the present invention, the step of training the weight determination model with adaptive step size includes: dividing the expanded second sample dataset X2 into a training set and a test set; sequentially inputting the sample data of each training sample in the training set into the weight determination model to be trained, and determining the corresponding predicted output value according to a preset step size and number of iterations; in response to the loss function value between the predicted output value and the corresponding true output value being greater than a preset threshold, increasing the step size or the number of iterations, so as to determine the corresponding predicted output value again according to the increased step size and / or the increased number of iterations; and in response to the loss function value being less than or equal to the threshold, determining that the training of the weight determination model is complete.

[0025] Furthermore, in some embodiments of the present invention, the step of determining the corresponding predicted output value based on a preset step size and number of iterations includes: determining the weight of each power in the product prediction model according to the following expression:

[0026]

[0027]

[0028] r=ρr+(1-ρ)g 2

[0029]

[0030] g′=μg′+(1-μ)g

[0031] Where, θ j Let be the weights raised to the power of j, α be the step size, and δ be a constant used to stabilize the values. Here, ρ is the corrected cumulative squared gradient, r is the cumulative squared gradient, g′ is the cumulative gradient in each iteration, μ is the momentum coefficient, and t is the current iteration number. The corrected cumulative gradient; and the weights of each of the powers are substituted into the product prediction model to determine the predicted output value corresponding to each of the sample data.

[0032] Furthermore, the hydrocracking product prediction system provided according to a second aspect of the present invention includes: a memory storing computer instructions thereon; and a processor connected to the memory and configured to execute the computer instructions stored in the memory to implement the hydrocracking product prediction method as described in the first aspect of the present invention.

[0033] Furthermore, according to a second aspect of the present invention, a computer-readable storage medium is provided thereon storing computer instructions. When the computer instructions are executed by a processor, the method for predicting hydrocracking products as described in the first aspect of the present invention is implemented. Attached Figure Description

[0034] The above-described features and advantages of the present invention will be better understood after reading the following detailed description of embodiments of the present disclosure in conjunction with the accompanying drawings. In the drawings, components are not necessarily drawn to scale, and components having similar related characteristics or features may have the same or similar reference numerals.

[0035] Figure 1 A schematic flowchart of a method for predicting hydrocracking products according to some embodiments of the present invention is shown.

[0036] Figure 2 A schematic diagram of the loss function of a hydrocracking model provided according to some embodiments of the present invention is shown. Detailed Implementation

[0037] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Although the description of the present invention is presented in conjunction with preferred embodiments, this does not mean that the features of the invention are limited to these embodiments. On the contrary, the purpose of describing the invention in conjunction with embodiments is to cover other options or modifications that may be derived based on the claims of the present invention. To provide a thorough understanding of the invention, many specific details will be included in the following description. The invention may also be implemented without using these details. Furthermore, to avoid confusion or obscuring the focus of the invention, some specific details will be omitted in the description.

[0038] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0039] Furthermore, the terms "upper," "lower," "left," "right," "top," "bottom," "horizontal," and "vertical" used in the following description should be understood as the orientations shown in the relevant paragraphs and accompanying drawings. These relative terms are for illustrative purposes only and do not imply that the described apparatus must be manufactured or operated in a specific orientation, and therefore should not be construed as limiting the invention.

[0040] It is understood that although terms such as "first," "second," and "third" may be used herein to describe various components, regions, layers, and / or parts, these components, regions, layers, and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers, and / or parts. Therefore, the first components, regions, layers, and / or parts discussed below may be referred to as second components, regions, layers, and / or parts without departing from some embodiments of the present invention.

[0041] Currently, commonly used adaptive algorithms include the momentum method and the RMSprop algorithm. The momentum method, based on gradient descent, adds a variable to accumulate and sum the gradients of each iteration, while simultaneously using a decay rate to attenuate the accumulated sum. However, when the step size is large, this method may miss the optimal solution or even lead to oscillations and divergence. The RMSprop algorithm uses the root mean square (RMS) to accumulate the gradient in the denominator, resulting in a smaller step size, which may also lead to excessively slow iteration speeds and severely reduced computational efficiency.

[0042] To overcome the aforementioned deficiencies in the prior art, the present invention provides a method for predicting hydrocracking products, a system for predicting hydrocracking products, and a computer-readable storage medium for adaptively adjusting the step size of the iteration to quickly and accurately obtain the molecular aggregate composition of the hydrocracking reaction.

[0043] In some non-limiting embodiments, the method for predicting hydrocracking products provided in the first aspect of the present invention can be implemented based on the hydrocracking product prediction system provided in the second aspect of the present invention. Specifically, the hydrocracking product prediction system is equipped with a memory and a processor. The memory includes, but is not limited to, the computer-readable storage medium provided in the third aspect of the present invention, on which computer instructions are stored. The processor is connected to the memory and configured to execute the computer instructions stored in the memory to implement the method for predicting hydrocracking products provided in the first aspect of the present invention.

[0044] The following will describe the working principle of the above-mentioned hydrocracking product prediction system with reference to some embodiments of the prediction methods for hydrocracking products. Those skilled in the art will understand that these embodiments of the hydrocracking product prediction methods are merely non-limiting implementations provided by the present invention, intended to clearly demonstrate the main concepts of the invention and provide specific solutions convenient for public implementation, rather than limiting all functions or all operating methods of the hydrocracking product prediction system. Similarly, these hydrocracking product prediction methods are also merely non-limiting implementations provided by the present invention, and do not constitute a limitation on the subject or order of execution of each step in these hydrocracking product prediction methods.

[0045] Specifically, due to the complexity of petroleum molecules, all refining processes contain thousands of molecules in their reaction systems, including hydrocracking. Under high temperature, high pressure, and hydrogen-rich operating conditions, the large molecules in the hydrocracking feedstock are broken down into hydrocarbons with lower molecular weight and lower boiling points, facilitated by a catalyst. The catalyst used in hydrocracking is a bifunctional catalyst, and the process consists of two steps. The first step is hydrotreating, where the active metal component in the catalyst removes metals, sulfur, nitrogen compounds, and some aromatics from the feedstock. The second step is hydrocracking, where the acidic support in the catalyst converts the heavy fractions from the previous step into lighter fractions. The main reaction types include hydrocracking, hydroring-opening, hydrodealkylation, and hydroisomerization. These reactions are highly interconnected, forming a complex reaction network.

[0046] Generally, the feedstock for hydrocracking consists of various hydrocarbons, which can be classified into four main groups based on their structure: alkanes (P), alkenes (O), cycloalkanes (N), and aromatics (A), collectively known as PONA. In addition, there are sulfur- and nitrogen-containing compounds. During the hydrocracking reaction, as various reactions occur, heavier hydrocarbons or non-hydrocarbon compounds decompose into substances with smaller molecular weights and lower distillation ranges. Therefore, this study considers refining the output based on each hydrocarbon group and non-hydrocarbon compound, using the lumped molecular composition as the output of the computational model.

[0047] Please refer to Figure 1 , Figure 1 A schematic flowchart of a method for predicting hydrocracking products according to some embodiments of the present invention is shown.

[0048] like Figure 1 As shown, the method for predicting hydrocracking products can first perform step S1: obtaining the hydrocracking dataset.

[0049] Specifically, this hydrocracking dataset may include sample data on various feedstock properties during the hydrocracking process, as well as the lumped molecular composition data of the corresponding hydrocracking products. Here, the lumped molecular composition data can be generated by process mechanism model simulation as the actual output value. The feedstock property data may include at least one of the following: feedstock density, boiling point, sulfur content, and nitrogen content.

[0050] Subsequently, the method for predicting hydrocracking products can perform step S2: preprocessing the collected sample data. This preprocessing includes preprocessing the original sample set X0 constructed from the original sample data of multiple training samples to remove outliers and missing values.

[0051] Subsequently, the method for predicting hydrocracking products can perform step S3: normalizing the preprocessed data.

[0052] Specifically, the prediction method for hydrocracking products normalizes the raw sample data in the pre-processed raw sample set X0 to obtain a first sample set X1 that conforms to a standard normal distribution, i.e.:

[0053] X=(X0-μ) / σ

[0054] Where X0 is the input original sample data, X is the normalized first sample data, μ is the mean, and σ is the variance.

[0055] The matrix form of the first sample set X1 is represented as:

[0056]

[0057] Among them, Xi This refers to the first sample data in the first sample set X1.

[0058] Furthermore, the training set after normalization is as follows:

[0059]

[0060] In this dataset, the first dataset X1 represents a set of input data. During data processing, the first column of the dataset is made entirely of 1s, effectively introducing a constant term into the model so that the model can learn the intercept during the fitting process. Clearly, the normalized data does not meet the requirements for input data in a multinomial regression equation.

[0061] Subsequently, the method for predicting hydrocracking products can perform step S4: eigenvalue expansion.

[0062] Specifically, because the training model, which uses the original input data from hydrocracking as training samples, has a poor fit between its calculation results and the original sample output, the normalized dataset does not satisfy the input of the multinomial regression equation.

[0063] To better capture the nonlinear relationships between independent variables, the sample data can be augmented according to the data format of a polynomial linear regression equation. This prediction method requires cross-multiplication of eigenvalues ​​to obtain cross terms, thereby augmenting dataset X1 to obtain a second sample set X2, ensuring that its input data format conforms to the following polynomial linear regression equation. The product prediction model represents the output data of the product prediction model, where y is the output data of the product prediction model, and x is the output data of the product prediction model. i x j This is the input data for the product prediction model, where n is the dimension of the input data, and θ0, θ... i θ ij These are the weights of each power in the product prediction model. In multinomial linear regression, the interaction term refers to the new feature obtained by multiplying different independent variables.

[0064] The steps for augmenting the first dataset X1 include multiplying each feature except the first feature in pairs to obtain the augmented second sample set X2:

[0065] Where, X′ i This refers to the second sample data in the second sample set X2.

[0066] Furthermore, this polynomial regression is expressed in matrix form as follows:

[0067]

[0068]

[0069]

[0070] Where p is the second sample data X′ in the second sample set X2. i The number of samples, where n is the number of samples in the first sample set X1.

[0071] Subsequently, the method for predicting hydrocracking products can perform step S5: constructing a hydrocracking prediction model with a fixed step size based on the mini-batch gradient descent algorithm.

[0072] Specifically, a weight determination model is constructed for each of the aforementioned weights. A small subset of data is selected for each calculation to improve training efficiency. The weight determination expression for this model is:

[0073]

[0074] Where α is the step size of each iteration, m represents the number of samples selected in each iteration, and x k For the input value of the k-th sample, Let y be the predicted output value for the k-th sample. (k) Let m be the true output value of the k-th sample, where k ≤ m.

[0075] After implementing a hydrocracking prediction model with a fixed step size, in order to further optimize it, the prediction method for hydrocracking products can perform step S6: using the Adam adaptive algorithm to train the hydrocracking prediction model with an adaptive step size.

[0076] Specifically, the method for predicting hydrocracking products can train the weight determination model with adaptive step size to determine the weights of each power and obtain a fully trained product prediction model.

[0077] First, the expanded second sample dataset X2 is divided into a training set and a test set according to a certain ratio. Currently, when the amount of data is large, the ratio used is training set: test set = 0.7: 0.3.

[0078] Furthermore, the sample data of each training sample in the training set are sequentially input into the weights to be trained to determine the model, and the corresponding predicted output value is determined according to the preset step size and number of iterations. Here, this method for predicting hydrocracking products can substitute the weights of each power into the product prediction model to determine the predicted output value corresponding to each sample data. The expression for the weights of each power is:

[0079]

[0080]

[0081] r=ρr+(1-ρ)g 2

[0082]

[0083] g′=μg′+(1-μ)g

[0084] Where, θ j Let be the weights raised to the power of j, α be the step size (i.e., the learning rate of the model), and δ be a small constant used to stabilize the values. Let ρ be the corrected cumulative squared gradient, ρ be the decay rate (typically taken as 0.999), r be the cumulative squared gradient, g′ be the cumulative gradient in each iteration (initially 0), μ be the momentum coefficient (typically taken as 0.9), and t be the current iteration number. This is the corrected cumulative gradient.

[0085] In the early stages of iteration, the initialization of the cumulative squared gradient r approaches 0, while the corrected cumulative gradient... The initial value approaches g, so the Adam algorithm will generate a larger step size, which speeds up the convergence and quickly approaches the optimal solution. At the end of the iteration, since the gradient change is small, the cumulative gradient after correction increases with the number of iterations. At this time, the Adam algorithm will generate a smaller step size to refine the parameter adjustment with a smaller step size.

[0086] Furthermore, the hydrocracking model can construct a loss function based on the mean squared error between the predicted and actual values ​​of each sample. In order to evaluate the model's effectiveness, among which, For the product prediction model x i The corresponding output value, y i For x i The corresponding true output value, where m is the number of samples selected in each iteration. If the loss function value is less than or equal to a threshold, the training of the weight-determining model is considered complete. Conversely, if the loss function value between the predicted output value and the corresponding true output value is greater than a preset threshold, the initial step size α (i.e., the model's learning rate) or the number of iterations is increased to determine the corresponding predicted output value again based on the increased step size and / or increased number of iterations.

[0087] In other words, the closer the loss function is to 0, the smaller the average difference between the model's predicted and actual values, and the better the prediction result. In this case, the model can be used as a prediction model for hydrocracking. If the loss function does not meet the expected requirements, the number of iterations can be increased to obtain better results, or the step size can be increased to accelerate the rate at which the loss function decreases.

[0088] Please refer to Figure 2 and Table 1, Figure 2 A schematic diagram of the loss function of a hydrocracking model provided according to some embodiments of the present invention is shown. Table 1 shows a comparison between the predicted output values ​​and the actual output values ​​of the hydrocracking model.

[0089] Table 1 Comparison of predicted and actual output values ​​of the hydrocracking model.

[0090]

[0091] like Figure 2 As shown in Table 1, after 10,000 iterations, the model's loss function is very close to 0, indicating that the average difference between the model's predicted and actual values ​​is very small, the prediction results are good, and the convergence speed is fast. The data model obtained after 40,000 iterations can be used as a prediction model for hydrocracking. Table 1 shows that the output of the prediction model is almost identical to the output of the actual mechanism model, further confirming the conclusion that this data model can be used as a prediction model for hydrocracking.

[0092] Finally, after training the prediction model, the prediction method can perform step S7: make a prediction using the hydrogenation prediction model.

[0093] Specifically, the method for predicting hydrocracking products can input multiple feedstock oil attribute data of the sample to be tested into a trained product prediction model to predict the molecular aggregate composition of the corresponding product.

[0094] In summary, the method, system, and computer-readable storage medium for predicting hydrocracking products provided by this invention can be used to adaptively adjust the step size of the iteration to quickly and accurately obtain the molecular aggregate composition of the hydrocracking reaction.

[0095] Although the methods described above are illustrated and depicted as a series of actions for the sake of simplicity, it should be understood and appreciated that these methods are not limited by the order of the actions, as some actions may occur in a different order and / or concurrently with other actions from the illustrations and descriptions herein or not illustrated and described herein but which may be understood by those skilled in the art, according to one or more embodiments.

[0096] The prior description of this disclosure is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not intended to be limited to the examples and designs described herein, but should be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of predicting a hydrocracking product, characterized by, Includes the following steps: Sample data of various feedstock properties are obtained during the hydrocracking process, and molecular aggregate composition data of hydrocracking products are generated through a process mechanism model as the actual output value. The feedstock property data includes density, boiling point, sulfur content and nitrogen content. constructing a product prediction model based on a polynomial linear regression equation wherein, is output data of the product prediction model, , is input data of the product prediction model, n is a dimension of the input data, , is a weight of each power in the product prediction model; Based on the mini-batch gradient descent algorithm, a weight determination model for each of the aforementioned weights is constructed: wherein, α is the step size for each iteration, m is the number of samples selected for each iteration, is the input value for the k th sample, is the predicted output value for the k th sample, is the true output value for the k th sample; a raw sample set constructed from raw sample data of a plurality of training samples preprocessing is performed to remove outliers and missing values therefrom; The preprocessed original sample set The original sample data in the dataset is normalized to obtain a first sample set that conforms to a standard normal distribution. ; By introducing cross terms, the first sample set Data augmentation is performed to obtain a second sample set that conforms to the input data format of the polynomial linear regression equation. ; The weight determination model is trained with adaptive step size to determine the weights of each power, and a completed product prediction model is obtained; and The various raw material oil attribute data of the sample to be tested are input into the trained product prediction model to predict the molecular aggregate composition of the corresponding product.

2. The method for predicting hydrocracking products as described in claim 1, characterized in that, The preprocessed original sample set The steps for normalizing the original sample data include: Where X0 is the input original sample data, and X is the normalized first sample data. The mean, Let Variance be the variance.

3. The method for predicting hydrocracking products as described in claim 1, characterized in that, The first sample set The matrix form is represented as: in, For the first sample set The first sample data in, The first sample set is modified by introducing a cross term. Data augmentation is performed to obtain a second sample set that conforms to the input data format of the polynomial linear regression equation. The steps include: The first sample set The first sample data, excluding the first feature value, are multiplied pairwise to obtain the expanded second sample set. : in, For the second sample set The second sample data in the dataset.

4. The method for predicting hydrocracking products as described in claim 3, characterized in that, The matrix form of the polynomial linear regression equation is expressed as: in, p For the second sample set Second sample data The number of n For the first sample set The number of samples in the middle.

5. The method for predicting hydrocracking products as described in claim 1, characterized in that, The steps of training the weight determination model with adaptive step size include: The expanded second sample dataset Divided into training set and test set; The sample data of each training sample in the training set are sequentially input into the weights to be trained to determine the model, and the corresponding predicted output value is determined according to the preset step size and number of iterations. In response to the loss function value between the predicted output value and the corresponding true output value being greater than a preset threshold, the step size or the number of iterations is increased to determine the corresponding predicted output value again based on the increased step size and / or the increased number of iterations; and When the loss function value is less than or equal to the threshold, the training of the weight determination model is deemed complete.

6. The method for predicting hydrocracking products as described in claim 5, characterized in that, The step of determining the corresponding predicted output value based on the preset step size and number of iterations includes: The weights of each power in the product prediction model are determined according to the following expression: in, The weight is a power of j. Let the step size be... For constants used to stabilize values, This is the corrected cumulative squared gradient. For decay rate, For the cumulative squared gradient, This represents the cumulative gradient in each iteration. Here, t is the momentum coefficient, and t is the current iteration number. The corrected cumulative gradient; and The weights of each power are substituted into the product prediction model to determine the predicted output value corresponding to each sample data.

7. A prediction system for hydrocracking products, characterized in that, include: Memory, on which computer instructions are stored; as well as A processor, connected to the memory, and configured to execute computer instructions stored in the memory to implement the method for predicting hydrocracking products as described in any one of claims 1 to 6.

8. A computer-readable storage medium storing computer instructions thereon, characterized in that, When the computer instructions are executed by the processor, the method for predicting hydrocracking products as described in any one of claims 1 to 6 is implemented.

Images