Asphalt product performance prediction model construction, performance prediction method and device

Abstract

The present application provides a kind of asphalt product performance prediction model construction, performance prediction method and device, the model construction method includes: determining the sample production characteristic vector of sample asphalt product, sample production characteristic vector includes the sample asphalt product corresponding different sample crude oil component characteristic vector and the sample performance characteristic vector of sample asphalt product;Sample Resid Four Component Characteristic Vector in sample production characteristic vector is extracted, and sample Resid Four Component Characteristic Vector is input into initial regression model, and the sample predicted performance characteristic vector output by initial regression model is obtained;Sample production characteristic vector and sample predicted performance characteristic vector are spliced, to obtain sample amplification characteristic vector, and based on sample amplification characteristic vector, asphalt product performance prediction model is trained to obtain.The present application can accurately predict the asphalt product performance corresponding to different crude oil varieties and proportion based on asphalt product performance prediction model, avoid the problem that there is deviation in predicted performance based on artificial prediction.

Description

Technical Field

[0001] This invention relates to the field of petrochemical technology, and in particular to a performance prediction model for asphalt products, a performance prediction method, and an apparatus. Background Technology

[0002] Asphalt is an important product in petroleum refining. Straight-run asphalt is produced directly from atmospheric and vacuum distillation units, resulting in low production costs. However, the performance of straight-run asphalt varies with the quality and blending ratio of crude oil. Therefore, selecting an appropriate crude oil blending ratio to ensure the performance of the asphalt product is crucial.

[0003] Currently, the prediction of asphalt product performance based on different crude oil grades and blending ratios relies heavily on manual reference to historical production data and crude oil evaluation reports. However, this method introduces too many subjective human factors, leading to deviations in the predicted asphalt product performance. Summary of the Invention

[0004] This invention provides a model for predicting the performance of asphalt products, a method for predicting performance, and an apparatus to address the shortcomings of low accuracy in predicting the performance of asphalt products in existing technologies.

[0005] This invention provides a method for constructing a performance prediction model for asphalt products, comprising:

[0006] Determine the sample production feature vector of the sample asphalt product, wherein the sample production feature vector includes the feature vector of different sample crude oil components corresponding to the sample asphalt product and the sample performance feature vector of the sample asphalt product.

[0007] Extract the sample residue oil four-component feature vector from the sample production feature vector, and input the sample residue oil four-component feature vector into the initial regression model to obtain the sample prediction performance feature vector output by the initial regression model;

[0008] By concatenating the sample production feature vector with the sample predicted performance feature vector, a sample amplified feature vector is obtained, and an asphalt product performance prediction model is trained based on the sample amplified feature vector.

[0009] According to the method for constructing a performance prediction model for asphalt products provided by the present invention, determining the sample production feature vector of the sample asphalt product includes:

[0010] Determine the feature vectors of different crude oil components in the sample asphalt product;

[0011] According to the preset blending weights, the feature vectors of crude oil components in each sample are weighted and summed to obtain the sample combination feature vector;

[0012] The sample production feature vector is obtained by concatenating the sample combination feature vector with the sample performance feature vector.

[0013] The method for constructing a performance prediction model for asphalt products according to the present invention further includes:

[0014] When the performance index corresponding to any sample performance feature vector is a range index, the corresponding sample performance feature vector is represented in the form of discrete variables;

[0015] In the case where any sample crude oil component feature vector is missing, the average value of the remaining sample crude oil component feature vectors in the column containing the missing sample crude oil component feature vector is taken as the missing sample crude oil component feature vector.

[0016] If the performance index corresponding to any sample performance feature vector exceeds the instrument's range, the maximum value of the column containing the corresponding sample performance feature vector shall be taken as the corresponding sample performance feature vector.

[0017] In the case of missing performance feature vectors for any sample, the row position of the missing sample performance feature vector is identified, and the data corresponding to the row position is deleted before training the asphalt product performance prediction model.

[0018] According to the present invention, a method for constructing a performance prediction model for asphalt products is provided, wherein the initial regression model includes an initial linear regression model and an initial logistic regression model, and the sample predicted performance feature vector includes a sample continuous performance feature vector and a sample discrete performance feature vector.

[0019] The step of inputting the four-component feature vector of the sample residue oil into the initial regression model to obtain the sample prediction performance feature vector output by the initial regression model includes:

[0020] The sample residue oil four-component feature vector is input into the initial linear regression model to obtain the sample continuous performance feature vector output by the initial linear regression model;

[0021] The sample residue oil four-component feature vector is input into the initial logistic regression model to obtain the sample discrete performance feature vector output by the initial logistic regression model.

[0022] The method for constructing a performance prediction model for asphalt products according to the present invention further includes:

[0023] When the sample performance feature vector is a continuous sample performance feature vector, the base learner of the asphalt product performance prediction model is constructed based on classification and regression trees;

[0024] When the sample performance feature vector is a discrete sample performance feature vector, the base learner of the asphalt product performance prediction model is constructed based on a multi-class regression tree, and the multi-class regression tree uses Softmax as the objective function.

[0025] This invention also provides a method for predicting the performance of asphalt products, comprising:

[0026] Determine the target residual oil four-component characteristic vector of the target asphalt product;

[0027] The target residual oil four-component feature vector is input into the initial regression model to obtain the target predicted performance feature vector output by the initial regression model;

[0028] By splicing the crude oil component feature vector of the target asphalt product with the target predicted performance feature vector, the target amplified feature vector is obtained;

[0029] The target amplified feature vector is input into the asphalt product performance prediction model to obtain the performance prediction result of the target asphalt product output by the asphalt product performance prediction model.

[0030] The asphalt product performance prediction model is determined based on the asphalt product performance prediction model construction method described above.

[0031] This invention provides an apparatus for constructing a performance prediction model for asphalt products, comprising:

[0032] The first determining unit is used to determine the sample production feature vector of the sample asphalt product, wherein the sample production feature vector includes the feature vector of different sample crude oil components corresponding to the sample asphalt product and the sample performance feature vector of the sample asphalt product.

[0033] The first prediction unit is used to extract the sample residue oil four-component feature vector from the sample production feature vector, and input the sample residue oil four-component feature vector into the initial regression model to obtain the sample prediction performance feature vector output by the initial regression model.

[0034] The model building unit is used to concatenate the sample production feature vector and the sample prediction performance feature vector to obtain the sample amplified feature vector, and to train the asphalt product performance prediction model based on the sample amplified feature vector.

[0035] This invention provides an asphalt product performance prediction device, comprising:

[0036] The second determining unit is used to determine the target residual oil four-component characteristic vector of the target asphalt product;

[0037] The second prediction unit is used to input the target residue oil four-component feature vector into the initial regression model to obtain the target predicted performance feature vector output by the initial regression model.

[0038] A vector splicing unit is used to splice the crude oil component feature vector of the target asphalt product with the target predicted performance feature vector to obtain the target amplified feature vector;

[0039] The performance prediction unit is used to input the target amplified feature vector into the asphalt product performance prediction model to obtain the performance prediction result of the target asphalt product output by the asphalt product performance prediction model.

[0040] The asphalt product performance prediction model is determined based on the asphalt product performance prediction model construction method described above.

[0041] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the asphalt product performance prediction model construction method or the asphalt product performance prediction method described above.

[0042] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the asphalt product performance prediction model construction method or the asphalt product performance prediction method as described above.

[0043] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the asphalt product performance prediction model construction method or the asphalt product performance prediction method as described above.

[0044] The asphalt product performance prediction model construction, performance prediction method and device provided by the present invention use the sample amplified feature vector obtained by concatenating the sample production feature vector and the sample predicted performance feature vector as the training set to train the asphalt product performance prediction model. In this way, the asphalt product performance prediction model can accurately predict the performance of asphalt products corresponding to different crude oil varieties and ratios, avoiding the problem of performance deviation caused by manual prediction. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0046] Figure 1This is a flowchart illustrating the method for constructing a performance prediction model for asphalt products provided by the present invention.

[0047] Figure 2 This is a flowchart illustrating the asphalt product performance prediction method provided by the present invention;

[0048] Figure 3 This is a flowchart illustrating another method for predicting the performance of asphalt products provided by the present invention.

[0049] Figure 4 This is a schematic diagram of the asphalt product performance prediction model construction device provided by the present invention;

[0050] Figure 5 This is a schematic diagram of the asphalt product performance prediction device provided by the present invention;

[0051] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0053] This invention provides a method for constructing a performance prediction model for asphalt products. Figure 1 This is a flowchart illustrating the method for constructing a performance prediction model for asphalt products provided by this invention. Figure 1 As shown, the method includes the following steps:

[0054] Step 110: Determine the sample production feature vector of the sample asphalt product. The sample production feature vector includes the feature vectors of different sample crude oil components corresponding to the sample asphalt product and the sample performance feature vector of the sample asphalt product.

[0055] Specifically, the sample asphalt products are processed from sample crude oil with different blend ratios. The sample production feature vector of the sample asphalt products can be determined based on the historical production data of the sample asphalt products. The historical production data can include crude oil type, crude oil blend ratio, and the performance of sample asphalt products under different crude oil blend ratios. The performance of the sample asphalt products can be characterized by key indicators of the sample asphalt products, such as penetration, ductility, and dynamic viscosity.

[0056] The sample crude oil component feature vector contains distillation data and compositional properties of each fraction of the sample crude oil, while the sample performance feature vector contains performance indicators of the sample asphalt product. Optionally, the sample performance feature vector may include key performance indicators of the sample asphalt product.

[0057] In some specific implementations, the sample combination feature vector and the sample performance feature vector can be concatenated to obtain the sample production feature vector.

[0058] Step 120: Extract the sample residue oil four-component feature vector from the sample production feature vector, and input the sample residue oil four-component feature vector into the initial regression model to obtain the sample prediction performance feature vector output by the initial regression model.

[0059] Specifically, the sample crude oil composition features include the sample residue four-component features. Therefore, after determining the sample production feature vector containing the sample crude oil composition feature vector, the sample residue four-component feature vector can be extracted from the sample production feature vector.

[0060] The four-component feature vector of the sample residue oil is used to characterize the fractional information of the sample residue oil, namely, information on saturated hydrocarbons, aromatic hydrocarbons, resins, and asphaltenes. The four-component feature vector of the sample residue oil is input into the initial regression model, which then predicts the performance of the sample asphalt products based on the four-component feature vector of the sample residue oil, thus obtaining the sample predicted performance feature vector.

[0061] Step 130: Combine the sample production feature vector with the sample prediction performance feature vector to obtain the sample amplified feature vector, and train the asphalt product performance prediction model based on the sample amplified feature vector.

[0062] Specifically, after obtaining the sample predicted performance feature vector, the sample predicted performance feature vector can be used as a reference vector for predicting the performance of the sample asphalt product. Therefore, in this embodiment of the invention, the sample production feature vector and the sample predicted performance feature vector are concatenated to amplify the sample production feature vector, resulting in a sample amplified vector, and an asphalt product performance prediction model is trained based on the sample amplified feature vector.

[0063] In some specific implementations, the sample crude oil component feature vector and the sample predicted performance feature vector in the sample amplification vector are used as inputs, and the sample performance feature vector in the sample amplification vector is used as a label to train the initial asphalt product performance prediction model, thereby obtaining the asphalt product performance prediction model.

[0064] Therefore, the asphalt product performance prediction model construction method provided in this embodiment of the invention uses the sample amplified feature vector obtained by concatenating the sample production feature vector and the sample prediction performance feature vector as the training set to train the asphalt product performance prediction model. This enables accurate prediction of asphalt product performance corresponding to different crude oil varieties and ratios based on the asphalt product performance prediction model, avoiding the problem of performance deviation caused by manual prediction.

[0065] Based on the above embodiments, the sample production feature vector of the sample asphalt product is determined, including:

[0066] Determine the feature vectors of different crude oil components in the sample asphalt product;

[0067] According to the preset blending weights, the feature vectors of crude oil components in each sample are weighted and summed to obtain the sample combination feature vector;

[0068] By concatenating the sample combination feature vector and the sample performance feature vector, the sample production feature vector is obtained.

[0069] In some specific implementations, historical production data of the sample asphalt products is collected using the LIMS (Laboratory Information Management System). This historical production data includes the crude oil type and blending ratio, operating conditions, and property analysis reports of the sample asphalt products. Based on the historical production data, the various test properties of the sample asphalt products and their corresponding dates are obtained. Then, based on the processing log, the type and proportion of sample crude oil processed on that day are obtained. According to the name of each sample crude oil and its corresponding date, the most recent evaluation report is retrieved from the sample crude oil database. The distillation data and compositional properties of each fraction are extracted from the report to form a sample crude oil component feature vector for each sample asphalt product.

[0070]

[0071] All sample crude oil component feature vectors A weighted average is performed based on preset mixing weights to obtain the sample combination feature vector for each sample asphalt product.

[0072]

[0073] Where, r i Let be the preset blending weight of the i-th type of crude oil, and have

[0074] Obtain the sample combination feature vector Then, the sample performance feature vector of the sample asphalt product is... Concatenated in the sample combination feature vector Then, the sample production feature vector of each sample asphalt product is obtained.

[0075]

[0076] Among them, y i1 ,y i2 ,…,y im Let m be the key indicators of the sample asphalt product.

[0077] The historical production data of all sample asphalt products are represented as sample production feature vectors, thus obtaining the sample production feature vectors for all sample asphalt products.

[0078]

[0079] Based on any of the above embodiments, the method further includes:

[0080] When the performance index corresponding to any sample performance feature vector is a range index, the corresponding sample performance feature vector is represented in the form of discrete variables;

[0081] In the case of missing feature vectors of any sample crude oil component, the average value of the feature vectors of the remaining sample crude oil components in the column containing the missing sample crude oil component feature vector is taken as the missing sample crude oil component feature vector.

[0082] If the performance index corresponding to any sample performance feature vector exceeds the instrument's range, the maximum value of the column containing the corresponding sample performance feature vector shall be taken as the corresponding sample performance feature vector.

[0083] If any sample performance feature vector is missing, the row position of the missing sample performance feature vector is identified, and the data corresponding to the row position is deleted before training the asphalt product performance prediction model.

[0084] In some specific implementations, when the performance index corresponding to any sample performance feature vector is a range index, it indicates that the corresponding sample performance index is difficult to represent with precise detection values, but is represented by a numerical range. In this case, the corresponding sample performance feature vector can be stored in a discrete format, such as converting the corresponding numerical range into a discrete variable representation of 0, 1, 2, 3, ... in sequence.

[0085] If any sample crude oil component feature vector is missing, it is filled by the average value of the remaining sample crude oil component feature vectors in the column containing the missing sample crude oil component feature vector, that is, the average value is used as the missing sample crude oil component feature vector.

[0086] If the performance index corresponding to any sample performance feature vector exceeds the instrument's range, the maximum value of the column containing the corresponding sample performance feature vector shall be taken as the corresponding sample performance feature vector.

[0087] In the case of missing performance feature vectors for any sample, the row position of the missing sample performance feature vector is identified so that the data corresponding to that row position is discarded during training. This can be understood as deleting the data corresponding to the row position before training the asphalt product performance prediction model.

[0088] Based on any of the above embodiments, the initial regression model includes an initial linear regression model and an initial logistic regression model, and the sample prediction performance feature vector includes a sample continuous performance feature vector and a sample discrete performance feature vector.

[0089] The four-component feature vectors of the sample residue oil are input into the initial regression model to obtain the sample prediction performance feature vectors output by the initial regression model, including:

[0090] The four component feature vectors of the sample residue oil are input into the initial linear regression model to obtain the sample continuous performance feature vectors output by the initial linear regression model.

[0091] The four component feature vectors of the sample residue oil are input into the initial logistic regression model to obtain the discrete performance feature vectors of the sample output by the initial logistic regression model.

[0092] Specifically, the initial regression models include an initial linear regression model and an initial logistic regression model. The initial linear regression model is used to predict the continuous performance feature vector of the samples, and the initial logistic regression model is used to predict the discrete performance feature vector of the samples. Here, the continuous performance feature vector refers to the performance vector represented by a range of values, and the discrete performance feature vector refers to the performance vector represented by discrete values.

[0093] In some specific implementations, feature vectors are generated from samples. Extracting the four component feature vectors of sample residue oil Sample residue oil four-component feature vector Each row of data is as follows:

[0094]

[0095] Among them, Ar i ,Sa i At i Re i This is the data for the four components of the sample residue oil.

[0096] Using the four-component feature vector of the sample residue oil Establish m initial linear regression models and initial logistic regression models. Use the initial linear regression models to predict the continuous performance feature vector of the sample asphalt products, and use the initial logistic regression models to predict the discrete performance feature vector of the sample asphalt products. Then, combine the predicted values ​​from both models. As a new feature vector, it is added to the sample production feature vector. In this process, the sample amplification feature vector is obtained.

[0097] Based on any of the above embodiments, the method further includes:

[0098] When the sample performance feature vector is a continuous sample performance feature vector, the base learner of the asphalt product performance prediction model is constructed based on classification and regression trees;

[0099] When the sample performance feature vector is a discrete sample performance feature vector, the base learner of the asphalt product performance prediction model is constructed based on a multi-class regression tree, and the multi-class regression tree uses Softmax as the objective function.

[0100] In some specific implementations, when the sample performance feature vector is a continuous sample performance feature vector, the base learner of the asphalt product performance prediction model adopts conventional classification and regression tree construction (CART), the training objective is regression, and the loss function is mean squared error (MSE). Optionally, the asphalt product performance prediction model can be a boosting tree model.

[0101] When the sample performance feature vector is discrete, the base learner of the asphalt product performance prediction model uses Softmax as the objective function of the multi-class regression tree, and its judgment form is as follows:

[0102]

[0103] Among them, z i For the output of the classification tree, Softmax(z) i ) represents the probability that the sample belongs to the i-th category.

[0104] Specifically, a gradient boosting decision tree model with regularization control based on second-order Taylor expansion is used in the modeling process. The objective function consists of two parts: a loss function and a regularization term. A second-order Taylor expansion is performed to remove the constant term and optimize the loss function; the regularization term is expanded to remove the constant term and optimize the regularization term; the coefficients of the first and second terms are combined to obtain the final objective function. 85% of the amplified feature vectors from the samples are randomly allocated to the training set, and 15% to the validation set. The asphalt product performance prediction model is trained using the training set, and its performance is observed on the validation set to control the model depth and prevent overfitting.

[0105] It is understandable that when determining the sample prediction performance feature vector, an independent regression model can be trained based on the performance indicators of each type of asphalt product. The form of each regression model is fixed as inputting an n-dimensional feature vector and outputting a sample prediction performance feature vector.

[0106] Based on any of the above embodiments, the present invention also provides a method for predicting the performance of asphalt products, such as... Figure 2 As shown, the method includes:

[0107] Step 210: Determine the target residual oil four-component feature vector of the target asphalt product;

[0108] Step 220: Input the target residual oil four-component feature vector into the initial regression model to obtain the target predicted performance feature vector output by the initial regression model;

[0109] Step 230: Combine the crude oil component feature vector of the target asphalt product with the target predicted performance feature vector to obtain the target amplified feature vector;

[0110] Step 240: Input the target amplified feature vector into the asphalt product performance prediction model to obtain the performance prediction result of the target asphalt product output by the asphalt product performance prediction model.

[0111] The asphalt product performance prediction model is determined based on the asphalt product performance prediction model construction method described in any of the above embodiments.

[0112] Here, the target asphalt product is the asphalt product whose performance needs to be predicted.

[0113] In some specific implementations, the target residue four-component feature vectors for the target asphalt product are determined by referring to the method described above for determining the four-component feature vectors of the sample residue oil. The target residue four-component feature vectors are then input into an initial regression model to obtain the target predicted performance feature vectors output by the initial regression model. These target predicted performance feature vectors include target continuous performance feature vectors and target discrete performance feature vectors. The target continuous performance feature vectors refer to performance vectors expressed using a range of values, while the target discrete performance feature vectors refer to performance vectors expressed using discrete values. The initial regression model includes an initial linear regression model and an initial logistic regression model. The initial linear regression model is used to predict the target continuous performance feature vectors, and the initial logistic regression model is used to predict the target discrete performance feature vectors.

[0114] After obtaining the target predicted performance feature vector, the crude oil component feature vector of the target asphalt product is concatenated with the target predicted performance feature vector to obtain the target amplified feature vector. The target amplified feature vector can then be input into the asphalt product performance prediction model to obtain the performance prediction result of the target asphalt product output by the asphalt product performance prediction model.

[0115] Based on any of the above embodiments, the present invention also provides a method for predicting the performance of asphalt products, such as... Figure 3 As shown, the method includes:

[0116] First, historical production data of sample asphalt products are obtained, and feature vectors of different sample crude oil components and sample performance feature vectors of sample asphalt products are extracted from the historical production data. The feature vectors of different sample crude oil components and sample performance feature vectors are then concatenated to obtain the sample production feature vector.

[0117] Extract the four component feature vectors of the sample residue oil from the sample production feature vector, and input the four component feature vectors of the sample residue oil into the initial regression model to obtain the sample prediction performance feature vector output by the initial regression model.

[0118] By splicing the sample production feature vector and the sample prediction performance feature vector, a sample amplified feature vector is obtained, and an asphalt product performance prediction model is trained based on the sample amplified feature vector.

[0119] After training and obtaining the product performance prediction model, the target residue oil four-component feature vector of the target asphalt product is determined by referring to the method described above for determining the feature vector of the four components of the sample residue oil. The target residue oil four-component feature vector is input into the initial regression model to obtain the target predicted performance feature vector output by the initial regression model; the crude oil component feature vector of the target asphalt product is concatenated with the target predicted performance feature vector to obtain the target amplified feature vector; the target amplified feature vector is input into the asphalt product performance prediction model to obtain the performance prediction result of the target asphalt product output by the asphalt product performance prediction model.

[0120] The device for constructing a prediction model for asphalt product performance provided by the present invention is described below. The device for constructing a prediction model for asphalt product performance described below can be referred to in correspondence with the method for constructing a prediction model for asphalt product performance described above.

[0121] Based on any of the above embodiments, the present invention also provides an apparatus for constructing an asphalt product performance prediction model, such as... Figure 4 As shown, the device includes:

[0122] The first determining unit 410 is used to determine the sample production feature vector of the sample asphalt product, wherein the sample production feature vector includes the feature vector of different sample crude oil components corresponding to the sample asphalt product and the sample performance feature vector of the sample asphalt product.

[0123] The first prediction unit 420 is used to extract the sample residue oil four-component feature vector from the sample production feature vector, and input the sample residue oil four-component feature vector into the initial regression model to obtain the sample prediction performance feature vector output by the initial regression model.

[0124] The model building unit 430 is used to concatenate the sample production feature vector and the sample prediction performance feature vector to obtain the sample amplification feature vector, and to train the asphalt product performance prediction model based on the sample amplification feature vector.

[0125] The asphalt product performance prediction device provided by the present invention is described below. The asphalt product performance prediction device described below can be referred to in correspondence with the asphalt product performance prediction method described above.

[0126] Based on any of the above embodiments, the present invention also provides an asphalt product performance prediction device, such as... Figure 5 As shown, the device includes:

[0127] The second determining unit 510 is used to determine the target residual oil four-component characteristic vector of the target asphalt product;

[0128] The second prediction unit 520 is used to input the target residue oil four-component feature vector into the initial regression model to obtain the target predicted performance feature vector output by the initial regression model.

[0129] The vector splicing unit 530 is used to splice the crude oil component feature vector of the target asphalt product with the target predicted performance feature vector to obtain the target amplified feature vector.

[0130] The performance prediction unit 540 is used to input the target amplified feature vector into the asphalt product performance prediction model to obtain the performance prediction result of the target asphalt product output by the asphalt product performance prediction model.

[0131] The asphalt product performance prediction model is determined based on the asphalt product performance prediction model construction method described in any of the above embodiments.

[0132] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 6 As shown, the electronic device may include a processor 610, a memory 620, a communication interface 630, and a communication bus 640, wherein the processor 610, memory 620, and communication interface 630 communicate with each other through the communication bus 640. The processor 610 can call logical instructions in the memory 620 to execute a method for constructing an asphalt product performance prediction model. This method includes: determining a sample production feature vector for a sample asphalt product, the sample production feature vector including feature vectors of different sample crude oil components corresponding to the sample asphalt product and a sample performance feature vector for the sample asphalt product; extracting the sample residue four-component feature vector from the sample production feature vector, and inputting the sample residue four-component feature vector into an initial regression model to obtain a sample predicted performance feature vector output by the initial regression model; concatenating the sample production feature vector with the sample predicted performance feature vector to obtain a sample amplified feature vector, and training an asphalt product performance prediction model based on the sample amplified feature vector.

[0133] Alternatively, an asphalt product performance prediction method may be implemented, comprising: determining the target residual oil four-component feature vector of the target asphalt product; inputting the target residual oil four-component feature vector into an initial regression model to obtain the target predicted performance feature vector output by the initial regression model; concatenating the crude oil component feature vector of the target asphalt product with the target predicted performance feature vector to obtain a target amplified feature vector; inputting the target amplified feature vector into an asphalt product performance prediction model to obtain the performance prediction result of the target asphalt product output by the asphalt product performance prediction model; wherein, the asphalt product performance prediction model is determined based on the asphalt product performance prediction model construction method described above.

[0134] Furthermore, the logical instructions in the aforementioned memory 620 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0135] On the other hand, the present invention also provides a computer program product, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, when the program instructions are executed by a computer, the computer is able to execute the asphalt product performance prediction model construction method provided by the above methods, the method including: determining a sample production feature vector of a sample asphalt product, the sample production feature vector including feature vectors of different sample crude oil components corresponding to the sample asphalt product and a sample performance feature vector of the sample asphalt product; extracting the sample residue four-component feature vector from the sample production feature vector, and inputting the sample residue four-component feature vector into an initial regression model to obtain a sample prediction performance feature vector output by the initial regression model; concatenating the sample production feature vector and the sample prediction performance feature vector to obtain a sample amplified feature vector, and training an asphalt product performance prediction model based on the sample amplified feature vector.

[0136] Alternatively, an asphalt product performance prediction method may be implemented, comprising: determining the target residual oil four-component feature vector of the target asphalt product; inputting the target residual oil four-component feature vector into an initial regression model to obtain the target predicted performance feature vector output by the initial regression model; concatenating the crude oil component feature vector of the target asphalt product with the target predicted performance feature vector to obtain a target amplified feature vector; inputting the target amplified feature vector into an asphalt product performance prediction model to obtain the performance prediction result of the target asphalt product output by the asphalt product performance prediction model; wherein, the asphalt product performance prediction model is determined based on the asphalt product performance prediction model construction method described above.

[0137] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the above-described methods for constructing asphalt product performance prediction models. The method includes: determining a sample production feature vector of a sample asphalt product, the sample production feature vector including feature vectors of different sample crude oil components corresponding to the sample asphalt product and a sample performance feature vector of the sample asphalt product; extracting the sample residue four-component feature vector from the sample production feature vector and inputting the sample residue four-component feature vector into an initial regression model to obtain a sample prediction performance feature vector output by the initial regression model; concatenating the sample production feature vector and the sample prediction performance feature vector to obtain a sample amplified feature vector, and training an asphalt product performance prediction model based on the sample amplified feature vector.

[0138] Alternatively, an asphalt product performance prediction method may be implemented, comprising: determining the target residual oil four-component feature vector of the target asphalt product; inputting the target residual oil four-component feature vector into an initial regression model to obtain the target predicted performance feature vector output by the initial regression model; concatenating the crude oil component feature vector of the target asphalt product with the target predicted performance feature vector to obtain a target amplified feature vector; inputting the target amplified feature vector into an asphalt product performance prediction model to obtain the performance prediction result of the target asphalt product output by the asphalt product performance prediction model; wherein, the asphalt product performance prediction model is determined based on the asphalt product performance prediction model construction method described above.

[0139] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. 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. Those skilled in the art can understand and implement this without any creative effort.

[0140] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical feature vectors; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a performance prediction model for asphalt products, characterized in that, include: Determine the sample production feature vector of the sample asphalt product, wherein the sample production feature vector includes the feature vector of different sample crude oil components corresponding to the sample asphalt product and the sample performance feature vector of the sample asphalt product. Extract the sample residue oil four-component feature vector from the sample production feature vector, and input the sample residue oil four-component feature vector into the initial regression model to obtain the sample prediction performance feature vector output by the initial regression model; By concatenating the sample production feature vector with the sample predicted performance feature vector, a sample amplified feature vector is obtained, and an asphalt product performance prediction model is trained based on the sample amplified feature vector. Also includes: When the sample performance feature vector is a continuous sample performance feature vector, the base learner of the asphalt product performance prediction model is constructed based on classification and regression trees; When the sample performance feature vector is a discrete sample performance feature vector, the base learner of the asphalt product performance prediction model is constructed based on a multi-class regression tree, and the multi-class regression tree uses Softmax as the objective function.

2. The method for constructing a performance prediction model for asphalt products according to claim 1, characterized in that, The determination of the sample production feature vector of the sample asphalt product includes: Determine the feature vectors of different crude oil components in the sample asphalt product; According to the preset blending weights, the feature vectors of crude oil components in each sample are weighted and summed to obtain the sample combination feature vector; The sample production feature vector is obtained by concatenating the sample combination feature vector with the sample performance feature vector.

3. The method for constructing a performance prediction model for asphalt products according to claim 1, characterized in that, Also includes: When the performance index corresponding to any sample performance feature vector is a range index, the corresponding sample performance feature vector is represented in the form of discrete variables; In the case where any sample crude oil component feature vector is missing, the average value of the remaining sample crude oil component feature vectors in the column containing the missing sample crude oil component feature vector is taken as the missing sample crude oil component feature vector. If the performance index corresponding to any sample performance feature vector exceeds the instrument's range, the maximum value in the column containing the corresponding sample performance feature vector shall be taken as the corresponding sample performance feature vector. In the case of missing performance feature vectors for any sample, the row position of the missing sample performance feature vector is identified, and the data corresponding to the row position is deleted before training the asphalt product performance prediction model.

4. The method for constructing a performance prediction model for asphalt products according to claim 1, characterized in that, The initial regression model includes an initial linear regression model and an initial logistic regression model, and the sample prediction performance feature vector includes a sample continuous performance feature vector and a sample discrete performance feature vector. The step of inputting the four-component feature vector of the sample residue oil into the initial regression model to obtain the sample prediction performance feature vector output by the initial regression model includes: The sample residue oil four-component feature vector is input into the initial linear regression model to obtain the sample continuous performance feature vector output by the initial linear regression model; The sample residue oil four-component feature vector is input into the initial logistic regression model to obtain the sample discrete performance feature vector output by the initial logistic regression model.

5. A method for predicting the performance of asphalt products, characterized in that, include: Determine the target residual oil four-component characteristic vector of the target asphalt product; The target residual oil four-component feature vector is input into the initial regression model to obtain the target predicted performance feature vector output by the initial regression model; By splicing the crude oil component feature vector of the target asphalt product with the target predicted performance feature vector, the target amplified feature vector is obtained; The target amplified feature vector is input into the asphalt product performance prediction model to obtain the performance prediction result of the target asphalt product output by the asphalt product performance prediction model. The asphalt product performance prediction model is determined based on the asphalt product performance prediction model construction method according to any one of claims 1 to 4.

6. A device for constructing a performance prediction model for asphalt products, characterized in that, include: The first determining unit is used to determine the sample production feature vector of the sample asphalt product, wherein the sample production feature vector includes the feature vector of different sample crude oil components corresponding to the sample asphalt product and the sample performance feature vector of the sample asphalt product. The first prediction unit is used to extract the sample residue oil four-component feature vector from the sample production feature vector, and input the sample residue oil four-component feature vector into the initial regression model to obtain the sample prediction performance feature vector output by the initial regression model. The model building unit is used to concatenate the sample production feature vector and the sample prediction performance feature vector to obtain the sample amplified feature vector, and to train the asphalt product performance prediction model based on the sample amplified feature vector. Also includes: When the sample performance feature vector is a continuous sample performance feature vector, the base learner of the asphalt product performance prediction model is constructed based on classification and regression trees; When the sample performance feature vector is a discrete sample performance feature vector, the base learner of the asphalt product performance prediction model is constructed based on a multi-class regression tree, and the multi-class regression tree uses Softmax as the objective function.

7. A device for predicting the performance of asphalt products, characterized in that, include: The second determining unit is used to determine the target residual oil four-component characteristic vector of the target asphalt product; The second prediction unit is used to input the target residue oil four-component feature vector into the initial regression model to obtain the target predicted performance feature vector output by the initial regression model. A vector splicing unit is used to splice the crude oil component feature vector of the target asphalt product with the target predicted performance feature vector to obtain the target amplified feature vector; The performance prediction unit is used to input the target amplified feature vector into the asphalt product performance prediction model to obtain the performance prediction result of the target asphalt product output by the asphalt product performance prediction model. The asphalt product performance prediction model is determined based on the asphalt product performance prediction model construction method according to any one of claims 1 to 4.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the asphalt product performance prediction model construction method as described in any one of claims 1 to 4, or when the processor executes the program, it implements the asphalt product performance prediction method as described in claim 5.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the asphalt product performance prediction model construction method as described in any one of claims 1 to 4, or when the computer program is executed by the processor, it implements the asphalt product performance prediction method as described in claim 5.

Images