Training method, prediction method and device of diesel cetane number prediction model

Abstract

The invention discloses a training method, a prediction method and a device of a diesel cetane number prediction model, and belongs to the technical field of petrochemical engineering, the training method of the diesel cetane number prediction model comprises the following steps: obtaining a training data set, the training data set comprising explanatory variables and corresponding target variables, the explanatory variables comprise physicochemical index data of different diesel oil samples, and the target variables comprise cetane numbers of different diesel oil samples; and based on the training data set, training a deep learning regression model to obtain a diesel cetane number prediction model. By utilizing the diesel cetane number prediction model constructed by the method, the cetane number of the diesel can be rapidly and accurately predicted.

Description

Technical Field

[0001] The invention belongs to the technical field of petrochemical industry, and in particular relates to a training method, a prediction method and a device for a diesel cetane number prediction model. Background Art

[0002] Cetane number is a key indicator for measuring diesel combustion performance. It refers to the result of comparing the volume percentage of n-cetane in diesel with that in standard fuel under specific conditions. This indicator is used to indicate the self-ignition property of diesel, that is, the ability of diesel to ignite itself in the engine. Diesel with a high cetane number has better self-ignition property, which makes it easier to start and burn, thereby significantly improving the performance and efficiency of the engine. Therefore, accurate determination of the cetane number is crucial to ensure the quality of diesel, which is directly related to the performance and emission level of the engine. If the cetane number is not measured accurately, it may cause unstable engine operation, increase maintenance costs, and even have a negative impact on the environment.

[0003] To determine the cetane number, specialized cetane number testing instruments can be used, which are measured by dielectric properties or other physical methods. The testing process usually includes steps such as diesel sample preparation, instrument connection, parameter setting, and automatic detection and analysis. However, cetane number testing instruments are usually expensive, complicated to operate, and time-consuming, so it is impossible to fully configure a standard cetane number test engine to test all diesel samples.

[0004] Another low-cost and simple method is to use the cetane index calculated by an empirical formula to characterize the cetane number. However, it should be noted that the cetane index cannot be used arbitrarily to replace the cetane number measured by the standard engine test device. In actual use, the cetane index has certain limitations. The correlation between the cetane index and the cetane number of diesel produced from different crude oil sources and different blending schemes is different. Especially in the case of high cetane numbers, the correlation between the two will drop significantly. Therefore, the cetane index can only be used as an auxiliary means to determine the cetane number of diesel, and cannot completely replace the standard cetane number determination method.

[0005] In summary, it is of great significance to design and develop a method that can quickly and accurately predict the cetane number of diesel. Summary of the invention

[0006] The purpose of the present invention is to meet practical needs and provide a training method, a prediction method and a device for a diesel cetane number prediction model. The diesel cetane number prediction model constructed by the method can quickly and accurately predict the cetane number of diesel.

[0007] To achieve the above-mentioned object of the invention, the first object of the present invention is to provide a training method for a diesel cetane number prediction model, comprising the following steps:

[0008] Acquire a training data set, wherein the training data set includes explanatory variables and corresponding target variables, wherein the explanatory variables include physical and chemical index data of different diesel samples, and the target variables include cetane numbers of different diesel samples;

[0009] Based on the training data set, a deep learning regression model is trained to obtain a diesel cetane number prediction model.

[0010] Preferably, obtaining a training data set includes:

[0011] Sampling and testing diesel fuel to obtain the physical and chemical index data of diesel fuel samples and the cetane number corresponding to each diesel fuel sample;

[0012] The training data set is constructed using the physical and chemical index data of diesel samples and the cetane number corresponding to each diesel sample.

[0013] Preferably, the step of obtaining the cetane number of each diesel sample comprises:

[0014] Under the standard operating conditions of the cetane number tester, the diesel sample to be tested is mixed with different standard fuels with known cetane numbers;

[0015] The mixture was subjected to the fire property test under the same conditions;

[0016] The cetane number of the diesel sample to be tested is obtained by comparison.

[0017] Preferably, the method for acquiring the deep learning regression model includes: establishing a regression model through the h2o package in R software.

[0018] Preferably, after obtaining the diesel cetane number prediction model, the fitting coefficient R 2 Measuring the predictive ability of the diesel cetane number prediction model; specifically including:

[0019] When R 2 When ≤0.3, the predicted value and the observed value have a poor fitting effect, and the prediction ability of the prediction model is poor;

[0020] When 0.3<R 2 When ≤0.4, the fitting effect between the predicted value and the observed value is weak, and the prediction ability of the prediction model is weak;

[0021] When 0.4<R 2 When ≤0.6, the fitting effect between the predicted value and the observed value is moderate, and the prediction ability of the prediction model is moderate;

[0022] When 0.6<R2 When ≤1.0, the predicted value and the observed value have a strong fitting effect, and the prediction model has a strong prediction ability.

[0023] Preferably, the physical and chemical index data include: density at 20°C, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, clogging tendency, cold filter point, sulfur content, chlorine content, freezing point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, 20°C kinematic viscosity, turbidity point, and total pollutant content.

[0024] A second object of the present invention is to provide a method for predicting the cetane number of diesel, comprising:

[0025] Obtain the physical and chemical index data of the diesel sample to be tested;

[0026] The physical and chemical index data of the diesel sample to be tested are input into the above diesel cetane number prediction model to obtain the cetane number of the diesel sample to be tested.

[0027] A third object of the present invention is to provide a training device for a diesel cetane number prediction model, comprising:

[0028] A data acquisition module, which acquires a training data set, wherein the training data set includes explanatory variables and corresponding target variables, wherein the explanatory variables include physical and chemical index data of different diesel samples, and the target variables include cetane numbers of different diesel samples;

[0029] The model training module trains a deep learning regression model based on the training data set to obtain a diesel cetane number prediction model.

[0030] Preferably, obtaining a training data set includes:

[0031] Sampling and testing diesel fuel to obtain the physical and chemical index data of diesel fuel samples and the cetane number corresponding to each diesel fuel sample;

[0032] The training data set is constructed using the physical and chemical index data of diesel samples and the cetane number corresponding to each diesel sample;

[0033] The steps for obtaining the cetane number of each diesel sample include:

[0034] Under the standard operating conditions of the cetane number tester, the diesel sample to be tested is mixed with different standard fuels with known cetane numbers;

[0035] The mixture was subjected to the fire property test under the same conditions;

[0036] The cetane number of the diesel sample to be tested is obtained by comparison.

[0037] Preferably, the method for acquiring the deep learning regression model includes: establishing a regression model through the h2o package in R software.

[0038] Preferably, after obtaining the diesel cetane number prediction model, the fitting coefficient R 2 Measuring the predictive ability of the diesel cetane number prediction model; specifically including:

[0039] When R 2 When ≤0.3, the predicted value and the observed value have a poor fitting effect, and the prediction ability of the prediction model is poor;

[0040] When 0.3<R 2 When ≤0.4, the fitting effect between the predicted value and the observed value is weak, and the prediction ability of the prediction model is weak;

[0041] When 0.4<R 2 When ≤0.6, the fitting effect between the predicted value and the observed value is moderate, and the prediction ability of the prediction model is moderate;

[0042] When 0.6<R 2 When ≤1.0, the predicted value and the observed value have a strong fitting effect, and the prediction model has a strong prediction ability.

[0043] Preferably, the physical and chemical index data include: density at 20°C, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, clogging tendency, cold filter point, sulfur content, chlorine content, freezing point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, 20°C kinematic viscosity, turbidity point, and total pollutant content.

[0044] A fourth object of the present invention is to provide a computer-readable storage medium storing a computer program, which implements the above-mentioned diesel cetane number prediction method when executed by a processor.

[0045] A fifth object of the present invention is to provide a computer program product, comprising a computer program, which implements the above-mentioned diesel cetane number prediction method when executed by a processor.

[0046] The advantages and positive effects of this application are:

[0047] The core of the present invention is to obtain the physical and chemical index data of diesel samples in advance, construct a data set, and use these data sets to perform an automated machine learning process, so as to construct a prediction model that can predict the cetane number of the diesel to be tested. Specifically, the present invention selects common physical and chemical indicators in the diesel sample detection process as explanatory variables, and these indicators include but are not limited to density at 20°C, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, clogging tendency, cold filter point, sulfur content, chlorine content, freezing point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, kinematic viscosity (measured value at 20°C), turbidity point and total pollutant content. These indicators will be used as input variables, and the cetane number will be used as the target variable. By adopting a deep learning algorithm, the present invention successfully established an efficient and accurate prediction model.

[0048] After a series of verification experiments, the results show that the prediction model constructed by the method provided by the present invention has a strong fitting effect. This means that the cetane number can be accurately predicted only by the physical and chemical index test results of the diesel sample. This method can not only effectively solve the complexity and the problem of being unable to monitor in real time in the traditional index detection process, thereby saving a lot of time and human resources and avoiding cumbersome detection processes, but also can deeply analyze the intrinsic response relationship between the physical and chemical indicators of diesel and the cetane number based on an interpretable machine learning method. In this way, researchers and engineers can better understand the performance characteristics of diesel, and then optimize the production and use of diesel. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0050] Figure 1 A flow chart showing a training method provided by a preferred embodiment of the present invention;

[0051] Figure 2 A flowchart of constructing a training data set in a preferred embodiment of the present invention is shown;

[0052] Figure 3 A flow chart for obtaining the cetane number of each diesel sample in a preferred embodiment of the present invention is shown;

[0053] Figure 4 A flowchart of a prediction method provided by a preferred embodiment of the present invention is shown;

[0054] Figure 5shows a neural network diagram selected in a preferred embodiment of the present invention;

[0055] Figure 6 A fitting relationship diagram between the detection value and the predicted value in a preferred embodiment of the present invention is shown;

[0056] Figure 7 A partial dependency diagram between the target variable and each explanatory variable in the prediction model of the preferred embodiment of the present invention is shown. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] See also Figure 1 ;

[0059] The first embodiment is a method for training a diesel cetane number prediction model, comprising the following steps:

[0060] Obtain a training data set. The training data set is the core foundation of model construction, and its selection has a decisive influence on the output results of the model. In order to build an ideal prediction model, the training data set used in this embodiment mainly includes explanatory variables and their corresponding target variables. The explanatory variables cover the physical and chemical index data of different diesel samples, while the target variables include the cetane number of the corresponding diesel samples;

[0061] Based on the provided training data set, the deep learning regression model is trained to obtain a model for predicting the diesel cetane number.

[0062] In order to more deeply understand the technical solution of the present invention, the following non-limiting examples are given:

[0063] See also Figure 2 In the process of obtaining the training data set, diesel samples must first be collected and tested to obtain the physical and chemical property index data of the samples and the corresponding cetane number; then, based on these physical and chemical index data and the corresponding cetane number, a data set for training is constructed.

[0064] See also Figure 3 , the steps of obtaining the cetane number of each diesel sample include:

[0065] First, under the standard operating conditions of the cetane number tester, the diesel sample to be tested is mixed with different standard fuels with known cetane numbers;

[0066] Subsequently, the mixture was tested for its ignition properties under the same conditions;

[0067] Finally, the cetane number of the diesel sample to be tested is obtained through comparative analysis.

[0068] See also Figure 5 In this embodiment, the method for obtaining the deep learning regression model includes the following steps: using the h2o package in the R software to build the required regression model. Specifically, the R software is a widely used statistical analysis and graphic representation tool that provides a wealth of packages and functions to handle various data analysis tasks. The h2o package is a package specifically used for big data analysis and automatic machine learning, which supports a variety of machine learning algorithms, including deep learning regression models.

[0069] By using the h2o package, users can easily train deep learning regression models. First, users need to install and load the h2o package, and then import the dataset to be processed. Next, users can use automatic machine learning to define and optimize the model and its parameters, such as the number and size of hidden layers, activation functions, loss functions, and optimization algorithms. After defining the model and its parameters, users can use the functions provided by the h2o package to train the model and evaluate the performance of the model through methods such as cross-validation.

[0070] Ultimately, in this way, users can successfully obtain a deep learning regression model that can be used to predict and analyze various complex data relationships. This method is not only concise and efficient, but also takes advantage of the powerful functions of R software and h2o package, making the construction of deep learning regression models easier and more reliable.

[0071] After obtaining the diesel cetane number prediction model, the fitting coefficient R 2 To evaluate the predictive performance of the model. The specific evaluation criteria are as follows:

[0072] If R 2 ≤0.3, it indicates that the fit between the predicted value and the actual observed value is poor, and the prediction efficiency of the prediction model is low;

[0073] If 0.3<R 2 ≤0.4, it means that the fitting effect between the predicted value and the actual observed value is weak, and the prediction efficiency of the prediction model is also weak;

[0074] If 0.4<R 2 ≤0.6, it means that the fitting effect between the predicted value and the actual observed value is at a medium level, and the prediction efficiency of the prediction model is medium;

[0075] If 0.6<R 2≤1.0, it means that the fit between the predicted value and the actual observed value is strong, and the prediction efficiency of the prediction model is high.

[0076] The physical and chemical index data include: density at 20°C, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, clogging tendency, cold filter point, sulfur content, chlorine content, freezing point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, 20°C kinematic viscosity, cloud point, and total pollutant content.

[0077] For the second embodiment, please refer to Figure 4 , a method for predicting diesel cetane number, comprising:

[0078] Obtain the physical and chemical index data of the diesel sample to be tested;

[0079] The physical and chemical index data of the diesel sample to be tested are input into the diesel cetane number prediction model of the first embodiment to obtain the cetane number of the diesel sample to be tested.

[0080] The third embodiment is a training device for a diesel cetane number prediction model, comprising:

[0081] Data acquisition module, training data set is the core foundation of model construction, and its selection has a decisive influence on the output results of the model. In order to build an ideal prediction model, the training data set used in this embodiment mainly includes explanatory variables and their corresponding target variables. The explanatory variables cover the physical and chemical index data of different diesel samples, while the target variables include the cetane number of the corresponding diesel samples;

[0082] The model training module trains the deep learning regression model based on the provided training data set to obtain a model for predicting the diesel cetane number.

[0083] In order to more deeply understand the technical solution of the present invention, the following non-limiting examples are given:

[0084] To construct the training data set, the following steps were performed: first, diesel samples were collected and their physical and chemical properties were tested to obtain the physical and chemical index data of the samples; second, the cetane number corresponding to each sample was recorded. Based on these physical and chemical index data and the corresponding cetane number, a training data set for subsequent analysis was constructed.

[0085] The steps to obtain the cetane number of each diesel sample are as follows:

[0086] First, under the standard operating environment of the cetane number tester, the diesel sample to be tested is mixed with a series of standard fuels with known cetane numbers;

[0087] Subsequently, the ignition characteristics of the mixture were analyzed under the same experimental conditions;

[0088] Finally, the cetane number of the diesel sample to be tested is determined through comparative analysis.

[0089] In this embodiment, the method for obtaining the deep learning regression model involves using the h2o package in the R software to build a regression model. Specifically, the h2o package needs to be installed and loaded in the R environment. After the installation is completed, a deep learning regression model can be created by calling the relevant functions in the h2o package. This model can be used to handle various regression problems, such as predicting continuous numerical results. By setting appropriate parameters, such as the number of layers, the number of neurons, the activation function, etc., the performance of the model can be further optimized. Ultimately, this deep learning regression model established by the h2o package can be used to analyze and predict the target variable in the data set.

[0090] After obtaining the diesel cetane number prediction model, the fitting coefficient R 2 Measuring the predictive ability of the diesel cetane number prediction model; specifically including:

[0091] When R 2 When ≤0.3, the predicted value and the observed value have a poor fitting effect, and the prediction ability of the prediction model is poor;

[0092] When 0.3<R 2 When ≤0.4, the fitting effect between the predicted value and the observed value is weak, and the prediction ability of the prediction model is weak;

[0093] When 0.4<R 2 When ≤0.6, the fitting effect between the predicted value and the observed value is moderate, and the prediction ability of the prediction model is moderate;

[0094] When 0.6<R 2 When ≤1.0, the predicted value and the observed value have a strong fitting effect, and the prediction model has a strong prediction ability.

[0095] The physical and chemical index data include: density at 20°C, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, clogging tendency, cold filter point, sulfur content, chlorine content, freezing point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, 20°C kinematic viscosity, cloud point, and total pollutant content.

[0096] A fourth embodiment is a computer-readable storage medium storing a computer program, which implements the above-mentioned diesel cetane number prediction method when executed by a processor.

[0097] A fifth embodiment is a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method for predicting the diesel cetane number is implemented.

[0098] The following is a detailed description of the physical and chemical index data of the above 17 parameters as an example:

[0099] Take samples of diesel and conduct tests to obtain the physical and chemical index data and cetane number of the diesel samples; specifically:

[0100] First, obtain a number of diesel samples and obtain their physical and chemical index data, and calculate the cetane index through some physical and chemical index data of diesel. Then use a professional measuring machine to calculate the cetane number; the physical and chemical index data include 20℃ density, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, clogging tendency, cold filter point, sulfur content, chlorine content, freezing point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, kinematic viscosity (20℃), cloud point, and total pollutant content;

[0101] The basis for selecting the above 17 physical and chemical indicators in this embodiment includes:

[0102] The relationship between cetane number and composition: among various diesel component hydrocarbons with the same number of carbon atoms, normal alkanes have the highest cetane number; polycyclic aromatic hydrocarbons have the lowest cetane number; olefins and cycloalkanes are between alkanes and aromatic hydrocarbons. The higher the degree of isomerization of hydrocarbons and the more rings they have, the lower their cetane number; the cetane number of cycloalkanes and aromatic hydrocarbons increases with the increase of side chain length; and the cetane number decreases with the increase of side chain branches.

[0103] The relationship between the cetane index and the cetane number: the cetane number of the diesel fraction cut from the intermediate base crude oil is consistent with the cetane index; the cetane number of the diesel fraction cut from the naphthenic base crude oil is mostly smaller than the cetane index; the cetane number of the diesel fraction cut from the paraffin base crude oil is mostly greater than the cetane index. The cetane number of the straight-run diesel has the best correlation with the cetane index, followed by the hydrorefined diesel, and the worst is the hydrocracking diesel.

[0104] The relationship between cloud point, condensation point and cold filter point and cetane number: cloud point, condensation point and cold filter point are the main indicators for evaluating diesel low-temperature fluidity. The more cycloalkanes and aromatic hydrocarbons there are in diesel, the better the low-temperature fluidity is. The lower the test results of cloud point, condensation point and cold filter point, the more normal alkanes there are, and the worse the low-temperature fluidity is. Therefore, the low-temperature fluidity index of diesel should be negatively correlated with the cetane number without adding low-temperature flow improvers.

[0105] Relationship between corrected wear spot diameter and cetane number: Studies have shown that the content of polar components in diesel is inversely proportional to the content of saturated hydrocarbons, that is, the less polar components, the higher the content of saturated hydrocarbons, and the worse the lubricity. The higher the content of normal alkanes, the larger the value of the corrected wear spot diameter, that is, the corrected wear spot diameter is positively correlated with the cetane number.

[0106] The relationship between density and cetane number: When the content of PAHs and aromatic hydrocarbons in diesel is high, the density is high, while when the content of straight-chain alkanes is high, the density is low. Therefore, when the density is high, the cetane number is low; when the density is low, the cetane number is relatively high. The density and cetane number are negatively correlated.

[0107] Relationship between kinematic viscosity and cetane number: The kinematic viscosity of diesel is mainly affected by the molecular weight, chemical structure and type of hydrocarbon molecules. Generally speaking, hydrocarbons with larger molecular weight or more complex chemical structure will lead to increased viscosity of diesel. Among all hydrocarbons, alkanes have the smallest viscosity because their molecular structure is relatively simple and the intermolecular forces are relatively small. In contrast, cycloalkanes and aromatic hydrocarbons have stronger intermolecular forces and therefore have higher viscosities. In summary, as the viscosity of diesel increases, the alkane content decreases, the cycloalkanes and aromatic hydrocarbons content increases, and the cetane number decreases.

[0108] Relationship between flash point and cetane number: Flash point is related to the chemical composition and fraction composition of oil products. Generally speaking, oil products with more alkanes have higher flash points than oil products with the same viscosity but more cycloalkanes and aromatics. On the other hand, the higher the boiling point of the oil product, the heavier the fraction, and the larger the relative molecular mass, the higher its flash point. The larger the relative molecular mass of the oil product, the higher the aromatic content is generally than the alkane content, so the flash point and cetane number are mainly negatively correlated.

[0109] The relationship between total pollutants, clogging tendency and cetane number: The total pollutant content of diesel refers to the insoluble substances in the diesel sample, mainly including dust, particulate matter, etc. These pollutants are mainly contaminated during the production and transportation of diesel and caused by its own oxidation. The clogging tendency of diesel refers to the ability of diesel to easily clog pipes or filters under specific conditions. This characteristic is mainly related to factors such as impurities, moisture, and insoluble matter in diesel, which can cause diesel to clog pipes or filters during flow, affecting its normal use. As the depth of diesel hydrorefining increases, the content of various impurities in the oil will decrease, and the content of saturated hydrocarbons in diesel will increase relatively, which is manifested as an increase in the cetane number, and the total pollutant content and clogging tendency test results become smaller, so these two indicators are negatively correlated with the cetane number.

[0110] Relationship between acidity and cetane number: As the degree of diesel refinement increases, the polar substances in the oil decrease, and the lubricity of diesel becomes worse. At this time, acidic lubricity improvers are often added to improve the lubricity of diesel. In this case, the acidity of diesel will increase, resulting in a situation where the cetane number of diesel gradually increases with the increase of acidity.

[0111] The relationship between 50%, 90%, 95% recovery temperature and cetane number: The distillation range of diesel refers to the temperature range obtained by distillation under specific conditions. Diesel with a light distillation range has good starting performance, is easy to evaporate and burns quickly, but a too light distillation may result in a high autoignition point and a long ignition delay period, which may cause knocking. The 50% recovery temperature characterizes the content of light components in diesel. Within a certain temperature range, the concentration of the diesel and air mixture is moderate and easy to burn, which is manifested as a higher cetane number. The 90% and 95% recovery temperatures represent the content of heavy components in diesel. Diesel with a heavy distillation range will cause incomplete combustion due to slow evaporation, and the cetane number is poor, mainly showing a negative correlation.

[0112] Relationship between sulfur content and cetane number: The sulfides in diesel are mainly benzothiophene and alkylbenzothiophene and dibenzothiophene and alkyldibenzothiophene sulfides. Thiophene and its derivatives are mainly distributed in the fraction below 200℃, benzothiophene and its derivatives are mainly distributed in 200-300℃, and in the fraction above 300℃, the sulfides are mainly dibenzothiophene and its derivatives, among which 4,6-DMDBT is in the fraction of 330-345℃, and the more complex dibenzothiophene with substituents at the 4th and 6th positions is mainly concentrated in the fraction above 350℃. The temperature range of the straight-run diesel fraction is about 200-365℃, which concentrates most of the sulfur content in diesel, and its cetane number is relatively high, so as the sulfur content increases, the cetane number gradually increases.

[0113] The relationship between chlorine content and cetane number: The chlorine content in diesel may come from the crude oil itself, pollutants in the refining process, pollution in the transportation process, and the use of additives. By upgrading the refining process and strengthening management, the first three effects can be reduced as much as possible, so the chlorine content in diesel mainly comes from the additives used in the refining process. The cetane number of diesel can be improved through the refining process, so the chlorine content is positively correlated with the cetane number.

[0114] The relationship between oxidation stability and cetane number: The ability of various hydrocarbons in the fuel to resist oxidation in the liquid phase varies. Aromatic hydrocarbons, alkanes, and cycloalkanes have good stability. Unsaturated hydrocarbons easily react with oxygen in the air when in the liquid phase at room temperature, which is the main reason for fuel oxidation deterioration. In addition to unsaturated hydrocarbons, sulfur- or nitrogen-containing compounds can also cause fuel discoloration or deterioration, reducing fuel stability. The sulfide content in the straight-run components of diesel is relatively high, which is more likely to cause diesel oxidation deterioration. Therefore, under the premise that the oxidation stability of diesel meets the requirements of product standards, the straight-run component content is higher, the oxidation stability test result is larger, and the cetane number is also higher.

[0115] Construct a training data set, using physical and chemical index data as explanatory variables and cetane number as the target variable;

[0116] The data set includes explanatory variables and target variables; wherein the explanatory variables include physical and chemical index data of different diesel samples, and the target variables include cetane numbers of different diesel samples;

[0117] The data set of the diesel sample is randomly divided into a training set and a test set, the training set is used to establish the prediction model, and the test set is used to verify the prediction ability of the prediction model. For example, 90% of the data set can be used as a training set and 10% of the data set can be used as a test set.

[0118] Based on the training data set, a deep learning regression model is trained to obtain a diesel cetane number prediction model;

[0119] For example, the regression model is first established by using the h2o package in the R software, and then the deep learning regression model is trained based on the training data set to obtain a diesel cetane number prediction model;

[0120] The diesel cetane number prediction model is used to predict the cetane number of the diesel to be tested; specifically including:

[0121] First, the physical and chemical index data of the diesel sample to be tested are obtained, and the density at 20°C, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, clogging tendency, cold filter point, sulfur content, chlorine content, pour point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, kinematic viscosity (20°C), turbidity point, and total pollutant content are used as explanatory variables. Then, the explanatory variable data are substituted into the diesel cetane number prediction model, and the obtained target variable value is the predicted result of the cetane number of the diesel sample to be tested.

[0122] If no specific experimental steps or conditions are specified in the following examples, the conventional experimental steps or conditions described in the literature in the art can be used. The raw materials or instruments used are all conventional products that can be purchased commercially, including but not limited to the raw materials or instruments used in the examples of this application.

[0123] Implementation case, establishment of diesel cetane number prediction model

[0124] This embodiment provides a method for establishing a diesel cetane number prediction model, and the specific steps are as follows:

[0125] (1) Diesel sample collection

[0126] Select diesel samples that are stored in the production and operation process for testing. According to the product standard GB / T 19147-2016 and the internal control indicators of purchased diesel, carry out testing including all national standards and additional indicators (including odor, nitrate type cetane number improver, cloud point, clogging tendency, water separation performance, chlorine content, silicon content).

[0127] (2) Testing of physical and chemical indicators of diesel

[0128] The following physical and chemical indicators of diesel are tested: density at 20°C, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, clogging tendency, cold filter point, sulfur content, chlorine content, pour point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, kinematic viscosity at 20°C, cloud point, and total pollutant content.

[0129] The 17 diesel physical and chemical indicators used as explanatory variables include: density at 20°C, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, plugging tendency, cold filter point, sulfur content, chlorine content, freezing point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, kinematic viscosity at 20°C, cloud point and total pollutant content. The calculation formula of the cetane index CI is as follows:

[0130]

[0131] Where: T 10N =T 10 -215;

[0132] T 50N =T 50 -260;

[0133] T 90N =T 90 -310;

[0134] T 10 Indicates the 10% recovery temperature of the sample, in °C;

[0135] T 50 Indicates the 50% recovery temperature of the sample, in °C;

[0136] T 90 Indicates the 90% recovery temperature of the sample, in °C;

[0137] B=[exp(-0.0035D N )]-1;

[0138] D N =D-850;

[0139] D represents the density of the sample at 15°C, in kg / m 3 ;

[0140] Table 1 Detection methods of diesel physical and chemical index data

[0141]

[0142]

[0143] (3) Cetane number test

[0144] Under the specified operating environment, the combustion characteristics of the fuel to be tested are compared with those of a standard fuel with a known cetane number to determine the combustion characteristics of the fuel to be tested. The specific operation is to adjust the compression ratio of the engine (indicated by the handwheel reading) to determine the "ignition lag" of the fuel to be tested, that is, the time interval from the start of injection to the start of combustion (in units of crankshaft angle). According to the compression ratio of the fuel to be tested, two standard fuels with a cetane number difference of no more than 5.6 units are selected, and their "ignition lag" is determined by the same method. When the compression ratio of the fuel to be tested is between the compression ratios of the two selected standard fuels, the cetane number of the fuel to be tested is calculated by interpolation based on the handwheel reading.

[0145] (4) Using deep learning algorithms to build prediction models

[0146] In this embodiment, the AutoML algorithm provided by the H2O package of the R language is used for data analysis. The implementation of H2OAutoML is based on the JDK17 environment and is written in JAVA language. The h2o.automl function, which belongs to the H2O package, is used to perform the selection of the automatic machine learning algorithm. When determining the model parameter range, in order to avoid evaluation bias, all built-in models and their parameter combinations are selected for comprehensive search.

[0147] After the original data is imported in CSV format, a general data frame structure is formed. Subsequently, this data is converted into an H2O data frame and input into the H2OAutoML model. The data is segmented using the ratio function built into the H2O package to ensure the consistency of the model operation results and avoid data leakage. In order to evaluate the optimization effect of the model, this embodiment uses the five-fold cross-validation method that comes with the training model in the H2O package.

[0148] H2OAutoML has the ability to automatically select the optimal algorithm. The available algorithms include deep learning, random forest (RF), generalized linear model (GLM), gradient boosting machine (GBM) and XGBoost.

[0149] Finally, the best algorithm selected by H2O AutoML is h2o.deeplearning (deep learning algorithm). Then, the algorithm is used to build a preliminary regression model, and the model is used to test the R 2 The MSE and ∑ni=0.01 value are used to evaluate the generalization performance.

[0150] This embodiment involves a total of 7304 sample data, 17 oil physical and chemical parameters are selected as explanatory variables, and cetane number is used as a predictive variable. By comparing the prediction effects when the ratio of different data sets (i.e. the ratio of the number of samples in the training set to the test set) is 0.7, 0.8, 0.85, and 0.9, it is finally determined that the prediction performance is optimal when the ratio is 0.9. In the study, 30 random seeds were set for cyclic experiments, which were run 30 times in total. Each deep learning model is developed based on the training set and applied to the test set to test its generalization ability.

[0151] In this embodiment, the model is predicted using the training set, and the mean square error (MSE) and fitting coefficient (R 2 ) as an evaluation indicator. MSE is usually used to measure the performance of the prediction results on the same dimension indicator. When predicting the same indicator, it is generally believed that the lower the MSE value, the better the R 2 The higher the value, the more R 2 As a key parameter for evaluating model performance. 2 The different ranges of values ​​can be used to divide the degree of fit between the predicted value and the actual observed value into several levels: R 2 Less than 0.3 indicates a poor fit, between 0.3 and 0.4 indicates a weak fit, between 0.4 and 0.6 indicates a moderate fit, and between 0.6 and 1.0 indicates a strong fit. In addition, the h2o.deeplearning algorithm in the H2O package and the partitioning of the data set are implemented by the “h2o” package and the “h2o.deeplearning” function in the R software (version 4.3.3), respectively.

[0152] Testing of the prediction model

[0153] (1) Model training effect based on training set

[0154] Based on the training set, 17 diesel physical and chemical indicators (as explanatory variables) were used for combined prediction. The fitting effect of the predicted value of cetane number and the actual observed value is shown in the following figure. Figure 6 The comparison between the predicted cetane number data and the actual detected values ​​of some samples in the training set is shown in Table 2.

[0155] Table 2 Training effect of cetane number prediction model based on training set

[0156]

[0157]

[0158] As shown in Table 2 and Figure 6 As shown in the figure, the prediction results of the cetane number in the training set by the prediction model are highly consistent with the actual detection value, with a mean square error (MSE) of 0.019 and a determination coefficient (R2 ) reaches 0.990, showing a strong fitting effect.

[0159] (2) Model prediction effect based on the test set

[0160] The prediction model obtained from the training set is applied to the test set to test the generalization performance of the model, that is, its prediction ability for new data sets, and the mean square error (MSE) and determination coefficient (R 2 ) as the evaluation index. The results of the prediction of the test set using the cetane number prediction model are presented in Figure 6 The black scattered points in the figure and the predicted and measured cetane number data of some samples in the test set are shown in Table 3.

[0161] Table 3 Test results of the cetane number prediction model based on the test set

[0162]

[0163]

[0164] As shown in Table 3 and Figure 6 As shown in the figure, the prediction model still achieves a high fitting effect on the prediction of cetane number in the new data set (test set), and its correlation coefficient R 2 It reaches 0.992, while the mean square error (MSE) is 0.019.

[0165] In the prediction model of cetane number, the partial dependence relationship between the target variable and each explanatory variable is shown in Figure 7 .

[0166] Figure 7 The order of the variables in the equation reflects their importance, thus identifying the cetane index as a key feature for predicting the cetane number. The one-dimensional partial dependence plot between the cetane index and the cetane number reveals a linear relationship between the two, i.e., an increase in the cetane index is accompanied by a gradual increase in the cetane number. In addition, variables such as cloud point and sulfur content also show a positive correlation with the cetane number. However, variables such as 90% recovery temperature, cloud point, and cetane number show a negative correlation. In addition, variables such as 95% recovery temperature and condensation point have a nonlinear relationship with the cetane number.

[0167] It can be seen that the prediction model constructed by the method provided by the present invention has excellent generalization ability and can accurately predict the cetane number of unknown diesel samples. The prediction model constructed by the present invention adopts a wide range of data, covering diesel of different qualities, and can therefore be widely used in the accurate prediction of diesel cetane number.

[0168] Cetane number prediction

[0169] According to the previously constructed prediction model (which is based on the training set data), the process of predicting the cetane number of the diesel sample to be tested is as follows: First, collect the physical and chemical index data of the diesel sample to be tested. These physical and chemical indicators include density at 20°C, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, plugging tendency, cold filter point, sulfur content, chlorine content, pour point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, kinematic viscosity at 20°C, turbidity point and total pollutant content. Subsequently, these physical and chemical index data are input into the cetane number prediction model, and the target variable value calculated by the model is the cetane number prediction result of the diesel sample to be tested.

[0170] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When the use is implemented in whole or in part in the form of a computer program product, the computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL) or wireless (e.g., infrared, wireless, microwave, etc.) mode) to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk SolidState Disk (SSD)), etc.

[0171] The above is only a preferred embodiment of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A training method for a diesel cetane number prediction model, characterized in that: include: Acquire a training data set, wherein the training data set includes explanatory variables and corresponding target variables, wherein the explanatory variables include physical and chemical index data of different diesel samples, and the target variables include cetane numbers of different diesel samples; Based on the training data set, a deep learning regression model is trained to obtain a diesel cetane number prediction model.

2. The training method of the diesel cetane number prediction model according to claim 1, characterized in that: Obtaining the training data set includes: Sampling and testing diesel fuel to obtain the physical and chemical index data of diesel fuel samples and the cetane number corresponding to each diesel fuel sample; The training data set is constructed using the physical and chemical index data of diesel samples and the cetane number corresponding to each diesel sample.

3. The training method of the diesel cetane number prediction model according to claim 2, characterized in that: The steps for obtaining the cetane number of each diesel sample include: Under the standard operating conditions of the cetane number tester, the diesel sample to be tested is mixed with different standard fuels with known cetane numbers; The mixture was subjected to the fire property test under the same conditions; The cetane number of the diesel sample to be tested is obtained by comparison.

4. The training method of the diesel cetane number prediction model according to claim 1, characterized in that: The method for obtaining the deep learning regression model includes: establishing a regression model through the h2o package in R software.

5. The training method of the diesel cetane number prediction model according to claim 1, characterized in that: After obtaining the diesel cetane number prediction model, the fitting coefficient R 2 Measuring the predictive ability of the diesel cetane number prediction model; specifically including: When R 2 When ≤0.3, the predicted value and the observed value have a poor fitting effect, and the prediction ability of the prediction model is poor; When 0.3<R 2 When ≤0.4, the fitting effect between the predicted value and the observed value is weak, and the prediction ability of the prediction model is weak; When 0.4<R 2 When ≤0.6, the fitting effect between the predicted value and the observed value is moderate, and the prediction ability of the prediction model is moderate; When 0.6<R 2 When ≤1.0, the predicted value and the observed value have a strong fitting effect, and the prediction model has a strong prediction ability.

6. The method for training a diesel cetane number prediction model according to any one of claims 1 to 5, characterized in that: The physical and chemical index data include: density at 20°C, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, clogging tendency, cold filter point, sulfur content, chlorine content, freezing point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, 20°C kinematic viscosity, cloud point, and total pollutant content.

7. A method for predicting diesel cetane number, characterized in that: include: Obtain the physical and chemical index data of the diesel sample to be tested; The physical and chemical index data of the diesel sample to be tested is input into the diesel cetane number prediction model according to any one of claims 1 to 6 to obtain the cetane number of the diesel sample to be tested.

8. A training device for a diesel cetane number prediction model, characterized in that: include: A data acquisition module, which acquires a training data set, wherein the training data set includes explanatory variables and corresponding target variables, wherein the explanatory variables include physical and chemical index data of different diesel samples, and the target variables include cetane numbers of different diesel samples; The model training module trains a deep learning regression model based on the training data set to obtain a diesel cetane number prediction model.

9. The training device for the diesel cetane number prediction model according to claim 8, characterized in that: Obtaining the training data set includes: Sampling and testing diesel fuel to obtain the physical and chemical index data of diesel fuel samples and the cetane number corresponding to each diesel fuel sample; The training data set is constructed using the physical and chemical index data of diesel samples and the cetane number corresponding to each diesel sample; The steps for obtaining the cetane number of each diesel sample include: Under the standard operating conditions of the cetane number tester, the diesel sample to be tested is mixed with different standard fuels with known cetane numbers; The mixture was subjected to the fire property test under the same conditions; The cetane number of the diesel sample to be tested is obtained by comparison.

10. The training device for the diesel cetane number prediction model according to claim 8, characterized in that: The method for obtaining the deep learning regression model includes: establishing a regression model through the h2o package in R software.

11. The training device for the diesel cetane number prediction model according to claim 8, characterized in that: After obtaining the diesel cetane number prediction model, the fitting coefficient R 2 Measuring the predictive ability of the diesel cetane number prediction model; specifically including: When R 2 When ≤0.3, the predicted value and the observed value have a poor fitting effect, and the prediction ability of the prediction model is poor; When 0.3<R 2 When ≤0.4, the fitting effect between the predicted value and the observed value is weak, and the prediction ability of the prediction model is weak; When 0.4<R 2 When ≤0.6, the fitting effect between the predicted value and the observed value is moderate, and the prediction ability of the prediction model is moderate; When 0.6<R 2 When ≤1.0, the predicted value and the observed value have a strong fitting effect, and the prediction model has a strong prediction ability.

12. The training device for the diesel cetane number prediction model according to any one of claims 8 to 11, characterized in that: The physical and chemical index data include: density at 20°C, 50% recovery temperature, 90% recovery temperature, 95% recovery temperature, clogging tendency, cold filter point, sulfur content, chlorine content, freezing point, flash point, cetane index, acidity, corrected wear scar diameter, oxidation stability, 20°C kinematic viscosity, cloud point, and total pollutant content.

13. A computer program product, characterized in that The invention comprises a computer program, which realizes the diesel cetane number prediction method according to claim 7 when being executed by a processor.

14. A computer-readable storage medium storing a computer program, characterized in that: When the program is executed by a processor, the method for predicting the diesel cetane number as claimed in claim 7 is implemented.

Images