Method for measuring quality of oil product and measuring system

Abstract

The application provides a method for determining oil quality, comprising the following steps: obtaining oil quality sample data; extracting data features of the oil quality sample; establishing at least two weak classifiers according to the data features of the oil quality sample; determining a total classifier according to integration of the weak classifiers; obtaining a to-be-predicted sample of oil quality; and inputting the to-be-predicted sample of oil quality into the total classifier, performing classification analysis on the prediction data, so as to determine the oil quality classification of the to-be-predicted sample.

Description

Technical Field

[0001] This invention relates to the technical field of oil quality determination, and in particular to a method for determining oil quality, a system for determining oil quality, and a computer-readable storage medium. Background Technology

[0002] With the emergence and development of Industry 4.0, major factories have begun to leverage information technology for industrial transformation, leading to a continuous improvement in the level of intelligence in production processes. In the oil refining industry, a wide variety of sensors can accurately measure various properties of oil products. However, sensors cannot directly acquire deeper information about the oil, such as whether it can be used as asphalt feed, catalyst feed, slag feed, or coking feed. Therefore, a series of processing steps are required to obtain the corresponding oil quality information from the sensor data, thereby guiding the refining process and ensuring that the refined oil better meets production requirements.

[0003] Currently, mainstream methods often employ Support Vector Machines (SVMs) and Neural Networks (NNNs) for oil quality prediction. However, these methods have significant limitations. SVMs struggle with multi-class classification problems and are highly sensitive to parameter and kernel function selection, often resulting in low accuracy in practical applications. Neural Networks alone offer poor interpretability and are insufficient for guiding the refining process. Decision Trees, due to their excellent interpretability, stand out among various methods, allowing for the selection of different approaches based on the task requirements. However, single decision trees often exhibit poor generalization performance, making them difficult to apply in real-world scenarios. The reasons are as follows: First, a single decision tree only considers existing oil attributes and cannot utilize hidden features that aid in classification; second, a single decision tree has weak learning capabilities, is highly susceptible to specific data, and has limited processing power; finally, traditional decision trees cannot handle missing value data, thus rendering them unsuitable for practical industrial applications.

[0004] In order to overcome the above-mentioned defects in the existing technology, there is an urgent need in the field for a method for determining oil quality, which can make the model for predicting oil quality more capable of learning and thus more accurately predict oil quality. Summary of the Invention

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

[0006] Specifically, the method for determining oil quality according to the first aspect of the present invention includes the following steps: acquiring oil quality sample data; extracting data features of the oil quality samples; establishing at least two weak classifiers based on the data features of the oil quality samples; determining a total classifier based on the integration of the weak classifiers; acquiring samples of oil quality to be predicted; and inputting the samples of oil quality to be predicted into the total classifier to perform classification analysis on the prediction data to determine the oil quality classification of the samples to be predicted.

[0007] Furthermore, in some embodiments of the present invention, the extraction of data features from the oil quality samples includes the following steps: preprocessing the sample data; establishing a fully connected neural network based on the preprocessed sample data; and using the results of the hidden layers of the fully connected neural network as the data features.

[0008] Furthermore, in some embodiments of the present invention, the above-mentioned establishment of a fully connected neural network includes the following steps: based on the above-mentioned oil quality sample data, using attribute values ​​without missing values ​​as input and classification results as supervision, training the neural network to construct an initial fully connected neural network; determining the backpropagation gradient and loss value based on the binary classification cross-entropy loss function; and determining the parameters of the above-mentioned fully connected neural network in response to the convergence of the above-mentioned loss value.

[0009] Furthermore, in some embodiments of the present invention, after determining the parameters of the fully connected neural network, the determination method further includes the following steps: selecting multiple dimensions of the hidden layers for multiple training sessions to obtain multiple models that extract features of multiple different dimensions; and determining the optimal dimension of the hidden layer corresponding to the model through a subsequent module.

[0010] Furthermore, in some embodiments of the present invention, the above-mentioned method of establishing at least two weak classifiers based on the data features of the oil quality samples includes the following steps: concatenating the data features with first sample data containing missing values ​​in the sample data to obtain second sample data; and constructing at least two classification trees as the weak classifiers based on the second sample data.

[0011] Furthermore, in some embodiments of the present invention, the above-mentioned construction of at least two classification and regression trees as the weak classifier includes the following steps: using the squared error as the loss function of the classification and regression trees; determining the regularization term of the classification and regression trees according to the loss function; determining the gain of the classification and regression trees for splitting according to the regularization term; and determining the split point of the classification and regression trees and the features of the classification and regression trees according to the gain of the classification and regression trees for splitting and a greedy algorithm to construct the classification and regression trees as the weak classifier.

[0012] Furthermore, in some embodiments of the present invention, the determination of the total classifier based on the integration of the weak classifiers includes the following steps: weighting the results of the weak classifiers according to the oil quality sample data to obtain the result of the total classifier; and optimizing the hyperparameters of the total classifier based on the result of the total classifier to determine an optimized total classifier.

[0013] Furthermore, in some embodiments of the present invention, the hyperparameters are selected from one or more of the following: the maximum depth of the tree, the weight of the regularization term, the learning rate, and the dimension of the hidden layer (considered as a hyperparameter).

[0014] Furthermore, in some embodiments of the present invention, the oil quality classification result of the above-mentioned sample to be predicted is selected from one or more of the following: whether it can be used as asphalt feedstock, whether it can be used as asphalt feedstock catalyst, whether it can be used as asphalt slag feedstock, and whether it can be used as asphalt coking feedstock.

[0015] Furthermore, the oil quality determination system provided according to a second aspect of the present invention includes: a memory; and a processor connected to the memory and configured with the oil quality determination method.

[0016] Furthermore, according to a third aspect of the present invention, a computer-readable storage medium is provided thereon storing computer instructions, characterized in that, when the computer instructions are executed by a processor, the above-described method for determining the quality of oil is implemented. Attached Figure Description

[0017] The above-described features and advantages of the present invention will be better understood after reading the following detailed description of embodiments of the present disclosure in conjunction with the accompanying drawings. In the drawings, components are not necessarily drawn to scale, and components having similar related characteristics or features may have the same or similar reference numerals.

[0018] Figure 1 An architectural diagram of an oil quality determination system according to some embodiments of the present invention is shown;

[0019] Figure 2 A flowchart of a method for determining oil quality according to some embodiments of the present invention is shown;

[0020] Figure 3 A schematic flowchart of a method for determining oil quality according to some embodiments of the present invention is shown;

[0021] Figure 4 A schematic diagram of a regression tree visualization provided according to some embodiments of the present invention is shown. Detailed Implementation

[0022] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Although the description of the present invention is presented in conjunction with preferred embodiments, this does not mean that the features of the invention are limited to these embodiments. On the contrary, the purpose of describing the invention in conjunction with embodiments is to cover other options or modifications that may be derived based on the claims of the present invention. To provide a thorough understanding of the invention, many specific details will be included in the following description. The invention may also be implemented without using these details. Furthermore, to avoid confusion or obscuring the focus of the invention, some specific details will be omitted in the description.

[0023] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0024] Furthermore, the terms "upper," "lower," "left," "right," "top," "bottom," "horizontal," and "vertical" used in the following description should be understood as the orientations shown in the relevant paragraphs and accompanying drawings. These relative terms are for illustrative purposes only and do not imply that the described apparatus must be manufactured or operated in a specific orientation, and therefore should not be construed as limiting the invention.

[0025] It is understood that although terms such as "first," "second," and "third" may be used herein to describe various components, regions, layers, and / or parts, these components, regions, layers, and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers, and / or parts. Therefore, the first components, regions, layers, and / or parts discussed below may be referred to as second components, regions, layers, and / or parts without departing from some embodiments of the present invention.

[0026] As mentioned above, current mainstream methods often employ Support Vector Machines (SVMs) and Neural Networks (NNNs) for oil quality prediction. However, these methods have significant limitations. SVMs struggle with multi-class classification problems and are highly sensitive to parameter and kernel function selection, often resulting in low accuracy in practical applications. Neural Networks alone offer poor interpretability and are insufficient for guiding the refining process. Decision trees, due to their excellent interpretability, stand out among various methods, allowing for the selection of different approaches based on the task requirements. However, single decision trees often exhibit poor generalization performance, making them difficult to apply in real-world scenarios. The reasons are as follows: First, a single decision tree only considers existing oil attributes and cannot utilize hidden features that aid in classification; second, a single decision tree has weak learning capabilities, is highly susceptible to specific data, and has limited processing power; finally, traditional decision trees cannot handle missing value data, thus rendering them unsuitable for practical industrial applications.

[0027] To overcome the aforementioned deficiencies in the existing technology, the present invention provides a method for determining oil quality, a system for determining oil quality, and a corresponding computer-readable storage medium, which enables the model for predicting oil quality to have stronger learning ability, thereby enabling more accurate prediction of oil quality.

[0028] In some non-limiting embodiments, the oil quality determination method provided in the first aspect of the present invention can be implemented via the oil quality determination system provided in the second aspect of the present invention. Specifically, the oil quality determination system is equipped with a memory and a processor. The memory includes, but is not limited to, the computer-readable storage medium provided in the third aspect of the present invention, on which computer instructions are stored. The processor is connected to the memory and configured to execute the computer instructions stored in the memory to implement the oil quality determination method provided in the first aspect of the present invention.

[0029] Please refer to the following first. Figure 1 , Figure 1 A schematic diagram of an oil quality determination system according to some embodiments of the present invention is shown.

[0030] Figure 1An oil quality measuring system according to some embodiments of the present invention is illustrated. The oil quality measuring system includes an internal communication bus 101, a processor 102, a read-only memory (ROM) 103, a random access memory (RAM) 104, a communication port 105, and a hard disk 107. The internal communication bus 101 enables data communication between components of the oil quality measuring system. The processor 102 can perform judgments and issue prompts. In some embodiments, the processor 102 may consist of one or more processors. The communication port 105 enables data transmission and communication between the oil quality measuring system and external input / output devices. In some embodiments, the oil quality measuring system can send and receive information and data from a network through the communication port 105. In some embodiments, the oil quality measuring system can transmit and communicate with external input / output devices in a wired manner through input / output terminals 106. The oil quality determination system also includes various types of program storage units and data storage units, such as hard disk 107, read-only memory (ROM) 103, and random access memory (RAM) 104, capable of storing various data files used for computer processing and / or communication, as well as possible program instructions executed by processor 102. Processor 102 executes these instructions to implement the main part of the method. The results processed by processor 102 are transmitted to an external output device via communication port 105 and displayed on the user interface of the output device.

[0031] The working principle of the oil quality determination system described above will be described below with reference to some embodiments of oil quality determination methods. Those skilled in the art will understand that these embodiments of the methods are merely non-limiting implementations provided by the present invention, intended to clearly demonstrate the main concepts of the invention and provide specific solutions convenient for public implementation, rather than limiting all functions or operating methods of the oil quality determination system. Similarly, the oil quality determination system is also only one non-limiting implementation provided by the present invention and does not limit the entities performing the steps in these oil quality determination methods.

[0032] Please refer to the reference. Figure 2 as well as Figure 3 , Figure 2 A flowchart of a method for determining oil quality according to some embodiments of the present invention is shown. Figure 3 A schematic flowchart of a method for determining oil quality according to some embodiments of the present invention is shown.

[0033] like Figure 2 As shown in step S1, in the process of determining oil quality, the method for determining oil quality can first obtain oil quality sample data. Then, as... Figure 2As shown in step S2, the above method can extract the data features of the oil quality sample. Then, as... Figure 2 As shown in step S3, the above method can establish at least two weak classifiers based on the data characteristics of the oil quality samples. After establishing the weak classifiers, the determination method can determine the overall classifier based on the ensemble of the weak classifiers. After determining the overall classifier, the above method can obtain the oil quality samples to be predicted. Figure 3 As shown, after obtaining the oil quality sample to be predicted, the oil quality determination method can input the oil quality sample to be predicted into the total classifier, perform classification analysis on the prediction data, and determine the oil quality classification of the sample to be predicted.

[0034] Optionally, in some embodiments of the present invention, those skilled in the art can obtain sample data of the oil quality and the sample to be predicted by installing a sensor for measuring the oil quality sample.

[0035] Furthermore, the above-described measurement method can preprocess the sample data. After preprocessing, the measurement method can construct a fully connected neural network based on the preprocessed sample data. Then, the measurement method can use the results of the hidden layers of the fully connected neural network as the data features.

[0036] Furthermore, the establishment of the fully connected neural network in the above-mentioned measurement method may include the following steps: based on the oil quality sample data, using attribute values ​​without missing values ​​as input and classification results as supervision, the neural network is trained to construct an initial fully connected neural network. Then, the above-mentioned measurement method can determine the backpropagation gradient and loss value based on the binary cross-entropy loss function. After determining the backpropagation gradient and loss value, when the loss value converges, the above-mentioned measurement method can determine the parameters of the fully connected neural network.

[0037] Specifically, the above measurement method can perform minimum-maximum normalization on each feature of the sample, mapping the attribute values ​​to (0, 1), which facilitates subsequent classification and feature extraction. The formula is as follows:

[0038]

[0039] Where x min x represents the minimum value in this attribute. maxThis represents the maximum value of the attribute. The above measurement method can then select 70% of the samples in the crude oil dataset as the training set, using the attribute values ​​without missing values ​​as input, and the classification results as supervised training for the neural network. Here, for the missing values, the above measurement method can allocate all missing value data to one side of the tree and calculate the gain to determine the optimal splitting strategy. The above measurement method can also use the binary cross-entropy function as the loss function. To prevent leakage of classification information, the above measurement method can omit the test set when training the feature extraction model. This measurement method can observe loss convergence during training and save the model when the classification accuracy on the training set reaches a high level.

[0040] Furthermore, after determining the parameters of the fully connected neural network, the measurement method may further include the following steps: the measurement method may select multiple dimensions of the hidden layers for multiple training iterations to obtain multiple models that extract features from multiple different dimensions. After determining the models, the measurement method can determine the optimized fully connected neural network based on the accuracy of the classification results of each model.

[0041] Specifically, the above measurement method can randomly sample the dimensions of the hidden layers and train the sampled hidden layers multiple times. Alternatively, the method can select multiple dimensions of the hidden layers for multiple training iterations to obtain multiple models that extract features from various dimensions. Subsequently, these different dimensions of the hidden layers can be used as hyperparameters for training to determine the optimal dimensions.

[0042] Furthermore, the above-mentioned method of establishing at least two weak classifiers based on the data characteristics of oil quality samples may include the following steps: the above-mentioned measurement method may concatenate the data characteristics with first sample data containing missing values ​​to obtain second sample data. After obtaining the second sample data, the above-mentioned measurement method may construct at least two classification trees as the weak classifiers based on the second sample data.

[0043] Specifically, the above measurement method can use the results of the hidden layer as the extracted feature x. feature The new attributes of the oil are obtained by concatenating the features with the original attributes that have missing values, as shown in the following formula:

[0044] x = concatenate(x) normalization ,x feature )

[0045] Furthermore, constructing at least two classification trees as the weak classifier may include the following steps: The measurement method may use the squared error as the loss function of the classification tree. After determining the loss function, the measurement method may determine the regularization term of the classification tree based on the loss function. After determining the regularization term of the classification tree, the measurement method may determine the splitting gain of the classification tree based on the regularization term. Then, the measurement method may determine the splitting point of the classification tree and the features of the classification tree based on the splitting gain and a greedy algorithm to construct the classification tree as the weak classifier.

[0046] Specifically, the above measurement method discretizes the input raw attributes and auxiliary features: a feature is sorted from largest to smallest, and the mean of two adjacent values ​​is used as the split point. That is, when there are n different feature values, there are n-1 split points, and the optimal split point is selected based on the gain. The split point divides the sample into two groups, R1 and R2, each corresponding to a regression value c1 and c2. The squared error between the regression value and the supervised value is the loss function, as shown in the following formula:

[0047]

[0048] For positive samples, the supervision value is 1, and for negative samples, the supervision value is 0. For a k-class classification problem, the above method can construct k trees, building a regression tree for each class as a positive sample, ultimately obtaining k regression values ​​w. k Then, the above measurement method can be used to obtain the probability corresponding to each category by calculating the softmax function, and the category with the highest probability can be used as the predicted classification result.

[0049] Those skilled in the art will understand that the softmax function described above is merely a non-limiting implementation of the present invention, intended to normalize the corresponding output value to obtain the probability corresponding to each category, rather than being used to limit the scope of protection of the present invention.

[0050] Furthermore, the aforementioned overall classifier can be XGBOOST (Extreme Gradient Boosting Ensemble Decision Tree), and the final prediction result of the overall classifier is for each weak learner f. t The sum of the results is:

[0051]

[0052] Those skilled in the art will understand that the XGBOOST classifier described above is merely a non-limiting implementation of the present invention, intended to integrate the aforementioned multiple weak classifiers, and not intended to limit the scope of protection of the present invention.

[0053] Furthermore, after determining the overall classifier, to simplify the model and prevent overfitting, the above measurement method can incorporate a regularization term σ(f) describing the complexity of the decision tree into the loss function. t ),Right now:

[0054]

[0055] Assuming the previous decision trees are fixed, their complexity and predicted values ​​are also determined. The predicted value can be considered as the sum of the current decision tree's prediction and the predictions of the previous total decision trees. This can be obtained by expanding the loss function using a second-order Taylor series and simplifying it.

[0056]

[0057]

[0058]

[0059] Where constant represents the constant term, g i Represents the loss function pair The first derivative, h i Represents the loss function pair The second derivative of the equation means that the final optimization objective omits the constant term and the already defined terms. Expanding the regularization terms and combining them yields:

[0060]

[0061]

[0062]

[0063] Where T represents the number of leaf nodes in the spanning tree, w j Let G represent the value corresponding to the leaf node. The simplification of the above formula rewrites the summation of all previous sample sets as the summation of the leaf node sets, thus allowing it to be combined with a portion of the complexity calculation. Since the number of samples in each leaf node is fixed, G... j and H j Replacing the original summation term, the independent variable of the optimization objective is w. j Setting the first derivative of the objective function to 0, we obtain the value corresponding to leaf node j.

[0064]

[0065] The value of the objective function is simplified to

[0066]

[0067] According to the above formula, when constructing a tree structure, the above measurement method can determine whether a split node is necessary based on the gain. The gain of a split node can be expressed as the new score of splitting the left and right subtrees minus the score of the node without splitting, minus the increased complexity due to adding a new node. This can be expressed by the following formula:

[0068]

[0069] Furthermore, after determining the construction of the above-mentioned total classifier, the above-mentioned measurement method can optimize the hyperparameters of the total classifier based on the results of the total classifier to determine the optimized total classifier.

[0070] In some embodiments of the present invention, the hyperparameter optimization described above can be performed using the Optuna hyperparameter optimization framework to optimize the hyperparameters of the overall classifier, thereby finding the hyperparameters with the highest accuracy and determining the overall classifier. In this way, the above-described method for determining oil quality can enhance the learning ability of the model predicting oil quality, thus enabling more accurate prediction and classification of oil quality.

[0071] Preferably, the hyperparameters in the above determination method can be selected from one or more of the following: the maximum depth of the classification tree, the weight of the regularization term, the learning rate, and the dimension of the hidden layer (considered as a hyperparameter).

[0072] Those skilled in the art will understand that the above-described hyperparameter optimization framework Optuna is merely a non-limiting implementation of the present invention, intended to conveniently and efficiently optimize the hyperparameters of the above-described total classifier, and not intended to limit the scope of protection of the present invention.

[0073] In some embodiments of the present invention, the oil quality classification results of the above-mentioned sample to be predicted are selected from one or more of the following: whether it can be used as asphalt material, whether it can be used as asphalt material catalyst, whether it can be used as asphalt material residue feed, and whether it can be used as asphalt material coking feed.

[0074] Please refer to Figure 4 , Figure 4 A schematic diagram of a regression tree visualization provided according to some embodiments of the present invention is shown.

[0075] like Figure 4 As shown, once the above model is trained, those skilled in the art can intuitively obtain the influence of the characteristics of each oil quality on the final classification result of the oil quality based on the visualized regression tree.

[0076] Please refer to the table below, which shows that the model of the above measurement method has the highest accuracy under different algorithm models.

[0077]

[0078] Although the methods described above are illustrated and depicted as a series of actions for the sake of simplicity, it should be understood and appreciated that these methods are not limited by the order of the actions, as some actions may occur in a different order and / or concurrently with other actions from the illustrations and descriptions herein or not illustrated and described herein but which may be understood by those skilled in the art, according to one or more embodiments.

[0079] Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or a combination of both. To clearly illustrate this interchangeability between hardware and software, the various illustrative components, blocks, modules, circuits, and steps are described above in a generalized manner in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. Those skilled in the art may implement the described functionality in different ways for each specific application, but such implementation decisions should not be construed as departing from the scope of the invention.

[0080] Although the oil quality measurement system described in the above embodiments can be implemented through a combination of software and hardware, it is understood that the oil quality measurement system can also be implemented in software or hardware. For hardware implementation, the oil quality measurement system can be implemented using one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, other electronic devices for performing the above functions, or a selection of combinations of the above devices. For software implementation, the oil quality measurement system can be implemented using independent software modules such as procedures and functions running on a general-purpose chip, each module performing one or more functions and operations described herein.

[0081] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of both. The software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor such that the processor can read and write information to / from the storage medium. In an alternative, the storage medium may be integrated into the processor. The processor and storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In an alternative, the processor and storage medium may reside as discrete components in the user terminal.

[0082] In one or more exemplary embodiments, the described functionality may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software as a computer program product, the functionality may be stored or transmitted as one or more instructions or code on or through a computer-readable medium. A computer-readable medium includes both computer storage media and communication media, encompassing any medium that facilitates the transfer of a computer program from one location to another. A storage medium may be any available medium accessible to a computer. By way of example and not limitation, such a computer-readable medium may include RAM, ROM, EEPROM, CD-ROM or other optical disc storage, disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and is accessible to a computer. Any connection is also legitimately referred to as a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of a medium. As used in this article, disk and disc include compact discs (CDs), laser discs, optical discs, digital multi-purpose discs (DVDs), floppy disks, and Blu-ray discs. Disks typically reproduce data magnetically, while discs reproduce data optically using lasers. Combinations of these should also be included within the scope of computer-readable media.

[0083] The prior description of this disclosure is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not intended to be limited to the examples and designs described herein, but should be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for determining the quality of oil, characterized in that, Includes the following steps: Obtain oil quality sample data; The sample data is preprocessed. Based on the preprocessed sample data, a fully connected neural network is established; Based on the oil quality sample data, the attribute values ​​with no missing values ​​are used as input, and the classification results are used as supervision to train the neural network to construct the initial fully connected neural network. Based on the binary cross-entropy loss function, determine the backpropagation gradient and the loss value; In response to the convergence of the loss value, the parameters of the fully connected neural network are determined; The results of the hidden layers of the fully connected neural network are used as data features; The data features are concatenated with the first sample data containing missing values ​​to obtain the second sample data; Based on the second sample data, construct at least two classification trees as weak classifiers; The overall classifier is determined based on the integration of the weak classifiers; Obtain samples of oil quality to be predicted; as well as The oil quality sample to be predicted is input into the overall classifier, and the prediction data is classified and analyzed to determine the oil quality classification of the sample to be predicted.

2. The determination method according to claim 1, characterized in that, After determining the parameters of the fully connected neural network, the determination method further includes the following steps: Multiple dimensions of the hidden layers are selected for multiple training iterations to obtain multiple models that extract features from various dimensions; and The optimal dimension of the hidden layer corresponding to the model is determined through subsequent modules.

3. The determination method according to claim 1, characterized in that, The construction of at least two classification trees as the weak classifier includes the following steps: The squared error is used as the loss function for the classification tree; Based on the loss function, determine the regularization term of the classification tree; Based on the regularization term, determine the gain of the classification tree split; and The split points of the classification tree and the features of the classification tree are determined based on the gain of the classification tree split and a greedy algorithm to construct the classification tree as the weak classifier.

4. The determination method according to claim 3, characterized in that, The process of determining the overall classifier based on the ensemble of the weak classifiers includes the following steps: Based on the oil quality sample data, the results of the weak classifiers are weighted and summed to obtain the result of the overall classifier; and Based on the results of the total classifier, the hyperparameters of the total classifier are optimized to determine the optimized total classifier.

5. The determination method according to claim 4, characterized in that, The hyperparameters are selected from one or more of the following: the maximum depth of the tree, the weight of the regularization term, the learning rate, and the dimension of the hidden layer.

6. The determination method according to claim 1, characterized in that, The oil quality classification result of the sample to be predicted is selected from one or more of the following: whether it can be used as asphalt feed, whether it can be used as asphalt catalyst, whether it can be used as asphalt residue feed, and whether it can be used as asphalt coking feed.

7. A system for determining the quality of oil, characterized in that, include: Memory; as well as A processor, connected to the memory, and configured to implement the method for determining the quality of oil as described in any one of claims 1 to 6.

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

Images