Oil dry point prediction calculation method, device, computer equipment and storage medium
Through the combination of neural network model and proxy model, the problem of inaccurate oil dry point calculation in the existing technology is solved, accurate prediction of oil dry point value is achieved, and the accuracy of the online optimization system is improved.
Patent Information
- Application Number
- CN202111402467.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-11-19
- Publication Date
- 2025-04-18
- Estimated Expiration
- 2041-11-19
AI Technical Summary
The existing commercial modeling software in the field of online optimization (RTO) of continuous reforming devices generally has inaccurate calculations in the calculation of dry oil points, resulting in a large gap between the calculated value of dry oil points of reforming and the real value, and loses the meaning of reference and prediction.
A neural network algorithm is used to generate a neural network model with at least one hidden layer, and transform the component weights and intercepts of the hidden layer through a preset matrix to generate a proxy model. Then, the calculation results of the proxy model are input into the modeling program for packaging, and a prediction module for predicting dry points is developed, and the prediction value of the dry points is obtained through this module.
This method realizes accurate calculation and prediction of oil dry point values, reducing the error with the real value, and improving the accuracy of the online optimization system.
Smart Images
Figure CN114300068B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of oil refining, and in particular to an oil dry point prediction calculation method, device, computer equipment and storage medium. Background Art
[0002] The "dry point" is when the oil product reaches the highest vapor phase temperature at the end of distillation. The "dry point" properties of the device logistics products are usually obtained through field tests, while the online optimization system obtains them through physical property calculations within the software.
[0003] At present, commercial modeling software in the field of continuous reforming unit online optimization (RTO) generally has the limitation of inaccurate calculation in the calculation of "dry point". The main reason is that the properties of isomers vary greatly. In particular, after complex fractionation calculations, many physical property matching results of process simulation software are not ideal. Taking the simulation of the reforming unit as an example, the dry point of the reforming oil is generally far from the actual measured value, and the error level has lost the meaning of reference and prediction. However, in the process of online optimization, relatively accurate dry point values are required, so it is necessary to improve the defects of the current process simulation software. Summary of the invention
[0004] Based on this, it is necessary to provide an oil dry point prediction calculation method, device, computer equipment and storage medium to address the above technical problems.
[0005] A method for predicting and calculating an oil dry point, comprising:
[0006] Get sample data;
[0007] Inputting the sample data into a preset neural network algorithm to generate a neural network model having at least one hidden layer;
[0008] Transforming each hidden layer of the neural network model using a preset matrix to generate a proxy model;
[0009] Inputting the calculation results of the proxy model into a preset modeling program for packaging to obtain a prediction module for predicting dry points;
[0010] The preset component values in the preset modeling program are input into the prediction module for calculation to obtain the dry point prediction value.
[0011] In one embodiment, the step of transforming each hidden layer of the neural network model using a preset matrix to generate a proxy model includes:
[0012] Obtaining the weights and intercepts of each component of each hidden layer of the neural network model;
[0013] The weights and intercepts of the components of each hidden layer are transformed using a preset matrix to generate the proxy model.
[0014] In one embodiment, the step of transforming the weights and intercepts of the components of each hidden layer using a preset matrix to generate the proxy model includes:
[0015] Using the first matrix, transforming the weights of each component of each hidden layer to obtain the transformed first matrix;
[0016] Using a second matrix, transforming the intercepts of each component of each hidden layer and the transformed first matrix to obtain the transformed second matrix;
[0017] The transformed second matrix is transformed using a third matrix to generate the proxy model, wherein the third matrix includes a logistic regression model classification function.
[0018] In one of the embodiments, the weights and intercepts of the components of each hidden layer are transformed using a preset matrix, and in the step of generating the proxy model, the first matrix, the second matrix and the third matrix are sequentially used to perform cyclic iterative transformations twice to generate the proxy model.
[0019] In one embodiment, the logistic regression model classification function is a sigmoid function.
[0020] In one embodiment, the number of hidden layers of the neural network model is two.
[0021] An oil dry point prediction and calculation device, comprising:
[0022] A sample data acquisition unit, used for acquiring sample data;
[0023] A neural network model generating unit, used for inputting the sample data into a preset neural network algorithm to generate a neural network model having at least one hidden layer;
[0024] A proxy model generating unit, used to transform each hidden layer of the neural network model using a preset matrix to generate a proxy model;
[0025] A prediction module obtaining unit, used for inputting the calculation result of the proxy model into a preset modeling program for packaging, and obtaining a prediction module for predicting dry points;
[0026] The dry point prediction value obtaining unit is used to input the preset component value in the preset modeling program into the prediction module for calculation to obtain the dry point prediction value.
[0027] In one embodiment, the proxy model generation unit includes:
[0028] A weight and intercept acquisition subunit, used to acquire the weight and intercept of each component of each hidden layer of the neural network model;
[0029] The proxy model generation subunit is used to transform the weights and intercepts of each component of each hidden layer using a preset matrix to generate the proxy model.
[0030] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the following steps when executing the computer program:
[0031] Get sample data;
[0032] Inputting the sample data into a preset neural network algorithm to generate a neural network model having at least one hidden layer;
[0033] Transforming each hidden layer of the neural network model using a preset matrix to generate a proxy model;
[0034] Inputting the calculation results of the proxy model into a preset modeling program for packaging to obtain a prediction module for predicting dry points;
[0035] The preset component values in the preset modeling program are input into the prediction module for calculation to obtain the dry point prediction value.
[0036] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:
[0037] Get sample data;
[0038] Inputting the sample data into a preset neural network algorithm to generate a neural network model having at least one hidden layer;
[0039] Transforming each hidden layer of the neural network model using a preset matrix to generate a proxy model;
[0040] Inputting the calculation results of the proxy model into a preset modeling program for packaging to obtain a prediction module for predicting dry points;
[0041] The preset component values in the preset modeling program are input into the prediction module for calculation to obtain the dry point prediction value.
[0042] The above-mentioned oil dry point prediction calculation method, device, computer equipment and storage medium use a large amount of sample data combined with a neural network algorithm to generate a neural network model, and then develop a corresponding proxy model based on the parameters generated by the neural network model, and develop a corresponding online optimization software module based on the proxy model, thereby realizing accurate calculation and prediction of the dry point value. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] Figure 1 A schematic diagram of a flow chart of a method for calculating an oil dry point prediction in one embodiment;
[0044] Figure 2 It is a structural block diagram of an oil dry point prediction and calculation device in one embodiment;
[0045] Figure 3 is an internal structure diagram of a computer device in one embodiment;
[0046] Figure 4 It is a flowchart of an oil dry point prediction calculation method in another embodiment;
[0047] Figure 5 is a schematic diagram of a neural network model in one embodiment;
[0048] Figure 6 It is a schematic diagram of the appearance interface of the prediction module and configuration icons;
[0049] Figure 7 It is a schematic diagram of the prediction module and configuration appearance interface;
[0050] Figure 8 This is a schematic diagram of the sample minimum value range statistics interface in the internal parameter interface of the prediction module;
[0051] Fig. 9 This is a schematic diagram of the internal parameter interface of the prediction module - the sample maximum value range statistics interface;
[0052] Fig.10 This is a schematic diagram of the internal parameter interface of the prediction module - the fine-tuning parameter interface. DETAILED DESCRIPTION
[0053] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0054] Embodiment 1
[0055] In this embodiment, Figure 1 As shown, a method for predicting and calculating an oil dry point is provided, which includes:
[0056] Step 110, obtaining sample data.
[0057] In this embodiment, the sample data is LIMS data, which is used as a sample of the neural network model. In this embodiment, the LIMS data is first screened to select suitable data, and then the screened data is sorted to obtain the sample data.
[0058] Step 120: input the sample data into a preset neural network algorithm to generate a neural network model having at least one hidden layer.
[0059] In this step, a neural network algorithm is applied according to the sample to generate a neural network model. In this embodiment, the sample data is the components of oil and gas, the neural network model inputs the components, and the output is the dry point temperature. The specific calculation process of the neural network model is to input the measured values of all corresponding 16 components at the same time node at the input end, and then the output end calculates the dry point temperature value of the corresponding time node.
[0060] Step 130, transform each hidden layer of the neural network model using a preset matrix to generate a proxy model.
[0061] In this embodiment, the parameters of each hidden layer of the neural network model are iteratively transformed using a preset matrix to generate a proxy model.
[0062] Step 140: Input the calculation result of the proxy model into a preset modeling program for packaging to obtain a prediction module for predicting dry points.
[0063] In this embodiment, the preset modeling program is a preset modeling software, which can be an existing preset modeling software, such as RoMeo software, or other preset modeling program software. In this embodiment, the result of the proxy model calculation is encapsulated into RoMeo software to make a prediction module for predicting dry points.
[0064] Step 150: input the preset component values in the preset modeling program into the prediction module for calculation to obtain the dry point prediction value.
[0065] In this embodiment, the preset component value is the component value calculated at the bottom of the RoMeo reforming oil fractionation tower. Specifically, when the prediction module is trained, the field data is used as input, and the field data is input into the prediction module to generate hidden layer weights and other parameters; when the optimization system is used, the prediction module input is the component value calculated at the bottom of the RoMeo reforming oil fractionation tower, that is, the component value calculated at the bottom of the RoMeo reforming oil fractionation tower is input into the prediction module, and the output is the corresponding prediction value. In this way, the big data model and the mechanism model can be unified.
[0066] In one embodiment, the step of transforming each hidden layer of the neural network model using a preset matrix to generate a proxy model includes: obtaining the weights and intercepts of each component of each hidden layer of the neural network model; transforming the weights and intercepts of each component of each hidden layer using a preset matrix to generate the proxy model. In one embodiment, the number of hidden layers of the neural network model is two.
[0067] In this embodiment, the neural network model includes two hidden layers, so the weights and intercepts of the components corresponding to each hidden layer can be obtained in sequence. Using a preset matrix, the weights and intercepts of the components of the two hidden layers are input into the preset matrix for calculation, thereby obtaining the proxy model.
[0068] In one embodiment, the weights and intercepts of each component of each hidden layer are transformed using a preset matrix, and the step of generating the proxy model includes: using a first matrix to transform the weights of each component of each hidden layer to obtain the transformed first matrix; using a second matrix to transform the intercepts of each component of each hidden layer and the transformed first matrix to obtain the transformed second matrix; using a third matrix to transform the transformed second matrix to generate the proxy model, wherein the third matrix includes a logistic regression model classification function.
[0069] In this embodiment, the generation rule of the proxy model is to iteratively transform the two hidden layers of the neural network model using the first matrix, the second matrix and the third matrix:
[0070] Among them, the first matrix is matrix 1, the second matrix is matrix 2, and the third matrix is matrix 3.
[0071]
[0072]
[0073] Matrix 3 = F (matrix 2), thus, by exchanging matrix 1, matrix 2 and matrix 3, a proxy model is generated.
[0074] In one embodiment, the logistic regression model classification function is a sigmoid function.
[0075] In this embodiment, in matrix 3 = F (matrix 2), F is a sigmoid function that can be used in classification problems in machine learning, for example, a logistic regression model classifier maps the output value to (0, 1); the sum is 1, which can be understood from a probability perspective, and the largest value (maximum probability) is selected as the predicted value when outputting. That is, it is transformed into:
[0076]
[0077] In one embodiment, the weights and intercepts of the components of each hidden layer are transformed using a preset matrix. In the step of generating the proxy model, the first matrix, the second matrix and the third matrix are sequentially used to perform cyclic iterative transformations twice to generate the proxy model.
[0078] In this embodiment, matrix 1, matrix 2 and matrix 3 are iterated twice in sequence, and the corresponding dry point value can be obtained after the final iteration.
[0079] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.
[0080] Embodiment 2
[0081] It should be understood that online optimization is the most advanced optimization technology in the current development of full-process optimization control technology. It applies optimization technology to process control, seeks a set of operating parameters that makes the objective function reach the optimal value while meeting various production technical indicators, and uses this set of parameters for the actual control of the device.
[0082] Online optimization technology can put the production process in the best operating state by adjusting operating parameters without modifying the process flow or adding production equipment. Online optimization technology can set the maximum benefit and high value-added yield as the objective function. Without modifying the process flow or adding or reducing production equipment, it can put the production process in the best operating state and obtain the highest expected value by adjusting parameters such as pressure, temperature, and load. In other words, without increasing major equipment investment, the operating potential of existing production equipment is fully utilized, and the main technical and economic indicators have reached or exceeded the international advanced level of similar equipment, effectively achieving the goals of increasing production, saving energy, and reducing consumption, creating new profits and taxes for the enterprise.
[0083] The core of supporting the above functions is that the online optimization system is based on the calculation and optimization results of a strict mechanism model. Currently, there are only two platforms of mechanism model software that can support online optimization: ROMEO and ASPEN PLUS.
[0084] Taking ROMEO, the online optimization software for existing reforming units, as an example, its model is assembled and configured by several submodules, which are then uniformly configured and debugged by professional technicians in combination with reforming process knowledge.
[0085] In view of the fact that it is difficult to accurately predict the properties of the "dry point" of the flow stream using the mechanism method at present, the reasons and current situation have been stated above, therefore, this application intends to use other methods to predict the "dry point".
[0086] However, current research is limited to using neural network or big data methods to generate neural network models, and using a single model to perform simple predictions of single variables. However, it is impossible to combine it with process simulation software to perform overall budgeting of multiple parameters at the device level, because it is impossible to achieve the effect of device-level optimization.
[0087] This application uses big data combined with a neural network algorithm to generate a model, and then develops a corresponding proxy model based on the parameters generated by the neural network model. The proxy model uses the parameters of the neural network model. Finally, the corresponding online optimization software module is developed based on the proxy model. Figure 4 As shown:
[0088] Step 1: Screen and organize data based on the LIMS data from the site for one and a half years to generate neural network samples.
[0089] Step 2: Apply the neural network algorithm based on the sample to generate a neural network model.
[0090] The input of this neural network model is the component and the output is the dry point temperature.
[0091] The steps for using the neural network model are to input the measurement values of all corresponding 16 components at the same time node at the input end, and then the output end calculates the dry point temperature value of the corresponding time node.
[0092] The neural network model calculates an accuracy rate of 99.9% for the test sample, and the neural network model predicts an accuracy rate of 96% for the new sample.
[0093] Step 3: Generate a proxy model.
[0094] Depend on Figure 5 It can be seen that the neural network model has two hidden layers. The weights and intercepts of the corresponding components of each hidden layer can be obtained sequentially.
[0095] The generation rule of the proxy model is to iteratively use the transformation from matrix 1 to matrix 3 for the two hidden layers.
[0096]
[0097]
[0098] Matrix 3 = F (matrix 2) F is the sigmoid function, which can be used in classification problems in machine learning, such as the logistic regression model classifier, which maps the output value to (0,1); the sum is 1, which can be understood from a probability perspective, so the largest value (maximum probability) is selected as the predicted value when outputting.
[0099] That is the transformation:
[0100]
[0101] The steps of matrix 1->matrix 3 are iterated twice. After the final iteration, the corresponding dry point value can be obtained.
[0102] The results of the proxy model calculation are encapsulated into the RoMeo software to make a prediction module for predicting dry point. When the module is trained, the field data is used as input to generate hidden layer weights and other parameters; when the optimization system is used, the module input is the component value calculated at the bottom of the RoMeo reforming oil fractionation tower, and the output is the predicted value at this time. In this way, the big data model and the mechanism model can be unified. The prediction module obtained is as follows Figures 6 to 9 shown.
[0103] in, Figure 6 It is a schematic diagram of the appearance interface of the prediction module and configuration icons;
[0104] Figure 7 It is a schematic diagram of the prediction module and configuration appearance interface;
[0105] Figure 8 This is a schematic diagram of the sample minimum value range statistics interface in the internal parameter interface of the prediction module;
[0106] Fig. 9 This is a schematic diagram of the internal parameter interface of the prediction module - the sample maximum value range statistics interface;
[0107] Fig.10 This is a schematic diagram of the internal parameter interface of the prediction module - the fine-tuning parameter interface.
[0108] The purpose of fine-tuning parameters is that this module is deployed in the industrial control network, while the original LIMS data is in the management network. Therefore, it is not realistic to update the hidden layer parameters in real time. Therefore, when there is a large gap between the actual situation on site and the original parameters of the neural network, and the customer does not want to update all parameters immediately, the above parameters can be manually fine-tuned to eliminate the errors between the dry point calculation values and the on-site values.
[0109] Embodiment 3
[0110] In this embodiment, Figure 2 As shown, a device for predicting and calculating an oil dry point is provided, comprising:
[0111] A sample data acquisition unit, used for acquiring sample data;
[0112] A neural network model generating unit, used for inputting the sample data into a preset neural network algorithm to generate a neural network model having at least one hidden layer;
[0113] A proxy model generating unit, used to transform each hidden layer of the neural network model using a preset matrix to generate a proxy model;
[0114] A prediction module obtaining unit, used for inputting the calculation result of the proxy model into a preset modeling program for packaging, and obtaining a prediction module for predicting dry points;
[0115] The dry point prediction value obtaining unit is used to input the preset component value in the preset modeling program into the prediction module for calculation to obtain the dry point prediction value.
[0116] In one embodiment, the proxy model generation unit includes:
[0117] A weight and intercept acquisition subunit, used to acquire the weight and intercept of each component of each hidden layer of the neural network model;
[0118] The proxy model generation subunit is used to transform the weights and intercepts of each component of each hidden layer using a preset matrix to generate the proxy model.
[0119] In one embodiment, the proxy model generation subunit is also used to use a first matrix to transform the weights of each component of each hidden layer to obtain the transformed first matrix; use a second matrix to transform the intercepts of each component of each hidden layer and the transformed first matrix to obtain the transformed second matrix; use a third matrix to transform the transformed second matrix to generate the proxy model, wherein the third matrix includes a logistic regression model classification function.
[0120] The proxy model generation subunit is further used to sequentially use the first matrix, the second matrix and the third matrix to perform cyclic iterative transformation twice to generate the proxy model.
[0121] In one embodiment, the logistic regression model classification function is a sigmoid function.
[0122] In one embodiment, the number of hidden layers of the neural network model is two.
[0123] The specific definition of the oil dry point prediction calculation device can refer to the definition of the oil dry point prediction calculation method above, which will not be repeated here. Each unit in the above oil dry point prediction calculation device can be implemented in whole or in part by software, hardware and their combination. The above units can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above units.
[0124] Embodiment 4
[0125] In this embodiment, a computer device is provided. Its internal structure diagram can be shown as follows: Figure 3 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program, and the non-volatile storage medium is deployed with a database, which is used to store sample data and a neural network model. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with other computer devices deployed with application software. When the computer program is executed by the processor, a method for predicting and calculating an oil dry point is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse, etc.
[0126] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0127] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:
[0128] Get sample data;
[0129] Inputting the sample data into a preset neural network algorithm to generate a neural network model having at least one hidden layer;
[0130] Transforming each hidden layer of the neural network model using a preset matrix to generate a proxy model;
[0131] Inputting the calculation results of the proxy model into a preset modeling program for packaging to obtain a prediction module for predicting dry points;
[0132] The preset component values in the preset modeling program are input into the prediction module for calculation to obtain the dry point prediction value.
[0133] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:
[0134] Obtaining the weights and intercepts of each component of each hidden layer of the neural network model;
[0135] The weights and intercepts of the components of each hidden layer are transformed using a preset matrix to generate the proxy model.
[0136] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:
[0137] Using the first matrix, transforming the weights of each component of each hidden layer to obtain the transformed first matrix;
[0138] Using a second matrix, transforming the intercepts of each component of each hidden layer and the transformed first matrix to obtain the transformed second matrix;
[0139] The transformed second matrix is transformed using a third matrix to generate the proxy model, wherein the third matrix includes a logistic regression model classification function.
[0140] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:
[0141] The first matrix, the second matrix and the third matrix are sequentially used to perform cyclic iterative transformation twice to generate the proxy model.
[0142] In one embodiment, the logistic regression model classification function is a sigmoid function.
[0143] In one embodiment, the number of hidden layers of the neural network model is two.
[0144] Embodiment 5
[0145] In this embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:
[0146] Get sample data;
[0147] Inputting the sample data into a preset neural network algorithm to generate a neural network model having at least one hidden layer;
[0148] Transforming each hidden layer of the neural network model using a preset matrix to generate a proxy model;
[0149] Inputting the calculation results of the proxy model into a preset modeling program for packaging to obtain a prediction module for predicting dry points;
[0150] The preset component values in the preset modeling program are input into the prediction module for calculation to obtain the dry point prediction value.
[0151] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0152] Obtaining the weights and intercepts of each component of each hidden layer of the neural network model;
[0153] The weights and intercepts of the components of each hidden layer are transformed using a preset matrix to generate the proxy model.
[0154] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0155] Using the first matrix, transforming the weights of each component of each hidden layer to obtain the transformed first matrix;
[0156] Using a second matrix, transforming the intercepts of each component of each hidden layer and the transformed first matrix to obtain the transformed second matrix;
[0157] The transformed second matrix is transformed using a third matrix to generate the proxy model, wherein the third matrix includes a logistic regression model classification function.
[0158] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:
[0159] The first matrix, the second matrix and the third matrix are sequentially used to perform cyclic iterative transformation twice to generate the proxy model.
[0160] In one embodiment, the logistic regression model classification function is a sigmoid function.
[0161] In one embodiment, the number of hidden layers of the neural network model is two.
[0162] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0163] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0164] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.
Claims
1. A method for predicting and calculating oil dry point, characterized in that: include: Acquiring sample data, wherein the sample data is components of oil and gas; Inputting the sample data into a preset neural network algorithm to generate a neural network model having at least one hidden layer, wherein the input of the neural network model is the composition of oil and gas, and the output of the neural network model is the dry point temperature; Transforming each hidden layer of the neural network model using a preset matrix to generate a proxy model; Inputting the calculation results of the proxy model into a preset modeling program for packaging to obtain a prediction module for predicting dry points; Inputting the preset component values in the preset modeling program into the prediction module for calculation to obtain the dry point prediction value; The step of transforming each hidden layer of the neural network model using a preset matrix to generate a proxy model comprises: Obtaining the weights and intercepts of each component of each hidden layer of the neural network model; The weights and intercepts of each component of each hidden layer are iteratively transformed using a preset matrix to generate the proxy model, including: Using the first matrix, transforming the weights of each component of each hidden layer to obtain the transformed first matrix; Using a second matrix, transforming the intercepts of each component of each hidden layer and the transformed first matrix to obtain the transformed second matrix; Using a third matrix to transform the transformed second matrix to generate the proxy model, wherein the third matrix includes a logistic regression model classification function; The first matrix is matrix1, the second matrix is matrix2, and the third matrix is matrix3. Matrix 3 = F(Matrix 2) F is the sigmoid function.
2. The method according to claim 1, characterized in that The weights and intercepts of each component of each hidden layer are transformed using a preset matrix. In the step of generating the proxy model, the first matrix, the second matrix and the third matrix are sequentially used to perform cyclic iterative transformation twice to generate the proxy model.
3. The method according to claim 1 or 2, characterized in that: The number of hidden layers of the neural network model is two.
4. An oil dry point prediction and calculation device, characterized in that: include: A sample data acquisition unit, used to acquire sample data, wherein the sample data is the components of oil and gas; A neural network model generating unit, used for inputting the sample data into a preset neural network algorithm to generate a neural network model having at least one hidden layer, wherein the input of the neural network model is the components of oil and gas, and the output of the neural network model is the dry point temperature; A proxy model generating unit, used to transform each hidden layer of the neural network model using a preset matrix to generate a proxy model; A prediction module obtaining unit, used for inputting the calculation result of the proxy model into a preset modeling program for packaging, and obtaining a prediction module for predicting dry points; A dry point prediction value obtaining unit, used for inputting the preset component value in the preset modeling program into the prediction module for calculation to obtain the dry point prediction value; The proxy model generation unit comprises: A weight and intercept acquisition subunit, used to acquire the weight and intercept of each component of each hidden layer of the neural network model; The proxy model generation subunit is used to iteratively transform the weights and intercepts of each component of each hidden layer using a preset matrix to generate the proxy model, including: Using the first matrix, transforming the weights of the components of each hidden layer to obtain the transformed first matrix; Using a second matrix, transforming the intercepts of each component of each hidden layer and the transformed first matrix to obtain the transformed second matrix; Using a third matrix to transform the transformed second matrix to generate the proxy model, wherein the third matrix includes a logistic regression model classification function; The first matrix is matrix1, the second matrix is matrix2, and the third matrix is matrix3. Matrix 3 = F(Matrix 2) F is the sigmoid function.
5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to claim 1 or 2 are implemented.
6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to claim 1 or 2 are implemented.
Citation Information
Patent Citations
Atmospheric overhead oil dry point on-line soft measurement method
CN102880905A