Control method and system for lubricating oil production

Abstract

The invention discloses a lubricating oil production control method and system, and relates to the technical field of lubricating oil production control. According to the technical scheme, raw materials are mixed according to a plan, and refined components are conveyed to a distillation tower. And the reaction controller monitors and controls each device and the acquisition device through the wireless sensor network. The first acquisition device acquires refined component image data, and the reaction controller extracts features and compares the similarity with a first refined product. And the second acquisition device acquires data of the reaction area, detects specific volatile gas generated when the first refined product is converted into the second refined product, generates a chromogenic reaction and transmits reaction data. And the reaction controller performs classification based on feature extraction and guides the feeding mechanism to convey the second refined product to the storage area. Automatic and accurate lubricating oil production control is realized in the whole process, and the production efficiency and the product quality are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of lubricating oil production control, and in particular to a lubricating oil production control method and system. Background Art

[0002] At present, the global market demand for medium and high viscosity heavy lubricant base oils is constantly increasing, and the production process of high viscosity heavy lubricant base oils is constantly being optimized. The process flow of "furfural refining - hydrotreatment / hydrofining - ketone benzene dewaxing - clay refining" has been widely used. In the actual production process, the degree of automation of this process is low, the process steps involved in different production links are relatively complex, and whether the refined product meets the production requirements of the next process is overly dependent on the experience of the staff. The refining process is not controllable and is easily affected by external factors. Therefore, it is particularly important to accurately identify the production stages and processes of high viscosity heavy lubricant base oils.

[0003] In the prior art, although there are a large number of various automated lubricant detection tools such as abrasive particle detection systems, contamination detection systems, and viscosity detection systems for accurate identification and judgment of the production process, most of the equipment is only applicable to a single process. In fact, a large number of technical and automated equipment arrangements are required in the lubricant production stage, which leads to an increase in production costs, and the changes in the properties of each fraction after hydrogenation and distillation of high-viscosity heavy lubricant base oil cannot be controlled. Furthermore, the lubricant production detection system based on machine vision, which has been widely used in recent years, can identify the image features of lubricants and quantify the water content and component properties of lubricants, but this method is only applicable to the detection of finished lubricants, and is not widely applicable to the detection effect of intermediate products in the production process. Therefore, the prior art lacks a control method for the entire process of lubricant production. Summary of the invention

[0004] In view of the above problems, the present invention provides a control method for lubricating oil production, which identifies the reaction state of the refined component through a first acquisition device to accurately obtain a first refined product, and identifies the reaction state of the first refined product through a second acquisition device to accurately obtain a second refined product, and classifies the second refined product into different types of products based on the characteristic recognition. Furthermore, the present invention also provides a control system for implementing the control method for lubricating oil production, which is used to realize automated production and integrated monitoring and control of the entire lubricating oil production process.

[0005] The invention objectives of this application can be achieved through the following technical means:

[0006] A method for controlling lubricating oil production comprises the following steps:

[0007] Step 1: a first collection device is arranged on the distillation tower, and a second collection device is arranged in a plurality of reaction zones, and both the first collection device and the second collection device are connected to a reaction controller;

[0008] Step 2: The batching mechanism transports the raw materials to the feeding mechanism according to the production plan, and the feeding mechanism mixes the raw materials to obtain refined components and transports them to the distillation tower;

[0009] Step 3: The first acquisition device periodically acquires image data of the refined component and sends the image data to the reaction controller;

[0010] Step 4: The reaction controller periodically extracts features of the image data. If the refined component is transformed into the first refined product, the process proceeds to step 5. Otherwise, the process returns to step 3.

[0011] Step 5: The feeding mechanism transports the first refined product to the reaction zone, and the homogenizing device mixes the first refined product uniformly;

[0012] Step 6: The second collection device periodically collects reaction data in the reaction zone and sends the reaction data to the reaction controller;

[0013] Step 7: If the reaction data has the first characteristic, the first refined product is converted into the second refined product and the process proceeds to step 8; otherwise, the process returns to step 6;

[0014] Step 8: extracting a second feature of the reaction data, assigning a reaction label to each reaction area based on the second feature, and any reaction area corresponding to a unique reaction label constitutes a classification criterion;

[0015] Step 9: The feeding mechanism obtains the classification criteria, and transports the second refined product in each reaction zone to the storage area based on the classification criteria.

[0016] In the present invention, the reaction controller includes image data features corresponding to the first refined product under the refined component, and the image data features constitute a training set and a test set.

[0017] In the present invention, in step 3, after the image data of the refined component is input into the reaction controller, the image data features are extracted, and the similarity between the refined component and the first refined product is compared based on the deep learning model.

[0018] In the present invention, the image data features include abrasive particle features, contamination features, viscosity features and moisture features.

[0019] In the present invention, in step 6, the first refined product continuously releases specific volatile gas during the process of being converted into the second refined product, and the specific volatile gas produces a color reaction with the second collection device.

[0020] In the present invention, the color development reaction is a 3×3 array RGB dot matrix diagram, and the reaction data is a plurality of groups of RGB response differences.

[0021] In the present invention, the first feature is the RGB response difference of the refined component corresponding to the second refined product, and the second feature is the main component of the RGB response difference of the refined component corresponding to the second refined product.

[0022] In the present invention, the principal components of the RGB response difference are obtained by performing principal component analysis on the response data, and the principal components are arranged from large to small according to the variance.

[0023] A control system for realizing a control method for lubricating oil production, comprising a distillation tower, a reaction controller, a batching mechanism, a feeding mechanism, a homogenizing device, a feeding mechanism, a first collecting device and a second collecting device, wherein:

[0024] The distillation tower connects the feeding mechanism with the reaction zone for conveying refined components;

[0025] The batching mechanism is connected to the feeding mechanism, and a storage area for various raw materials is arranged on the batching mechanism;

[0026] The feeding mechanism mixes various raw materials by stirring, and has a raw material bin on it, which controls the speed of conveying the refined components to the distillation tower according to the production plan;

[0027] The homogenizing equipment is arranged in the reaction zone and is used to mix and refine the components;

[0028] The feeding mechanism connects the reaction area and the storage area, and a storage area corresponding to any reaction area has a unique feeding mechanism;

[0029] The reaction controller controls the distillation tower, the batching mechanism, the feeding mechanism, the homogenizing device, the feeding mechanism, the first collection device and the second collection device through a wireless sensor network;

[0030] The first collecting device is arranged above the distillation tower;

[0031] The second acquisition device is a sensor array. A plurality of sensor units constitute the sensor array. A single color-sensitive material and a silica gel plate constitute a sensor unit, which is arranged in the reaction area.

[0032] In the present invention, the color-sensitive material is composed of tetraphenylporphyrin, tetraphenylporphyrin copper, tetraphenylporphyrin zinc, tetraphenylporphyrin ferric chloride, tetraphenylporphyrin palladium and tetraphenylporphyrin cobalt.

[0033] The control method and system for lubricating oil production of the present invention have the beneficial effect that the method mainly utilizes technical means such as deep learning model, image data feature extraction, reaction data feature extraction, and combines equipment such as sensor array to achieve accurate production control of lubricating oil and optimize production process. By comparing similarity between image data feature extraction and deep learning model, the changes between refined components and the first refined product can be accurately monitored, thereby achieving accurate control and adjustment of the production process, and effectively improving production efficiency and reducing manual intervention costs. The reaction data collected by the sensor array can be transmitted to the reaction controller in real time and fed back to the production process, so that the production process can be adjusted and optimized according to the real-time data to ensure product quality stability. In addition, by analyzing and processing the reaction data, the reaction zone can be classified and labeled to achieve accurate distribution and storage of the second refined product, reducing waste and loss in the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A flow chart of a method for controlling lubricating oil production according to the present invention;

[0035] Figure 2 A hardware block diagram of a control system for implementing a control method for lubricating oil production according to the present invention;

[0036] Figure 3 This is a hardware structure diagram of the second acquisition device of the present invention. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0038] Embodiment 1

[0039] In the technical solution disclosed in the present invention, the lubricating oil involved should be understood as a high-viscosity heavy lubricating oil base oil, and the vacuum distillate oil is treated by isomerization and decondensation and hydrogenation to obtain a lubricating oil with obvious hierarchical differentiation and a finished oil with higher application value. The specific production process involves multiple steps. As one aspect of the present application, in order to solve the automation of the lubricating oil production process and accurately control each stage of lubricating oil production, the present application provides a specific embodiment of a lubricating oil production control method, which includes two implementation stages:

[0040] System configuration phase:

[0041] A production plan is formulated, based on which standard refined components can be generated. The specific implementation of this technical solution is generated and configured according to the requirements of the production plan.

[0042] The reaction controller receives the production plan and calls the image data features, which are the criteria for determining whether the refined component under the production plan obtains the first refined product. In this embodiment, the reaction controller includes the image data features corresponding to the first refined product under the refined component, and the image data features constitute a training set and a test set.

[0043] The reaction controller receives the production plan and calls the first feature and the second feature. The first feature is the judgment criterion for obtaining the second refined product from the first refined product under the production plan, which is understood as the response difference of the RGB of the corresponding second refined product under the refined component; the classification criterion for the second refined product under the production plan is generated based on the second feature, which is understood as the principal component of the response difference of the RGB of the corresponding second refined product under the refined component. The principal component is obtained by performing principal component analysis on the reaction data, and the principal components are arranged from large to small according to the variance.

[0044] In this embodiment, the production plan can be used as the input of the technical solution. The preferred production plan is a configuration file, which is used to formulate basic rules for the system implementation stage.

[0045] In this embodiment, based on the volatile gas components generated during the lubricating oil production process, the benchmark values ​​of the color-sensitive materials screened through multiple experiments are shown in the following table.

[0046]

[0047] System implementation phase:

[0048] A method for controlling lubricating oil production, referring to Figure 1 , including the following steps:

[0049] Step 1: A first collection device is arranged on the distillation tower, and a second collection device is arranged in multiple reaction zones, and both the first collection device and the second collection device are connected to a reaction controller. In this embodiment, a wireless sensor network is formed between the first collection device, the second collection device and the reaction controller, wherein any one reaction zone has multiple second collection devices, and the second collection devices are evenly arranged in the reaction zone.

[0050] Furthermore, the first collection device and the second collection device transmit data via LoRa technology, thereby reducing communication interference in a narrow area.

[0051] Step 2: The batching mechanism transports the raw materials to the feeding mechanism according to the production plan, and the feeding mechanism mixes the raw materials to obtain refined components, and transports them to the distillation tower. In this embodiment, the refined components are the products of mixing multiple raw materials, which also include dewaxed oil and furfural refined components necessary for the lubricating oil production process.

[0052] Step 3: The first acquisition device periodically acquires image data of the refined component and sends the image data to the reaction controller. After the image data of the refined component is input into the reaction controller, the image data features are extracted, and the similarity between the refined component and the first refined product is compared based on the deep learning model.

[0053] In this embodiment, the first acquisition device is provided with an automatic trigger system to realize automatic image capture at specific time intervals periodically and perform real-time image processing on the captured images. The processing process includes denoising, enhancement and feature analysis to obtain corrected image data.

[0054] Step 4: The reaction controller periodically extracts features of the image data. If the reaction controller identifies the refined component as the first refined product, it proceeds to step 5. Otherwise, it returns to step 3. In this embodiment, the image data features include abrasive features, contamination features, viscosity features, and moisture features.

[0055] Step 5: The feeding mechanism conveys the first refined product to the reaction zone, and the homogenizing device mixes the first refined product uniformly. In this embodiment, the homogenizing device can place the first refined product uniformly in layers in the reaction zone through a multi-layer conveyor belt, and the mixing time is determined by the production plan.

[0056] Step 6: The second collection device periodically collects reaction data in the reaction zone and sends the reaction data to the reaction controller, wherein the first refined product continuously releases specific volatile gas during the process of being converted into the second refined product, and the specific volatile gas produces a color reaction with the second collection device. The color reaction is a 3×3 array RGB dot matrix diagram, and the reaction data is the response difference of multiple groups of RGB.

[0057] Step 7: When the reaction data has the first characteristic, if the first refined product is converted into the second refined product, proceed to step 8, otherwise, return to step 6;

[0058] Step 8: extracting a second feature of the reaction data, assigning a reaction label to each reaction area based on the second feature, and any reaction area corresponding to a unique reaction label constitutes a classification criterion;

[0059] Step 9: The feeding mechanism obtains the classification criteria, and transports the second refined product in each reaction zone to the storage area based on the classification criteria.

[0060] Embodiment 2

[0061] When the first acquisition device obtains the image data of the distillation tower, the original image is preprocessed. The original image is in RGB space. In order to better extract the features of the image data, the RGB space is converted into the HSV space, and the distribution of each pixel point V is counted. The abnormal pixel points are screened based on the 3σ principle, and the brightness V of each point is normalized to obtain the corrected pixel value.

[0062] Based on the color features generated by the image data of the refined components, multiple feature variables are extracted and the mean of each RGB channel is calculated. variance slope Among them, P ij is the grayscale value of a pixel at the coordinate point (i, j), and N is the total number of pixels in the region. In this embodiment, the characteristic variables are hue histogram, hue variance, hue mean, hue slope, saturation mean, saturation variance, saturation slope, lightness mean, lightness variance, and lightness slope.

[0063] The image data is converted into the statistical moment of the grayscale histogram. The preferred statistical moment in this embodiment is the global texture feature of the grayscale histogram, the average grayscale value Standard Deviation Among them, z i is grayscale, p(z i ) is the gray level z i The frequency of, L is the number of all gray levels in the gray level histogram.

[0064] According to the grayscale moment of inertia, grayscale entropy and grayscale correlation of the grayscale co-occurrence matrix, the frequency of the grayscale co-occurrence matrix F(i,j,d,θ)={[(x,y),(x+a,y+b)|f(x,y)=if(x+a,y+b)=j]} is calculated, where θ is the generation direction of the grayscale co-occurrence matrix. The preferred θ in this embodiment is 0°, 45°, 90° and 135°, d is the distance generated by the grayscale co-occurrence matrix, and the frequency F(i,j,d,θ) of the grayscale co-occurrence matrix is ​​used as the texture feature.

[0065] The color image is subjected to two-dimensional discrete wavelet decomposition, and the Haar wavelet is selected as the basis function. This decomposes the image into sub-bands of different frequencies, including high-frequency sub-bands GD2 and DG2 in the horizontal and vertical directions, and extracts information from the GD2 and DG2 sub-bands to calculate feature variables, and the high-frequency sub-bands contain high-frequency texture information in the image.

[0066] Further, for each sub-band, calculate the inertia values K1 and K2. The inertia values will be used to evaluate the distribution uniformity of the image. For sub-band GD2, calculate the pixel mean μGD2 of sub-band GD2, calculate the pixel standard deviation σGD2 of sub-band GD2, and calculate the inertia value K1GD2 of sub-band GD2 according to the mean and standard deviation. For sub-band DG2, calculate the pixel mean μDG2 of sub-band DG2, calculate the pixel standard deviation σDG2 of sub-band DG2, and calculate the inertia value K2DG2 of sub-band DG2 according to the mean and standard deviation, obtaining two characteristic variables K1 and K2.

[0067] Input the two characteristic variables K1 and K2 into the neural network model. If the two characteristic variables K1 and K2 successfully match the image data characteristics, it is determined that the refined component is transformed into the first refined product.

[0068] Example 3

[0069] The second acquisition device periodically acquires reaction data in the reaction zone and generates a first feature and a second feature based on the reaction data. Among them, the first feature is used as the determination criterion for obtaining the second refined product from the first refined product under this production plan, and the classification criterion for the second refined product under this production plan is generated based on the second feature. In a preferred embodiment of the present invention, the generation of the reaction data includes the following steps:

[0070] Step 101: Construct a 3×3 sensor array, input the reference value g(R1, G1, B1) of the color-sensitive material into the second acquisition device. The color-sensitive material of the second acquisition device contacts the volatile gas components in the reaction zone to generate 3×3 groups of RGB channel images.

[0071] Step 102: The first acquisition device extracts the features of the RGB channel images. The feature is the change value g(R2, G2, B2) of the color-sensitive material. Calculate the response difference △g(△R, △G, △B) of RGB according to the reference value of the color-sensitive material, where △R = 丨R1 - R2丨, △G = 丨G1 - G2丨, △B = 丨B1 - B2丨. In this embodiment, when the response difference of the RGB satisfies the first feature, the production process is stopped, and the second acquisition device stops acquiring reaction data.

[0072] Step 103: Calculate the covariance matrix of the RGB response differences. The element c of the covariance matrix C ij represents the covariance between variable x and variable y. By analyzing the eigenvalues and eigenvectors of the covariance matrix C, the variance of the principal component is the eigenvalue, and the direction of the principal component is the eigenvector.

[0073] Step 104: Select n principal components from the eigenvalues in descending order and project them onto the selected principal components. For any sample x, its projection value O projected onto the principal component y xy=△R x ·v yR2 +△Gx·v yG2 +△Bx·v yB2 , where vy R2 is the eigenvector of y principal components on channel R, v yG2 is the eigenvector of y principal components on channel G, v yB2 is the eigenvector of y principal components on channel B, △R x is the corresponding difference of channel R corresponding to sample x, △B x is the corresponding difference of channel B corresponding to sample x, △G x is the corresponding difference of channel G corresponding to sample x.

[0074] Further, after selecting the first n principal components, the weights of the n principal components in the classification criteria need to be determined, and the weights represent the contribution of each principal component to the classification. In this embodiment, the principal component with a larger eigenvalue contributes more to the data variance, and thus can be assigned a higher weight.

[0075] Step 105: Calculate a comprehensive score for the second refined product. The comprehensive score S y =w1·O xy1 +w2·O xy2 +…+w n ·O xyn , according to S y The scoring result is used as a classification criterion, the feeding mechanism obtains the classification criterion, and transports the second refined product in each reaction zone to the storage area based on the classification criterion, wherein the classification criterion is based on principal component analysis and projection value.

[0076] Embodiment 4

[0077] As a second aspect of the present application, in order to realize a more efficient control method for lubricating oil production, a preferred embodiment of the present application also provides a control system for realizing the control method for lubricating oil production, referring to Figure 2 , including a distillation tower, a reaction controller, a batching mechanism, a feeding mechanism, a homogenizing device, a feeding mechanism, a first collecting device and a second collecting device;

[0078] The distillation tower is connected to the feeding mechanism and the reaction zone for conveying the refined components. The batching mechanism is connected to the feeding mechanism, and a storage area for various raw materials is arranged on the batching mechanism. The feeding mechanism mixes various raw materials by stirring, and a raw material bin is arranged on it, and the raw material bin controls the speed of conveying the refined components to the distillation tower according to the production plan. The homogenizing device is arranged in the reaction zone for mixing the refined components. The feeding mechanism connects the reaction zone and the storage area, and a storage area corresponding to any reaction zone has a unique feeding mechanism.

[0079] The reaction controller controls the distillation tower, the batching mechanism, the feeding mechanism, the homogenizing device, the feeding mechanism, the first collecting device and the second collecting device through the wireless sensor network. The first collecting device is arranged above the distillation tower. The second collecting device is arranged in the reaction zone and is composed of a color-sensitive material and a silica gel plate.

[0080] In this embodiment, refer to Figure 3 The second acquisition device is a sensor array, a plurality of sensor units constitute the sensor array, and a single color-sensitive material and a silicone plate constitute a sensor unit.

[0081] In this embodiment, the color-sensitive material is composed of tetraphenylporphyrin, tetraphenylporphyrin copper, tetraphenylporphyrin zinc, tetraphenylporphyrin ferric chloride, tetraphenylporphyrin palladium and tetraphenylporphyrin cobalt.

[0082] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for controlling lubricating oil production, characterized in that: The following steps are involved: Step 1: a first collection device is arranged on the distillation tower, and a second collection device is arranged in a plurality of reaction zones, and both the first collection device and the second collection device are connected to a reaction controller; Step 2: The batching mechanism transports the raw materials to the feeding mechanism according to the production plan, and the feeding mechanism mixes the raw materials to obtain refined components and transports them to the distillation tower; Step 3: The first acquisition device periodically acquires image data of the refined component and sends the image data to the reaction controller; Step 4: The reaction controller periodically extracts features of the image data. If the refined component is transformed into the first refined product, the process proceeds to step 5. Otherwise, the process returns to step 3. Step 5: The feeding mechanism transports the first refined product to the reaction zone, and the homogenizing device mixes the first refined product uniformly; Step 6: The second collection device periodically collects reaction data in the reaction zone and sends the reaction data to the reaction controller; Step 7: If the reaction data has the first characteristic, the first refined product is converted into the second refined product and the process proceeds to step 8; otherwise, the process returns to step 6; Step 8: extracting a second feature of the reaction data, assigning a reaction label to each reaction area based on the second feature, and any reaction area corresponding to a unique reaction label constitutes a classification criterion; Step 9: The feeding mechanism obtains the classification criteria, and transports the second refined product in each reaction zone to the storage area based on the classification criteria.

2. The method for controlling lubricating oil production according to claim 1, characterized in that: The reaction controller includes image data features corresponding to the first refined product under the refined component, and the image data features constitute a training set and a test set.

3. The method for controlling lubricating oil production according to claim 1, characterized in that: In step 3, after the image data of the refined component is input into the reaction controller, the image data features are extracted, and the similarity between the refined component and the first refined product is compared based on the deep learning model.

4. The method for controlling lubricating oil production according to claim 3, characterized in that: The image data features include abrasive particle features, contamination features, viscosity features and moisture features.

5. The method for controlling lubricating oil production according to claim 1, characterized in that: In step 6, the first refined product continuously releases specific volatile gas during the process of being converted into the second refined product, and the specific volatile gas reacts with the second collection device to produce a color development reaction.

6. The method for controlling lubricating oil production according to claim 1, characterized in that: The color development reaction is a 3×3 array RGB dot matrix diagram, and the reaction data is the response difference of multiple groups of RGB.

7. The method for controlling lubricating oil production according to claim 1, characterized in that: The first feature is the RGB response difference of the second refined product under the refined component, and the second feature is the main component of the RGB response difference of the second refined product under the refined component.

8. The method for controlling lubricating oil production according to claim 7, characterized in that: The principal components of the RGB response differences are obtained by performing principal component analysis on the response data, and the principal components are arranged from large to small according to the variance.

9. A control system for the lubricating oil production control method according to claim 1, characterized in that: It includes a distillation tower, a reaction controller, a batching mechanism, a feeding mechanism, a homogenizing device, a feeding mechanism, a first collecting device and a second collecting device, wherein: The distillation tower connects the feeding mechanism with the reaction zone for conveying refined components; The batching mechanism is connected to the feeding mechanism, and a storage area for various raw materials is arranged on the batching mechanism; The feeding mechanism mixes various raw materials by stirring, and has a raw material bin on it, which controls the speed of conveying the refined components to the distillation tower according to the production plan; The homogenizing equipment is arranged in the reaction zone and is used to mix and refine the components; The feeding mechanism connects the reaction area and the storage area, and a storage area corresponding to any reaction area has a unique feeding mechanism; The reaction controller controls the distillation tower, the batching mechanism, the feeding mechanism, the homogenizing device, the feeding mechanism, the first collection device and the second collection device through a wireless sensor network; The first collecting device is arranged above the distillation tower; The second acquisition device is a sensor array. A plurality of sensor units constitute the sensor array. A single color-sensitive material and a silica gel plate constitute a sensor unit, which is arranged in the reaction area. In the present invention, the color-sensitive material is composed of tetraphenylporphyrin, tetraphenylporphyrin copper, tetraphenylporphyrin zinc, tetraphenylporphyrin ferric chloride, tetraphenylporphyrin palladium and tetraphenylporphyrin cobalt.

Images