Online Detection Method and System for Copper Strip Corrosion of Petroleum Products Based on Computer Vision

Abstract

The present invention relates to the technical field of copper corrosion detection, and discloses an on-line detection method and system for copper sheet corrosion of petroleum products based on computer vision, including: constructing a macroscopic data collection point of a test copper sheet and a microscopic data collection point of the test copper sheet; respectively performing data preprocessing on historical microscopic data and current microscopic data, establishing a concentration change model of the microscopic data collection point, and analyzing the corrosion diffusion degree corresponding to the current data through the concentration change model; selecting corrosion abnormal points from the macroscopic data collection points; performing equally spaced processing on the historical macroscopic data to generate a grey model of equally long historical data, performing equally spaced processing on the current macroscopic data, and outputting the corrosion severity corresponding to the equally long current data through the grey model; selecting an image collection point from the corrosion abnormal points, performing image collection at the image collection point, and using the collected image to perform corrosion detection on the image collection point. The present invention can reduce the amount of data during copper sheet corrosion detection.

Description

Technical Field

[0001] The present invention relates to an online detection method and system for copper strip corrosion of petroleum products based on computer vision, belonging to the technical field of copper corrosion detection. Background Art

[0002] In the quality control process of petroleum products, the copper strip corrosion test is an important indicator, which can reflect the corrosivity of petroleum products to metallic copper. Traditional copper strip corrosion detection methods mainly rely on manual visual inspection and comparison with standard color plates, which have problems such as strong subjectivity, low efficiency, and low accuracy. Therefore, developing a device that can automatically detect the degree of copper strip corrosion online is of great significance for improving the detection efficiency and quality control level of petroleum products.

[0003] Currently, when the volume and area of copper strips are large, image recognition technology cannot completely capture all copper strip areas for detection. The primary reason is that the area of copper strips recognized by image recognition technology at one time cannot completely cover the total area occupied by the copper strips. That is to say, it is necessary to use image recognition technology to scan the copper strip area multiple times and at multiple positions to complete the acquisition of image data on the copper strip surface. For example, when the area occupied by the copper strip is 20m * 20m and the area of the copper strip recognized by image recognition technology at one time is 5m * 5m, it is necessary to use an image collector to scan the copper strip surface 4 times to collect the complete image of the copper strip. Secondly, when collecting the copper strip image, it is also necessary to collect a copper strip sample from the complete copper strip, and to collect the copper strip sample, it is necessary to take a copper strip image sample with an unaffected color when the copper strip is not contaminated with petroleum. However, such opportunities are extremely rare because when the copper strip is in use (at this time the copper strip is in contact with petroleum), it is very difficult to obtain a copper strip that is not contaminated with petroleum. Especially when all copper strip areas are not contaminated with petroleum to capture the copper strip images of all area regions, which leads to difficulties in corrosion detection of copper strip areas. For example, when the area occupied by the copper strip is 20m * 20m and the area of the copper strip not contaminated with petroleum is 4m * 4m, only the 4m * 4m area can be used for image acquisition, and for other areas contaminated with petroleum, it is necessary to wait for the petroleum to subside before image data can be collected. Moreover, since it is not possible to determine which areas need to be focused on for collection and which areas need to be ignored, it is still necessary to collect image data for the entire copper strip area, resulting in a long time cycle and a huge amount of image data to be collected and processed, and the efficiency of copper strip corrosion detection is low.

[0004] Therefore, the amount of copper strip image data during copper strip corrosion detection is large. Summary of the Invention

[0005] The present invention provides an online detection method and system for copper strip corrosion of petroleum products based on computer vision, and its main purpose is to reduce the amount of data during copper strip corrosion detection.

[0006] To achieve the above object, an on-line detection method for copper sheet corrosion of petroleum products based on computer vision provided by the present invention includes:

[0007] Construct macroscopic data collection points of the test copper sheet, and construct microscopic data collection points of the test copper sheet, and collect historical microscopic data and current microscopic data at the microscopic data collection points;

[0008] Perform data preprocessing on the historical microscopic data and the current microscopic data respectively to obtain pre-historical data and pre-current data. According to the pre-historical data, establish a concentration change model of the microscopic data collection point, and analyze the corrosion diffusion degree corresponding to the pre-current data through the concentration change model;

[0009] According to the microscopic data collection point and the corrosion diffusion degree, select corrosion abnormal points from the macroscopic data collection points, and collect historical macroscopic data and current macroscopic data at the corrosion abnormal points;

[0010] Perform equally spaced processing on the historical macroscopic data to obtain equally long historical data, generate a grey model of the equally long historical data, perform equally spaced processing on the current macroscopic data to obtain equally long current data, and output the corrosion severity corresponding to the equally long current data through the grey model;

[0011] Based on the corrosion severity, select image collection points from the corrosion abnormal points, perform image collection at the image collection points to obtain a collected image, obtain a training image corresponding to the collected image, perform image preprocessing on the training image to obtain a preprocessed image, and based on the preprocessed image, generate a trained encoder and a trained decoder of the collected image. According to the trained encoder and the trained decoder, use the collected image to perform corrosion detection on the image collection points to obtain the corrosion detection result of the test copper sheet.

[0012] Optionally, the construction of the macroscopic data collection points of the test copper sheet includes:

[0013] Obtain the macroscopic data type of the test copper sheet;

[0014] Based on the macroscopic data type, set the macroscopic data collection device of the test copper sheet;

[0015] Query the number of ranges of the collection ranges of the macroscopic data collection devices included in the range corresponding to the test copper sheet;

[0016] Use the number of ranges to determine the number of devices of the macroscopic data collection device;

[0017] Generate random positions of the macroscopic data acquisition device within the range corresponding to the test copper sheet according to the number of the devices;

[0018] When the coverage area of the macroscopic data acquisition device at the random position is not less than the range corresponding to the test copper sheet, use the random position as the macroscopic data acquisition point of the test copper sheet.

[0019] Optionally, establishing the concentration change model of the microscopic data acquisition point according to the pre-historical data includes:

[0020] Query the microscopic concentration and copper sheet concentration of the microscopic data acquisition point from the pre-historical data;

[0021] Compare the copper sheet concentration with the microscopic concentration to obtain a concentration comparison result;

[0022] Use the concentration comparison result to determine the diffusion parameter of the microscopic data acquisition point;

[0023] According to the pre-historical data and the diffusion parameter, establish the concentration change model of the microscopic data acquisition point by using the following formula:

[0024]

[0025] Wherein, represents the concentration change model, s(z,t) represents the concentration at point z at time t, represents the concentration within the neighborhood range of point z in the pre-historical data of the point, represents the diffusion parameter between point z and the point, represents in the pre-historical data the volume of the point, D z represents the neighborhood of point z.

[0026] Optionally, analyzing the corrosion diffusion degree corresponding to the pre-current data through the concentration change model includes:

[0027] Analyze the current concentration value corresponding to the pre-current data by using the concentration change model;

[0028] According to the current concentration value, calculate the corrosion diffusion degree corresponding to the pre-current data by using the following formula:

[0029]

[0030] Wherein, q(z′,t) represents the corrosion diffusion degree, s(z′,t) represents the current concentration value, s1 represents the standard concentration of the test copper sheet, and s2 represents the standard concentration of the petroleum.

[0031] Optionally, selecting corrosion anomaly points from the macro data collection points according to the micro data collection points and the corrosion diffusion degree includes:

[0032] Determining whether the corrosion diffusion degree is an abnormal diffusion degree;

[0033] When the corrosion diffusion degree is an abnormal diffusion degree, querying the first target collection point corresponding to the corrosion diffusion degree in the micro data collection points;

[0034] Identifying the second target collection point in the macro data collection points that covers the first target collection point;

[0035] Taking the second target collection point as the corrosion anomaly point.

[0036] Optionally, performing equally spaced processing on the historical macro data to obtain equally long historical data includes:

[0037] Extracting the macro data time of the historical macro data; sorting the macro data time to obtain a data time sequence; calculating the standard time interval of the data time sequence using the following formula:

[0038]

[0039] where Δt represents the standard time interval, m represents the length of the data time sequence, t m represents the maximum time in the data time sequence, and t1 represents the minimum time in the data time sequence;

[0040] According to the standard time interval, performing equally spaced processing on the macro data sequence corresponding to the data time sequence using the following formula to obtain equally long historical data:

[0041]

[0042] where y(t j ) represents the equally long historical data, x(t j ) represents the historical macro data at time t in the macro data sequence, j represents the serial number of the time in the data time sequence, t j represents the j-th time in the data time sequence, x(t j -1) represents the predecessor of x(t j ) in the macro data sequence, and Δt represents the standard time interval. j ) in the macro data sequence, and Δt represents the standard time interval.

[0043] Optionally, generating the grey model of the equally long historical data includes:

[0044] Accumulate the equal-length historical data using the following formula to obtain the accumulated historical data:

[0045]

[0046] where y (1) (k) represents the accumulated historical data, y (0) (i) represents the i-th data in the equal-length historical data, n represents the number of values in the same type of the equal-length historical data, and k represents the sequence number less than n;

[0047] Construct the grey model of the equal-length historical data using the following formula according to the accumulated historical data:

[0048]

[0049] Y n = BU

[0050]

[0051] where, represents the grey model, a and u in U represent unknown parameters, U represents the matrix composed of unknown parameters, y (0) (2), y (0) (3), y (0) (n) represent the values of y (0) (i) when taking different i, Y n represents the matrix of the equal-length historical data composed of y (0) (2), y (0) (3), y (0) (n), y (1) (1), y (1) (2), y (1) (3), y (1) (n - 1), y (1) (n) represent the values of y (1) (k) when taking different k, B represents the matrix composed of the accumulated historical data, U represents the calculation result of U, a represents the calculation result of a, and u represents the calculation result of u, represents the value of the accumulated historical data in the next time period, y (1) (t) represents any type of data in the accumulated historical data, and t represents the time.

[0052] Optionally, outputting the corrosion severity corresponding to the equal-length current data through the grey model includes:

[0053] Calculating the planned current data of the equal-length historical data corresponding to the grey model according to the grey model;

[0054] Query the historical corrosion degree corresponding to the equal-length historical data;

[0055] Construct a grey analysis model for the historical corrosion degree;

[0056] Determine whether the planned current data is consistent with the equal-length current data;

[0057] When the planned current data is consistent with the equal-length current data, use the grey analysis model to output the corrosion severity corresponding to the equal-length current data;

[0058] When the planned current data is inconsistent with the equal-length current data, take severe copper sheet corrosion as the corrosion severity corresponding to the equal-length current data.

[0059] Optionally, the trained encoder and decoder for generating the acquisition image based on the preprocessed image include:

[0060] Divide the preprocessed image into multi-view sample anchors, positive samples, and negative samples;

[0061] In a preset encoder, use the following formula to map the multi-view sample anchor, the positive sample, and the negative sample to the feature latent space respectively, to obtain the anchor feature, the positive sample feature, and the negative sample feature:

[0062] f a =Φ(s a ),f p =Φ(s p ),f n =Φ(s n )

[0063] where, f a represents the anchor feature, f p represents the positive sample feature, f n represents the negative sample feature, s a represents the multi-view sample anchor, Φ represents the encoder, s p represents the positive sample, s n represents the negative sample;

[0064] Use the following formula to calculate the first loss value corresponding to the anchor feature, the positive sample feature, and the negative sample feature:

[0065]

[0066] L2=||Rec(f a )-s a ||

[0067] Among them, L1 represents the loss value between the anchor feature and the positive sample feature and the negative sample feature in the first loss value, L2 represents the loss value between the anchor feature and the multi-view sample anchor in the first loss value, and f a represents the anchor feature, f p represents the positive sample feature, f n represents the negative sample feature, s a represents the multi-view sample anchor, Rec represents the reconstruction model, and Rec is used to reconstruct f a back to the original image and ensure that the latent features extracted by the extraction encoder Φ are complete;

[0068] Obtain the feature sets corresponding to the anchor feature, the positive sample feature, and the negative sample feature; in a preset decoder, use the following formula to perform feature decoding on the feature sets to obtain the decoded category:

[0069] o i = Ψ(f i )

[0070] where, o i represents the decoded category corresponding to the i-th feature in the feature set, f i represents the i-th feature in the feature set, Ψ represents the decoder, and the decoder Ψ is composed of 2 fully connected layers. The decoder Ψ is used to directly map the latent features to the category space;

[0071] Use the following formula to calculate the second loss value corresponding to the decoded category:

[0072] L3 = Softmax(o i , g i )

[0073] where, L3 represents the second loss value, o i represents the decoded category corresponding to the i-th feature in the feature set, g i represents the label corresponding to o i ;

[0074] Based on the first loss value and the second loss value, train the encoder and the decoder to obtain a trained encoder and a trained decoder.

[0075] To solve the above problems, the present invention also provides an on-line detection system for copper sheet corrosion of petroleum products based on computer vision. The system includes:

[0076] A microscopic acquisition module, configured to construct a macroscopic data acquisition point of a test copper sheet and construct a microscopic data acquisition point of the test copper sheet, and acquire historical microscopic data and current microscopic data at the microscopic data acquisition point;

[0077] A diffusion analysis module, which is used to perform data preprocessing on the historical microdata and the current microdata respectively to obtain pre-historical data and pre-current data, establish a concentration change model of the microdata acquisition point according to the pre-historical data, and analyze the corrosion diffusion degree corresponding to the pre-current data through the concentration change model;

[0078] A macro acquisition module, which is used to select corrosion abnormal points from the macrodata acquisition points according to the microdata acquisition point and the corrosion diffusion degree, and collect historical macrodata and current macrodata at the corrosion abnormal points;

[0079] A degree output module, which is used to perform equally spaced processing on the historical macrodata to obtain equally long historical data, generate a grey model of the equally long historical data, perform equally spaced processing on the current macrodata to obtain equally long current data, and output the corrosion severity corresponding to the equally long current data through the grey model;

[0080] A corrosion detection module, which is used to select an image acquisition point from the corrosion abnormal points based on the corrosion severity, perform image acquisition at the image acquisition point to obtain a collected image, obtain a training image corresponding to the collected image, perform image preprocessing on the training image to obtain a preprocessed image, generate a trained encoder and a trained decoder of the collected image based on the preprocessed image, and perform corrosion detection on the image acquisition point by using the collected image according to the trained encoder and the trained decoder to obtain the corrosion detection result of the test copper sheet.

[0081] Compared with the problems in the background art, in the embodiment of the present invention, a concentration change model of the microdata acquisition point is established according to the pre-historical data to be used to identify whether corrosion has occurred on the copper sheet surface by using the change in the petroleum concentration on the test copper sheet. If it is calculated that the copper sheet concentration is reduced compared with the copper sheet concentration of the standard version, it means that the copper sheet has corroded. In the embodiment of the present invention, corrosion abnormal points are selected from the macrodata acquisition points according to the microdata acquisition point and the corrosion diffusion degree to select a part and a small number of acquisition points from a large number of microdata acquisition points, thereby reducing the amount of corrosion detection data. Further, in the embodiment of the present invention, the corrosion detection result of the test copper sheet is obtained by performing corrosion detection on the image acquisition point by using the collected image to reduce the amount of corrosion detection data of the copper sheet. Therefore, the online corrosion detection method of petroleum product copper sheet based on computer vision proposed by the present invention can reduce the amount of data during copper sheet corrosion detection. Description of the Drawings

[0082] Figure 1Schematic flowchart of the online detection method for copper strip corrosion of petroleum products based on computer vision provided by an embodiment of the present invention;

[0083] Figure 2 Encoder schematic diagram of the online detection method for copper strip corrosion of petroleum products based on computer vision provided by an embodiment of the present invention;

[0084] Figure 3 Decoder schematic diagram of the online detection method for copper strip corrosion of petroleum products based on computer vision provided by an embodiment of the present invention;

[0085] Figure 4 Module schematic diagram for implementing the online detection method for copper strip corrosion of petroleum products based on computer vision provided by an embodiment of the present invention.

[0086] The implementation, functional features, and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. Detailed implementation manners

[0087] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0088] An embodiment of the present application provides an online detection method for copper strip corrosion of petroleum products based on computer vision. The execution subject of the online detection method for copper strip corrosion of petroleum products based on computer vision includes, but is not limited to, at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the online detection method for copper strip corrosion of petroleum products based on computer vision can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to: a single server, a server cluster, a cloud server, or a cloud server cluster, etc.

[0089] Embodiment 1:

[0090] Refer to Figure 1 As shown, it is a schematic flowchart of the online detection method for copper strip corrosion of petroleum products based on computer vision provided by an embodiment of the present invention. In this embodiment, the online detection method for copper strip corrosion of petroleum products based on computer vision includes:

[0091] S1. Construct macroscopic data collection points for the test copper strip, and construct microscopic data collection points for the test copper strip, and collect historical microscopic data and current microscopic data at the microscopic data collection points.

[0092] In the embodiment of the present invention, the test copper strip refers to a copper strip in contact with petroleum, the macroscopic data collection point refers to the position where the data collection device for collecting macroscopic data on the copper strip is located, and the macroscopic data refers to temperature data and time data.

[0093] In one embodiment of the present invention, the construction of the macroscopic data acquisition points of the test copper sheet includes: obtaining the macroscopic data types of the test copper sheet; setting the macroscopic data acquisition device of the test copper sheet based on the macroscopic data types; querying the number of ranges of the acquisition ranges of the macroscopic data acquisition devices included in the corresponding range of the test copper sheet; determining the number of devices of the macroscopic data acquisition device using the number of ranges; generating random positions of the macroscopic data acquisition device within the corresponding range of the test copper sheet according to the number of devices; when the coverage area of the macroscopic data acquisition device at the random position is not less than the corresponding range of the test copper sheet, taking the random position as the macroscopic data acquisition point of the test copper sheet.

[0094] Among them, the macroscopic data types include temperature type and time type, the macroscopic data acquisition device includes a temperature acquisition device and a time recording device, the corresponding range of the test copper sheet refers to the area and region of the test copper sheet, the acquisition range refers to the maximum area and region where each macroscopic data acquisition device can acquire data, the number of devices is the same as the number of ranges, and the coverage area refers to the area occupied by the area and region where multiple macroscopic data acquisition devices acquire data on the test copper sheet.

[0095] Furthermore, in an embodiment of the present invention, the microscopic data acquisition point refers to the position where the data acquisition device for collecting the microscopic data of the oil contaminated on the copper sheet and the position where the device for collecting the concentration of the copper sheet are located. The microscopic data refers to the concentration data and volume data of the material points in the oil contaminated on the copper sheet and the concentration data of the copper sheet, where the material points refer to metal atoms. The historical microscopic data refers to the microscopic data within a historical period, and the current microscopic data refers to the microscopic data at the current moment.

[0096] Optionally, the process of constructing the microscopic data acquisition points of the test copper sheet is similar to the above process of constructing the macroscopic data acquisition points of the test copper sheet, and will not be elaborated further here.

[0097] Optionally, the process of collecting historical microscopic data and current microscopic data at the microscopic data acquisition point refers to the process of measuring the concentration and volume of metal atoms and the concentration of metal atoms of the entire copper sheet, which can be realized by professional instruments, such as atomic absorption spectrometry instruments, etc.

[0098] S2. Respectively perform data preprocessing on the historical microscopic data and the current microscopic data to obtain pre-historical data and pre-current data. According to the pre-historical data, establish a concentration change model for the microscopic data acquisition point, and analyze the corrosion diffusion degree corresponding to the pre-current data through the concentration change model.

[0099] Optionally, the process of respectively performing data preprocessing on the historical microdata and the current microdata to obtain pre-historical data and pre-current data refers to the process of reducing or enlarging the unit of the microdata. For example, when the unit is too small, the value will be too large. In this case, the unit is enlarged, and when enlarging the unit of this value, it is also necessary to match the units of other data. Therefore, the units of other data also need to be enlarged or reduced accordingly.

[0100] Further, in an embodiment of the present invention, a concentration change model of the microdata acquisition point is established according to the pre-historical data, so as to use the change in the petroleum concentration on the test copper sheet to identify whether corrosion has occurred on the copper sheet surface. If it is calculated that the copper sheet concentration is reduced relative to the copper sheet concentration of the standard version, it means that the copper sheet has corroded.

[0101] Among them, the concentration change model refers to the influence of the change in the concentration of petroleum material points in the neighborhood range of the material points on the copper sheet over time on the concentration of the material points on the copper sheet.

[0102] In an embodiment of the present invention, the establishment of the concentration change model of the microdata acquisition point according to the pre-historical data includes: querying the micro-concentration and copper sheet concentration of the microdata acquisition point from the pre-historical data; comparing the copper sheet concentration with the micro-concentration to obtain a concentration comparison result; using the concentration comparison result to determine the diffusion parameter of the microdata acquisition point; and establishing the concentration change model of the microdata acquisition point according to the pre-historical data and the diffusion parameter by using the following formula:

[0103]

[0104] Among them, represents the concentration change model, s(z, t) represents the concentration at point z at time t, represents the concentration of points within the neighborhood range of point z in the pre-historical data points, represents the diffusion parameter between point z and points, represents in the pre-historical data the volume of points, D z represents the neighborhood of point z.

[0105] Among them, the micro-concentration refers to the concentration of material points points in the petroleum, and the copper sheet concentration refers to the concentration data in a spatial region far larger than point z, rather than the concentration of copper sheet material points in this spatial region. Point z refers to the material point in the copper sheet, and the neighborhood of point z is the spherical region formed by the spherical radius of the material points in the copper sheet. Find the nearest point on the copper sheet near the point, and use this point as point z.

[0106] Optionally, the process of determining the diffusion parameter of the microscopic data acquisition point using the concentration comparison result means that when the copper sheet concentration is greater than the microscopic concentration, the dissolution rate of the copper sheet in the petroleum is used as the diffusion parameter, and when the copper sheet concentration is less than the microscopic concentration, the diffusion coefficient of copper sheet examples in the petroleum is used as the diffusion parameter.

[0107] In an embodiment of the present invention, analyzing the corrosion diffusion degree corresponding to the pre-current data through the concentration change model includes: analyzing the current concentration value corresponding to the pre-current data using the concentration change model; according to the current concentration value, calculating the corrosion diffusion degree corresponding to the pre-current data using the following formula:

[0108]

[0109] where q(z′, t) represents the corrosion diffusion degree, s(z′, t) represents the current concentration value, s1 represents the standard concentration of the test copper sheet, and s2 represents the standard concentration of the petroleum.

[0110] wherein the current concentration value refers to the value of s(z, t) under the premise of the pre-current data.

[0111] S3. According to the microscopic data acquisition point and the corrosion diffusion degree, select corrosion abnormal points from the macroscopic data acquisition points, and collect historical macroscopic data and current macroscopic data at the corrosion abnormal points.

[0112] In the embodiment of the present invention, by selecting corrosion abnormal points from the macroscopic data acquisition points according to the microscopic data acquisition point and the corrosion diffusion degree, it is used to select a part of the microscopic data acquisition points from a large number of microscopic data acquisition points, thereby reducing the amount of corrosion detection data.

[0113] In an embodiment of the present invention, selecting corrosion abnormal points from the macroscopic data acquisition points according to the microscopic data acquisition point and the corrosion diffusion degree includes: determining whether the corrosion diffusion degree is an abnormal diffusion degree; when the corrosion diffusion degree is an abnormal diffusion degree, querying the first target acquisition point corresponding to the corrosion diffusion degree in the microscopic data acquisition points; identifying the second target acquisition point covering the first target acquisition point in the macroscopic data acquisition points; and taking the second target acquisition point as the corrosion abnormal point.

[0114] Optionally, the process of determining whether the corrosion diffusion degree is an abnormal diffusion degree is that when q(z′, t) is non-zero, it represents an abnormal diffusion degree.

[0115] Further, in the embodiments of the present invention, the historical macroscopic data refers to the macroscopic data within a historical time period, and the current macroscopic data refers to the macroscopic data at the current moment.

[0116] Optionally, the process of collecting historical macroscopic data and current macroscopic data at the corrosion anomaly point can be implemented by a clock and a temperature sensor.

[0117] S4. Perform equally-spaced processing on the historical macroscopic data to obtain equally-long historical data, generate a grey model for the equally-long historical data, perform equally-spaced processing on the current macroscopic data to obtain equally-long current data, and output the corrosion severity corresponding to the equally-long current data through the grey model.

[0118] In one embodiment of the present invention, the performing equally-spaced processing on the historical macroscopic data to obtain equally-long historical data includes: extracting the macroscopic data time of the historical macroscopic data; sorting the macroscopic data times to obtain a data time sequence; calculating the standard time interval of the data time sequence by using the following formula:

[0119]

[0120] where Δt represents the standard time interval, m represents the length of the data time sequence, t m represents the maximum time in the data time sequence, and t1 represents the minimum time in the data time sequence;

[0121] According to the standard time interval, perform equally-spaced processing on the macroscopic data sequence corresponding to the data time sequence by using the following formula to obtain equally-long historical data:

[0122]

[0123] where y(t j ) represents the equally-long historical data, x(t j ) represents the historical macroscopic data at time t j in the macroscopic data sequence, j represents the serial number of the time in the data time sequence, t j represents the j-th time in the data time sequence, x(t j -1) represents the predecessor of x(t j ) in the macroscopic data sequence, and Δt represents the standard time interval.

[0124] It should be noted that when sorting the macroscopic data times, only the data within the same type are sorted. For example, the temperature data at different times are sorted, and the time lengths at different times are sorted, rather than mixing and sorting the temperature and time.

[0125] In one embodiment of the present invention, the grey model for generating the equi-length historical data includes: performing data accumulation on the equi-length historical data using the following formula to obtain the accumulated historical data:

[0126]

[0127] where y (1) (k) represents the accumulated historical data, y (0) (i) represents the i-th data in the equi-length historical data, n represents the number of values in the same type of the equi-length historical data, and k represents the sequence number less than n;

[0128] Construct the grey model of the equi-length historical data using the following formula according to the accumulated historical data:

[0129]

[0130] Y n = BU

[0131]

[0132] where, represents the grey model, a and u in U represent unknown parameters, U represents the matrix composed of unknown parameters, y (0) (2), y (0) (3), y (0) (n) represent the values of y (0) (i) when taking different i, Y n represents the matrix of the equi-length historical data composed of y (0) (2), y (0) (3), y (0) (n), y (1) (1), y (1) (2), y (1) (3), y (1) (n - 1), y (1) (n) represent the values of y (1) (k) when taking different k, B represents the matrix composed of the accumulated historical data, U represents the calculation result of U, a represents the calculation result of a, u represents the calculation result of u, y (1) (t + 1) represents the value of the accumulated historical data in the next time period, y (1) (t) represents any type of data in the accumulated historical data, and t represents the time.

[0133] It should be noted that the process of performing data accumulation on the equi-length historical data is also to accumulate only the data within the same type.

[0134] In an embodiment of the present invention, outputting the corrosion severity corresponding to the equal-length current data through the gray model includes: according to the gray model, calculating the planned current data of the equal-length historical data corresponding to the gray model by using the following formula:

[0135]

[0136] Wherein, represents the planned current data, represents the value of the accumulated historical data in the next time period, y (1) (t) represents any type of data in the accumulated historical data, and t represents the time;

[0137] Query the historical corrosion degree corresponding to the equal-length historical data; construct a gray analysis model of the historical corrosion degree; judge whether the planned current data is consistent with the equal-length current data; when the planned current data is consistent with the equal-length current data, use the gray analysis model to output the corrosion severity corresponding to the equal-length current data; when the planned current data is inconsistent with the equal-length current data, take the severe copper sheet corrosion as the corrosion severity corresponding to the equal-length current data.

[0138] Wherein, the planned current data refers to the macroscopic data at the time consistent with the time of the equal-length current data, and the gray analysis model refers to the gray model of the historical corrosion degree. It should be noted that when the planned current data is inconsistent with the equal-length current data, it means that an abnormality occurs when using the gray model to analyze the data in the future time period. Therefore, it is necessary to use image recognition technology in the follow-up to further identify whether the equal-length current data has copper sheet corrosion. Therefore, take the severe copper sheet corrosion as the corrosion severity corresponding to the equal-length current data. The historical corrosion degree and the corrosion severity refer to the severe level of copper sheet corrosion, such as uncorroded, slightly corroded, severely corroded, etc.

[0139] S5. Based on the corrosion severity, select image acquisition points from the corrosion abnormal points, perform image acquisition at the image acquisition points to obtain acquisition images, obtain the training images corresponding to the acquisition images, perform image preprocessing on the training images to obtain preprocessed images, based on the preprocessed images, generate a trained encoder and a trained decoder for the acquisition images, and according to the trained encoder and the trained decoder, use the acquisition images to perform corrosion detection on the image acquisition points to obtain the corrosion detection results of the test copper sheet.

[0140] Optionally, the process of selecting image acquisition points from the corrosion abnormal points based on the corrosion severity is similar to the process of selecting corrosion abnormal points from the macro data acquisition points according to the microscopic data acquisition points and the corrosion diffusion degree, and will not be elaborated further here.

[0141] Among them, the training images refer to the image data of the surfaces of all copper sheets.

[0142] In an embodiment of the present invention, the preprocessing the training images to obtain preprocessed images includes: performing color space conversion on the training images to obtain color conversion images; performing histogram equalization on the training images to obtain color enhancement images; performing edge detection on the training images to obtain edge detection images; performing image filtering on the training images to obtain filtered images; enhancing the texture features of the training images to obtain feature enhancement images; fusing the color conversion images, the color enhancement images, the edge detection images, the filtered images and the feature enhancement images to obtain multi-channel images; removing the noise of the multi-channel images to obtain denoised images; performing image normalization on the denoised images to obtain normalized images;

[0143] Optionally, the process of preprocessing the training images to obtain preprocessed images includes: converting the RGB images to the Lab color space to highlight the color differences in the images; using the luminance channel (L) in the Lab space to perform CLAHE (contrast-limited adaptive histogram equalization) to enhance the brightness contrast of the images and make the colors more obvious; using Canny edge detection to extract the edge information of the images and enhance the texture features; using multi-scale and multi-directional Gabor filters to extract the texture features of specific scales and directions; using local binary pattern (LBP) to extract the local texture patterns in the images and enhance the subtle surface texture features; fusing the enhanced color images, edge images, Gabor filtering results, and LBP features into a multi-channel image; using bilateral filtering to remove noise and retain the edge information; normalizing the images so that their values are between 0 and 1 for easy processing by the neural network.

[0144] In an embodiment of the present invention, the trained encoder and decoder for generating the acquisition images based on the preprocessed images include: dividing the preprocessed images into multi-view sample anchors, positive samples, and negative samples; in a preset encoder, using the following formula to map the multi-view sample anchors, the positive samples, and the negative samples to the feature latent space respectively to obtain anchor features, positive sample features, and negative sample features:

[0145] f a =Φ(s a ),f p= Φ(s p ), f n = Φ(s n )

[0146] Among them, f a represents the anchor feature, f p represents the positive sample feature, f n represents the negative sample feature, s a represents the multi-view sample anchor, Φ represents the encoder, s p represents the positive sample, s n represents the negative sample;

[0147] Use the following formula to calculate the first loss value corresponding to the anchor feature, the positive sample feature, and the negative sample feature:

[0148]

[0149] L2 = ||Rec(f a ) - s a ||

[0150] Among them, L1 represents the loss value between the anchor feature and the positive sample feature and the negative sample feature in the first loss value, L2 represents the loss value between the anchor feature and the multi-view sample anchor in the first loss value, f a represents the anchor feature, f p represents the positive sample feature, f n represents the negative sample feature, s a represents the multi-view sample anchor, Rec represents the reconstruction model, and Rec is used to reconstruct f a back to the original image and ensure that the latent features extracted by the encoder Φ are complete;

[0151] Obtain the feature sets corresponding to the anchor feature, the positive sample feature, and the negative sample feature; in the preset decoder, use the following formula to perform feature decoding on the feature sets to obtain the decoded categories:

[0152] o i = Ψ(f i )

[0153] Among them, o i represents the decoded category corresponding to the i-th feature in the feature set, f i represents the i-th feature in the feature set, Ψ represents the decoder, and the decoder Ψ is composed of 2 fully connected layers. The decoder Ψ is used to directly map the latent features to the category space;

[0154] Use the following formula to calculate the second loss value corresponding to the decoded category:

[0155] L3 = Softmax(oi ,g i )

[0156] Among them, L3 represents the second loss value, o i Indicates the decoding category corresponding to the i-th feature in the feature set, g i Indicates o i Corresponding label tag;

[0157] The encoder and the decoder are trained based on the first loss value and the second loss value to obtain a trained encoder and a trained decoder.

[0158] Among them, the network structure of encoder Φ can explore the best structure in practical applications, which can be the mainstream ResNet, the latest ViT, or a simple 4-layer convolutional neural network. The patent of this invention aims to propose a universal algorithm, and the network structure can be arbitrarily designed. The network structure of Rec is the same as that of encoder Φ.

[0159] It should be noted that It is used to shorten the distance between the anchor point and the positive sample feature and to increase the distance between the anchor point and the negative sample. L2=||Rec(f a )-s a ||Used to ensure that the feature encoder has a complete understanding of the sample, L3

[0160] =Softmax(o i ,g i ) is used for end-to-end learning. The first loss value is first used to optimize the encoder parameters to obtain more discriminative features. These features can then be mapped to the final degree of corrosion through the decoder. In this process, only the decoder needs to be learned. This is an end-to-end learning process. The output of the decoder is compared with the manually labeled labels, and the loss value of the comparison result is back-propagated to learn the decoder.

[0161] See Figure 2 , which is a schematic diagram of an encoder for an online detection method for copper sheet corrosion of petroleum products based on computer vision provided by an embodiment of the present invention.

[0162] See Figure 3 , which is a schematic diagram of a decoder for an online detection method for copper corrosion of petroleum products based on computer vision provided by an embodiment of the present invention.

[0163] Furthermore, the embodiment of the present invention uses the trained encoder and the trained decoder to perform corrosion detection on the image acquisition points using the acquired image, so as to reduce the amount of corrosion detection data of the copper sheet.

[0164] Optionally, the process of using the collected image to perform corrosion detection on the image acquisition point to obtain the corrosion detection result of the test copper sheet refers to the process of performing image recognition using an encoder and a decoder with optimized parameters, so as to obtain the corrosion category. Similar to the aforementioned corrosion severity, it includes uncorroded, slightly corroded, severely corroded, etc.

[0165] Compared with the problems in the background art, in the embodiment of the present invention, by establishing a concentration change model of the microscopic data acquisition point according to the pre-historical data, it is used to identify whether corrosion has occurred on the copper sheet surface by using the change in the petroleum concentration on the test copper sheet. If it is calculated that the copper sheet concentration is relatively reduced compared to the standard version of the copper sheet concentration, it indicates that the copper sheet has corroded. In the embodiment of the present invention, by selecting corrosion abnormal points from the macro data acquisition points according to the microscopic data acquisition point and the corrosion diffusion degree, it is used to select a part and a small number of acquisition points from a large number of microscopic data acquisition points, thereby reducing the amount of corrosion detection data. Further, in the embodiment of the present invention, by using the collected image to perform corrosion detection on the image acquisition point to obtain the corrosion detection result of the test copper sheet, it is used to reduce the amount of corrosion detection data of the copper sheet. Therefore, the online copper sheet corrosion detection method for petroleum products based on computer vision proposed by the present invention can reduce the amount of data during copper sheet corrosion detection.

[0166] Embodiment 2:

[0167] As Figure 4 shown, it is a functional module diagram of an online copper sheet corrosion detection system for petroleum products based on computer vision according to the present invention.

[0168] The online copper sheet corrosion detection system 400 for petroleum products based on computer vision according to the present invention can be installed in an electronic device. According to the implemented functions, the online copper sheet corrosion detection system for petroleum products based on computer vision can include a microscopic acquisition module 401, a diffusion analysis module 402, a macroscopic acquisition module 403, a degree output module 404, and a corrosion detection module 405. The modules in the present invention can also be referred to as units, which refer to a series of computer program segments that can be executed by a processor of an electronic device and can complete fixed functions, and are stored in the memory of the electronic device.

[0169] In the embodiment of the present invention, the functions of each module / unit are as follows:

[0170] The microscopic acquisition module 401 is used to construct macroscopic data acquisition points of the test copper sheet, construct microscopic data acquisition points of the test copper sheet, and collect historical microscopic data and current microscopic data on the microscopic data acquisition points;

[0171] The diffusion analysis module 402 is configured to perform data preprocessing on the historical micro data and the current micro data respectively to obtain pre-historical data and pre-current data, establish a concentration change model of the micro data acquisition point according to the pre-historical data, and analyze the corrosion diffusion degree corresponding to the pre-current data through the concentration change model;

[0172] The macro acquisition module 403 is configured to select corrosion abnormal points from the macro data acquisition points according to the micro data acquisition point and the corrosion diffusion degree, and acquire historical macro data and current macro data at the corrosion abnormal points;

[0173] The degree output module 404 is configured to perform equal-interval processing on the historical macro data to obtain equal-length historical data, generate a grey model of the equal-length historical data, perform equal-interval processing on the current macro data to obtain equal-length current data, and output the corrosion severity corresponding to the equal-length current data through the grey model;

[0174] The corrosion detection module 405 is configured to select image acquisition points from the corrosion abnormal points based on the corrosion severity, perform image acquisition at the image acquisition points to obtain an acquired image, obtain a training image corresponding to the acquired image, perform image preprocessing on the training image to obtain a preprocessed image, generate a trained encoder and a trained decoder of the acquired image based on the preprocessed image, and perform corrosion detection on the image acquisition points by using the acquired image according to the trained encoder and the trained decoder to obtain the corrosion detection result of the test copper sheet.

[0175] Specifically, each module in the online corrosion detection system 300 of copper sheet for petroleum products based on computer vision in the embodiment of the present invention adopts the same technical means as those in the Figure 1 online corrosion detection method of copper sheet for petroleum products based on computer vision described above, and can produce the same technical effects, which will not be elaborated here.

[0176] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms.

[0177] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An on-line detection method for copper strip corrosion of petroleum products based on computer vision, characterized in that, The method includes: Construct macroscopic data collection points of the test copper sheet, and construct microscopic data collection points of the test copper sheet, and collect historical microscopic data and current microscopic data at the microscopic data collection points; Perform data preprocessing on the historical microscopic data and the current microscopic data respectively to obtain pre-historical data and pre-current data. According to the pre-historical data, establish a concentration change model of the microscopic data collection point, and analyze the corrosion diffusion degree corresponding to the pre-current data through the concentration change model; Select corrosion abnormal points from the macroscopic data collection points according to the microscopic data collection points and the corrosion diffusion degree, and collect historical macroscopic data and current macroscopic data at the corrosion abnormal points; Perform equal-interval processing on the historical macroscopic data to obtain equal-length historical data, generate a grey model of the equal-length historical data, perform equal-interval processing on the current macroscopic data to obtain equal-length current data, and output the corrosion severity corresponding to the equal-length current data through the grey model; Based on the corrosion severity, select image collection points from the corrosion abnormal points, perform image collection at the image collection points to obtain a collected image, obtain a training image corresponding to the collected image, perform image preprocessing on the training image to obtain a preprocessed image, based on the preprocessed image, generate a trained encoder and a trained decoder of the collected image, and use the trained encoder and the trained decoder to perform corrosion detection on the image collection points by using the collected image to obtain the corrosion detection result of the test copper sheet.

2. The on-line detection method for copper strip corrosion of petroleum products based on computer vision according to claim 1, wherein The construction of the macroscopic data collection points of the test copper sheet includes: Obtain the macroscopic data type of the test copper sheet; Based on the macroscopic data type, set the macroscopic data collection device of the test copper sheet; Query the number of ranges of the collection range of the macroscopic data collection device included in the range corresponding to the test copper sheet; Use the number of ranges to determine the number of devices of the macroscopic data collection device; According to the number of devices, generate random positions of the macroscopic data collection device within the range corresponding to the test copper sheet; When the coverage area of the macroscopic data collection device at the random position is not less than the range corresponding to the test copper sheet, use the random position as the macroscopic data collection point of the test copper sheet.

3. The online detection method for copper strip corrosion of petroleum products based on computer vision according to claim 1, characterized in that, The establishment of the concentration change model of the microscopic data collection point according to the pre-historical data includes: Query the microscopic concentration and the copper sheet concentration of the microscopic data collection point from the pre-historical data; Compare the copper sheet concentration with the microscopic concentration to obtain a concentration comparison result; Use the concentration comparison result to determine the diffusion parameter of the microscopic data collection point; According to the pre-historical data and the diffusion parameter, establish the concentration change model of the microscopic data collection point by using the following formula: Among them, represents the concentration change model, and s(z,t) represents the concentration at point z at time t. represents the concentration of points within the neighborhood range of point z in the pre-historical data at that point. represents the diffusion parameter between point z and the point. represents the volume of the point in the pre-historical data, and D z represents the neighborhood of point z.

4. The on-line detection method for copper strip corrosion of petroleum products based on computer vision according to claim 1, wherein The analysis of the corrosion diffusion degree corresponding to the pre-current data through the concentration change model includes: Analyze the current concentration value corresponding to the pre-current data by using the concentration change model; According to the current concentration value, calculate the corrosion diffusion degree corresponding to the pre-current data by using the following formula: Among them, represents the degree of corrosion diffusion, represents the current concentration value, s1 represents the standard concentration of the test copper sheet, and s2 represents the standard concentration of the petroleum.

5. The on-line detection method for copper sheet corrosion of petroleum products based on computer vision according to claim 1, characterized in that Selecting corrosion abnormal points from the macro data acquisition points according to the micro data acquisition points and the corrosion diffusion degree includes: Judging whether the corrosion diffusion degree is an abnormal diffusion degree; When the corrosion diffusion degree is an abnormal diffusion degree, query the first target acquisition point corresponding to the corrosion diffusion degree in the micro data acquisition points; Identify the second target acquisition point in the macro data acquisition points that covers the first target acquisition point; Take the second target acquisition point as the corrosion abnormal point.

6. The online detection method for copper strip corrosion of petroleum products based on computer vision according to claim 1, wherein Performing equidistant processing on the historical macro data to obtain equi-length historical data includes: Extract the macro data time of the historical macro data; perform time sorting on the macro data time to obtain a data time sequence; calculate the standard time interval of the data time sequence by using the following formula: where Δt represents the standard time interval, m represents the length of the data time series, t m represents the maximum time in the data time series, and t1 represents the minimum time in the data time series; According to the standard time interval, perform equidistant processing on the macro data sequence corresponding to the data time sequence by using the following formula to obtain equi-length historical data: Among them, y(t j ) represents the equal-length historical data, and x(t j ) represents the historical macro data at time t in the macro data sequence. j represents the serial number of the time in the data time sequence. t j represents the j-th moment in the data time sequence. x(t j ) represents x(t j -1) represents the predecessor of x(t j ) in the macro data sequence, and Δt represents the standard time interval.

7. The on-line detection method for copper strip corrosion of petroleum products based on computer vision according to claim 1, wherein Generating the grey model of the equi-length historical data includes: Perform data accumulation on the equi-length historical data by using the following formula to obtain the accumulated historical data: Among them, y (1) (k) represents the accumulated historical data, y (0) (i) represents the i-th data in the equal-length historical data, n represents the number of values in the same type of the equal-length historical data, and k represents the serial number less than n; Construct the grey model of the equi-length historical data according to the accumulated historical data by using the following formula: Among them, represents the grey model, where a and u in U represent unknown parameters, U represents the matrix composed of unknown parameters, and y (0) (2), y (0) (3), y (0) (n) represents the values of y (0) (i) when taking different i, and Y n represents the matrix of equal-length historical data composed of y (0) (2), y (0) (3), y (0) (n), and y (1) (1), y (1) (2), y (1) (3), y (1) (n - 1), y (1) (n) represents the values of y (1) (k) when taking different k, B represents the matrix composed of the accumulated historical data, U represents the calculation result of U, a represents the calculation result of a, u represents the calculation result of u, and y (1) (t + 1) represents the value of the accumulated historical data in the next time period, and y (1) (t) represents any type of data in the accumulated historical data, and t represents the time.

8. The online detection method for copper strip corrosion of petroleum products based on computer vision according to claim 1, characterized in that Outputting the corrosion severity corresponding to the equi-length current data through the grey model includes: According to the grey model, calculate the planned current data of the equi-length historical data corresponding to the grey model; Query the historical corrosion degree corresponding to the equi-length historical data; Construct a grey analysis model of the historical corrosion degree; Judge whether the planned current data is consistent with the equi-length current data; When the planned current data is consistent with the equi-length current data, use the grey analysis model to output the corrosion severity corresponding to the equi-length current data; When the planned current data is inconsistent with the equi-length current data, take copper sheet severe corrosion as the corrosion severity corresponding to the equi-length current data.

9. The on-line detection method for copper strip corrosion of petroleum products based on computer vision according to claim 1, characterized in that, Generating the trained encoder and decoder of the acquisition image based on the preprocessed image includes: Divide the preprocessed image into multi-view sample anchors, positive samples and negative samples; In a preset encoder, map the multi-view sample anchors, the positive samples and the negative samples to the feature latent space respectively by using the following formula to obtain anchor features, positive sample features and negative sample features: f a = Φ(s a ), f p = Φ(s p ), f n = Φ(s n ) Among them, f a represents the anchor feature, f p represents the positive sample feature, f n represents the negative sample feature, s a represents the multi-view sample anchor, Φ represents the encoder, s p represents the positive sample, s n represents the negative sample; Calculate the first loss value corresponding to the anchor features, the positive sample features and the negative sample features by using the following formula: L2 = ||Rec(f a ) - s a || Among them, L1 represents the loss values between the anchor feature in the first loss value and the positive sample feature and the negative sample feature respectively, L2 represents the loss value between the anchor feature in the first loss value and the multi-view sample anchor, f a represents the anchor feature, f p represents the positive sample feature, f n represents the negative sample feature, s a represents the multi-view sample anchor, Rec represents the reconstruction model, and Rec is used to reconstruct f a back to the original image and ensure that the latent features extracted by the encoder Φ are complete; Obtain the feature set corresponding to the anchor features, the positive sample features and the negative sample features; in a preset decoder, perform feature decoding on the feature set by using the following formula to obtain the decoded category: o i = Ψ(f i ) Among them, o i represents the decoded category corresponding to the i-th feature in the feature set, and f i represents the i-th feature in the feature set. Ψ represents the decoder, and the decoder Ψ is composed of two fully connected layers. The decoder Ψ is used to directly map the hidden features to the category space; Calculate the second loss value corresponding to the decoded category by using the following formula: L3 = Softmax(o i , g i ) Among them, L3 represents the second loss value, o i represents the decoded category corresponding to the i-th feature in the feature set, g i represents o i corresponding label tag; Train the encoder and the decoder based on the first loss value and the second loss value to obtain the trained encoder and the trained decoder.

10. An on-line detection system for copper strip corrosion of petroleum products based on computer vision, characterized in that, The system includes: A microscopic acquisition module, configured to construct macroscopic data acquisition points of a test copper sheet, construct microscopic data acquisition points of the test copper sheet, and acquire historical microscopic data and current microscopic data at the microscopic data acquisition points; A diffusion analysis module, configured to perform data preprocessing on the historical microscopic data and the current microscopic data respectively to obtain pre-historical data and pre-current data, establish a concentration change model of the microscopic data acquisition point according to the pre-historical data, and analyze the corrosion diffusion degree corresponding to the pre-current data through the concentration change model; A macroscopic acquisition module, configured to select corrosion abnormal points from the macroscopic data acquisition points according to the microscopic data acquisition points and the corrosion diffusion degree, and acquire historical macroscopic data and current macroscopic data at the corrosion abnormal points; A degree output module, configured to perform equally-spaced processing on the historical macroscopic data to obtain equally-long historical data, generate a grey model of the equally-long historical data, perform equally-spaced processing on the current macroscopic data to obtain equally-long current data, and output the corrosion severity corresponding to the equally-long current data through the grey model; A corrosion detection module, configured to select image acquisition points from the corrosion abnormal points based on the corrosion severity, perform image acquisition at the image acquisition points to obtain an acquired image, obtain a training image corresponding to the acquired image, perform image preprocessing on the training image to obtain a preprocessed image, generate a trained encoder and a trained decoder of the acquired image based on the preprocessed image, and perform corrosion detection on the image acquisition points using the acquired image according to the trained encoder and the trained decoder to obtain a corrosion detection result of the test copper sheet.

Images