Real-time monitoring method of hard carbon roller furnace for sodium ion batteries based on machine vision

Abstract

The present invention discloses a real-time monitoring method for a sodium ion battery hard carbon roller furnace based on machine vision, comprising the following steps: obtaining furnace combustion data; combining the material flow image with an image processing algorithm to identify material flow characteristics; building a hard carbon combustion degree evaluation model in combination with the roller furnace operating parameters and material flow characteristics to obtain a hard carbon combustion degree value corresponding to each material flow image; calculating a hard carbon sintering effect value based on the roller furnace operating parameters and the hard carbon combustion degree value, and establishing a hard carbon sintering process effect data group; building a temperature-transport rate-sintering effect surface based on the established hard carbon sintering process effect data group, and finding the optimal furnace combustion environment temperature and optimal material conveying speed corresponding to the maximum hard carbon sintering effect value in the temperature-transport rate-sintering effect surface. The present invention solves the problem of strong instability of material flow in the prior art.

Description

Technical Field

[0001] The invention relates to the technical field of real-time monitoring of a sodium ion battery hard carbon roller furnace, and in particular to a real-time monitoring method of a sodium ion battery hard carbon roller furnace based on machine vision. Background Art

[0002] With the growing demand for energy storage, especially its widespread application in renewable energy, smart grids, electric vehicles and other fields, the research and development of battery technology has received more and more attention. In this context, sodium-ion battery (SIB), as a new battery technology with cost-effectiveness and resource advantages, is gradually becoming a strong competitor to lithium-ion batteries. Sodium-ion batteries not only have lower production costs in theory, but also because of the wide distribution of sodium elements, they avoid the risks brought by lithium resource shortages and price fluctuations, showing strong market potential.

[0003] Machine vision is a technology that uses cameras or sensors to capture image data, obtains images or videos through image sensors (such as cameras), and extracts useful information through image processing algorithms (such as image recognition, target detection, image segmentation, etc.). Sodium-ion battery (SIB) is a new type of battery technology that uses sodium ions as charge carriers. Compared with traditional lithium-ion batteries, it has the advantages of lower cost and richer resources. Real-time monitoring refers to the real-time collection of various types of data in the production process through various sensors or monitoring equipment, and processing and analysis, timely feedback of information and necessary adjustments.

[0004] For example, the invention patent with the publication number CN116309252A discloses an intelligent analysis method for an aircraft engine rotor borescope video, which includes a data acquisition module, a region selection module, a direction determination module, a hash similarity calculation module, a full-circle rotation determination module, a tiling merging module, and a coating peeling ratio calculation module. The present invention automatically determines the direction of movement of the rotor in the rotor video, and uses normalized hash similarity to determine whether the rotor rotates one circle. This process is determined and calculated by the computer itself, with a high degree of automation and high accuracy. The coating peeling ratio obtained by directly using the manual marking method to count the coating peeling ratio is very accurate, far exceeding the recognition accuracy of the traditional method relying on visual estimation.

[0005] For example, the invention patent with the publication number of CN119048825A discloses a pathological image classification method based on state space duality. The invention proposes a computational pathology multi-instance learning (MIL) method based on state space duality (SSD), and proposes a new Mamba2MIL model, which can comprehensively utilize the sequential dependency and sequential independence features in WSIs to improve the classification accuracy of pathological images. The present invention relates to the field of computational pathology, and in particular to a multi-instance learning method based on state space duality (State Space Duality, SSD) for automatic analysis and diagnosis of whole slide images (WSIs). The present invention belongs to the field of image data processing (G06T) and medical diagnosis (A61B). The model uses preprocessing, feature extraction, serialization processing, and the application of the state space duality model, and finally makes classification decisions through a multi-layer perceptron, effectively processing long sequence data and improving the accuracy of diagnosis. Experiments have shown that the Mamba2MIL method has important application value in improving the efficiency of pathological image analysis and diagnostic accuracy, providing a valuable auxiliary tool for pathologists and has broad clinical application prospects.

[0006] Hard carbon is a common negative electrode material in sodium-ion batteries. It has good electrical conductivity, cycle stability and high specific capacity. In the production process of hard carbon for sodium-ion batteries, roller furnaces are used to sinter hard carbon materials. Machine vision technology can monitor the flow of materials, temperature changes and the environment in the roller furnace in real time to ensure the stability and uniformity of the sintering process, thereby improving the quality and performance of the battery. For the production process of hard carbon for sodium-ion batteries in roller furnaces, real-time monitoring can ensure that the production environment meets the process requirements and prevent factors such as uneven temperature and unstable raw materials from affecting production quality.

[0007] The roller hearth furnace allows the hard carbon material to be heat treated at high temperatures to ensure good battery performance. However, current technology has problems with the material flow in the roller hearth furnace, which may result in over-sintering of the hard carbon material in some areas and under-sintering in other areas.

[0008] In the prior art, there is a problem of strong instability of material flow. Summary of the invention

[0009] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a real-time monitoring method for a sodium ion battery hard carbon roller furnace based on machine vision. The flow state of the material on the roller can be monitored in real time through machine vision technology. The operating parameters of the roller furnace can be adjusted by combining data such as material morphology, particle size and flow speed to solve the problem of strong instability of material flow in the prior art.

[0010] To achieve the above object, the present invention provides the following technical solutions:

[0011] A real-time monitoring method for a hard carbon roller furnace for sodium ion batteries based on machine vision comprises the following steps: obtaining furnace combustion data under different combustion environments, the furnace combustion data comprising roller furnace operation parameters and material flow images, the roller furnace operation parameters comprising the roller furnace operation parameters comprising furnace combustion environment temperature and material conveying speed to the material flow image; identifying material flow characteristics in combination with an image processing algorithm, the material flow characteristics comprising material morphology edge characteristics and material particle distribution characteristics; constructing a hard carbon combustion degree evaluation model in combination with roller furnace operation parameters and material flow characteristics, and obtaining a hard carbon combustion degree value corresponding to each material flow image; calculating a hard carbon sintering effect value according to the roller furnace operation parameters and the hard carbon combustion degree value, and establishing a hard carbon sintering process effect data group; constructing a temperature-conveying rate-sintering effect surface according to the established hard carbon sintering process effect data group, and finding an optimal furnace combustion environment temperature and an optimal material conveying speed corresponding to the maximum hard carbon sintering effect value in the temperature-conveying rate-sintering effect surface.

[0012] In a preferred embodiment, the material morphology edge features are obtained by the following method: the acquired material flow image is converted into grayscale and denoised using a Gaussian filter; the gradient value and direction of each pixel in the material flow image after grayscale conversion are calculated to obtain an edge strength-direction gradient map; the edge strength-direction gradient map is subjected to non-maximum gradient value suppression to obtain a local maximum value in the edge; the gradient value of the local maximum value in the edge is compared with a preset double threshold value to obtain the material morphology edge features.

[0013] In a preferred embodiment, non-maximum gradient value suppression is performed on the edge strength-directional gradient map to obtain a local maximum value in the edge, specifically: the pixel gradient value of each pixel is compared with the pixel gradient values ​​of neighboring pixels in the gradient direction; the pixel with the largest pixel gradient value among the neighboring pixels is obtained, the pixel is marked as the first edge pixel, and the pixel gradient values ​​of other compared pixels are suppressed to zero.

[0014] In a preferred embodiment, the gradient value of the local maximum value in the edge is compared with a preset double threshold value to obtain the material morphology edge feature, specifically as follows: pixels with gradient values ​​higher than a high threshold value are classified as strong edge pixels, and the strong edge pixels are directly marked as second edge pixels; pixels with gradient values ​​between a high threshold value and a low threshold value are classified as weak edge pixels, and the weak edge pixels are connected to the second edge pixels. If the connection is successful, the weak edge pixels are also marked as second edge pixels. If the connection fails, the pixel gradient value of the weak edge pixels is suppressed to zero; pixels with gradient values ​​lower than a low threshold value are classified as non-edge pixels, and the pixel gradient value of the non-edge pixels is suppressed to zero; the second edge pixel value obtained after comparison is used as the material morphology edge feature.

[0015] In a preferred embodiment, the material particle distribution characteristics are obtained by the following method: the acquired material flow image is gray-converted, and the converted material flow image is divided into equal areas, and the number of pixels in each area is calculated; the gray values ​​of several pixels are calculated one by one, and a gray-scale histogram is established; a global optimal threshold is preset, and the material flow image of each area is divided into two categories: foreground and background; the material flow image classified as foreground is binarized to obtain the material particle distribution characteristics.

[0016] In a preferred embodiment, the material flow image classified as the foreground is binarized to obtain the material particle distribution characteristics, specifically: the grayscale values ​​of the pixels in the area are compared one by one, if the grayscale value of the pixel is less than the preset global optimal threshold, the grayscale value of the pixel is set to zero; if the grayscale value of the pixel is greater than the preset global optimal threshold, the grayscale value of the pixel is set to 1 and marked as the location of the particle; the ratio of the particle location to all locations in the area is calculated as the material particle distribution characteristic.

[0017] In a preferred embodiment, a temperature-conveying rate-sintering effect surface is constructed according to the established hard carbon sintering process effect data group, and in the temperature-conveying rate-sintering effect surface, the optimal furnace combustion environment temperature and the optimal material conveying speed corresponding to the maximum hard carbon sintering effect value are found, specifically: the x-axis variable of the temperature-conveying rate-sintering effect surface is set to the furnace combustion environment temperature; the y-axis variable of the temperature-conveying rate-sintering effect surface is set to the material conveying speed; the z-axis variable of the temperature-conveying rate-sintering effect surface is set to the hard carbon sintering effect value; the surface constraints are preset, and the x-axis and y-axis of the temperature-conveying rate-sintering effect surface that do not meet the surface constraints are cut off to obtain the cut temperature-conveying rate-sintering effect surface; the optimal furnace combustion environment temperature and the optimal material conveying speed corresponding to the maximum hard carbon sintering effect value are found in the cut surface, and the furnace combustion environment temperature and the material conveying speed are set to the optimal furnace combustion environment temperature and the optimal material conveying speed during the hard carbon combustion process.

[0018] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0019] 1. Through the real-time monitoring method of the sodium ion battery hard carbon roller furnace based on machine vision, the combustion data in the furnace is combined with the image analysis of material flow, thereby realizing the real-time monitoring and accurate evaluation of the combustion degree of hard carbon, effectively solving the problem of strong instability of material flow in the existing technology.

[0020] 2. By adopting image processing algorithms, the material morphology edge features and particle distribution in the material flow image can be identified, thereby achieving the extraction of more detailed material features.

[0021] 3. By utilizing technologies such as grayscale conversion, edge detection, gradient calculation and image binarization, the monitoring accuracy of the combustion process is further improved.

[0022] 4. By establishing the "temperature-conveying rate-sintering effect surface" and combining the hard carbon combustion degree and actual operation constraints, the optimal furnace combustion environment temperature and material conveying speed can be accurately found, thereby maximizing the hard carbon sintering effect, improving the performance of sodium-ion batteries, and reducing energy waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A flow chart of a method for real-time monitoring of a sodium ion battery hard carbon roller furnace based on machine vision provided in an embodiment of the present application; DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0026] Embodiment 1, Figure 1 The present invention provides a real-time monitoring method for a sodium ion battery hard carbon roller furnace based on machine vision, comprising the following steps:

[0027] S1, acquiring furnace combustion data under different combustion environments, wherein the furnace combustion data includes operating parameters in the roller hearth furnace and a material flow image, wherein the operating parameters in the roller hearth furnace include the furnace combustion environment temperature and the material conveying speed;

[0028] S2, combining the material flow image with an image processing algorithm to identify material flow characteristics, wherein the material flow characteristics include material morphology edge characteristics and material particle distribution characteristics;

[0029] S3, combining the operating parameters and material flow characteristics in the roller furnace to build a hard carbon combustion degree evaluation model, and obtain the hard carbon combustion degree value corresponding to each material flow image;

[0030] S4, calculating the hard carbon sintering effect value according to the roller furnace operation parameters and the hard carbon combustion degree value, and establishing a hard carbon sintering process effect data group;

[0031] S5, constructing a temperature-conveying rate-sintering effect surface according to the established hard carbon sintering process effect data set, and finding the optimal furnace combustion environment temperature and the optimal material conveying rate corresponding to the maximum hard carbon sintering effect value in the temperature-conveying rate-sintering effect surface.

[0032] The operating parameters in the roller furnace include the combustion environment temperature in the furnace and the material conveying speed;

[0033] The initial format of the material flow image is a color image, which is collected by a high-resolution camera combined with an infrared thermal imaging device.

[0034] The material flow image is combined with an image processing algorithm to identify material flow characteristics, which include material morphology edge characteristics and material particle distribution characteristics, specifically including:

[0035] The material morphology edge features are obtained as follows:

[0036] The acquired material flow image is converted into grayscale and denoised using a Gaussian filter;

[0037] Calculate the gradient value and direction of each pixel in the material flow image after grayscale conversion to obtain an edge strength-direction gradient map;

[0038] Performing non-maximum gradient value suppression on the edge strength-direction gradient map to obtain a local maximum value in the edge;

[0039] The gradient value of the local maximum value in the edge is compared with the preset double threshold value to obtain the material morphology edge feature.

[0040] The edge strength-direction gradient map is subjected to non-maximum gradient value suppression to obtain the local maximum value in the edge, specifically:

[0041] Compare the pixel gradient value of each pixel with the pixel gradient values ​​of neighboring pixels in the gradient direction;

[0042] A pixel with the largest pixel gradient value among neighboring pixels is obtained, the pixel is marked as a first edge pixel, and the pixel gradient values ​​of other compared pixels are suppressed to zero.

[0043] The gradient value of the local maximum value in the edge is compared with the preset double threshold value to obtain the edge feature of the material morphology, which is as follows:

[0044] Classify pixels with gradient values ​​higher than a high threshold as strong edge pixels, and directly mark the strong edge pixels as second edge pixels;

[0045] Classifying pixels whose gradient values ​​are between a high threshold and a low threshold as weak edge pixels, connecting the weak edge pixels with the second edge pixels, and if the connection is successful, marking the weak edge pixels as second edge pixels; if the connection fails, suppressing the pixel gradient value of the weak edge pixels to zero;

[0046] Classify pixels with gradient values ​​lower than a low threshold as non-edge pixels, and suppress the pixel gradient values ​​of non-edge pixels to zero;

[0047] The second edge pixel value obtained after comparison is used as the edge feature of the material shape.

[0048] The calculation formula for obtaining the edge features of the material shape is as follows:

[0049]

[0050] Where: is the edge feature of the material shape, is a strong edge pixel with a gradient value higher than the high threshold, It is a weak edge pixel that is successfully connected to a strong edge pixel.

[0051] Local maximum suppression ensures that only possible edge pixels are considered before double threshold processing. Only when a pixel is suppressed by the local maximum and retained can it enter the double threshold classification stage. Through double threshold processing, it can be determined which local maxima (possible edges) are actually retained. Local maximum suppression is to refine the edge image, and only the most significant edge points are considered, while double threshold processing is to further screen these edge points to determine which are strong edge pixels and which are weak edge pixels, thereby clearly defining the final material morphology edge feature structure.

[0052] The material particle distribution characteristics are obtained as follows:

[0053] The acquired material flow image is converted into grayscale, and the converted material flow image is divided into equal areas, and the number of pixels in each area is calculated;

[0054] Calculate the grayscale values ​​of several pixels one by one and establish a grayscale histogram;

[0055] The global optimal threshold is preset to divide the material flow image of each area into two categories: foreground and background;

[0056] The material flow image classified as foreground is binarized to obtain the material particle distribution characteristics.

[0057] Grayscale histogram, defined as follows:

[0058]

[0059] Where: represents the number of pixels with gray level g, is the height of the material flow image, is the width of the material flow image, is the indicator function.

[0060] It should be noted that when the exponential function value is the corresponding gray level, is 1 if the value is true, otherwise it is 0.

[0061] The material flow image classified as the foreground is binarized to obtain the material particle distribution characteristics, which are specifically:

[0062] Compare the grayscale values ​​of pixels in the area one by one. If the grayscale value of a pixel is less than a preset global optimal threshold, set the grayscale value of the pixel to zero.

[0063] If the gray value of a pixel is greater than the preset global optimal threshold, the gray value of the pixel is set to 1 and marked as the location of the particle;

[0064] The ratio of the position of the particle to all positions in the area is calculated as the material particle distribution characteristic.

[0065] The material particle distribution characteristics are calculated as follows:

[0066]

[0067] Where: is the particle distribution characteristic of the material, is the total number of pixels in the i-th region, is the preset global optimal threshold, is the gray value of the pixel in the area where the image is binarized, is the gray value of the pixel compared with the preset global optimal threshold, and Q is the number of regions.

[0068] In image processing, grayscale images use grayscale levels (usually integer values ​​between 0 and 255) to represent the brightness or intensity of pixels. Each pixel has a corresponding grayscale level. The grayscale level represents the grayscale value of each pixel in the image, usually an integer. A grayscale value of 0 represents complete black, 255 represents complete white, and values ​​between the two represent different grayscale levels.

[0069] A grayscale histogram is a graphical representation that shows the frequency of each gray level in an image. The horizontal axis of the grayscale histogram represents the gray level, and the vertical axis represents the number of pixels with the corresponding gray level in the image. The number of grayscale values ​​is usually between 0 and 255.

[0070] When the grayscale value of the image is divided into foreground and background by the global optimal threshold, the variance between the two categories is the largest; it should be noted that the larger the inter-class variance, the more obvious the difference between the foreground and the background, and the better the segmentation effect.

[0071] Image binarization is to convert each pixel in the image into two values ​​by applying a threshold to the grayscale image, which is used to further and more clearly distinguish the specific distribution position of particles in the image classified as the foreground area.

[0072] Hard carbon undergoes morphological changes during combustion, so the edge characteristics can reflect the extent of combustion, while the particle distribution may change with changes in temperature and the advancement of the combustion process. The higher the degree of combustion, the smaller or more evenly distributed the particles may be.

[0073] The hard carbon combustion degree evaluation model is specifically:

[0074]

[0075] Where: For the hard carbon combustion degree assessment model, is the combustion environment temperature in the furnace, It is expressed as the coefficient factor of the influence of temperature on combustion, which is a constant; is the edge feature of the material shape, is the particle distribution characteristic of the material, is the material conveying speed, is the error term.

[0076] It should be noted that temperature directly affects the combustion reaction rate and combustion degree of hard carbon; the flow rate of the material in the roller furnace affects the contact time with the high-temperature airflow, and thus affects the completion of the combustion reaction. Too fast or too slow material speed will affect the combustion efficiency; edge information can reflect the combustion changes of the material, especially during the combustion process, the shape and edge characteristics of the material will change significantly with the change of the combustion degree; particle distribution information can reflect the thermal distribution of the material in the furnace, and thus affect the combustion degree of hard carbon.

[0077] For each material flow image, the material morphology edge features and material particle distribution features are extracted through image processing technology to obtain the current combustion temperature and material conveying speed in the furnace. These features are input into the trained hard carbon combustion degree evaluation model. The output of the model is the combustion degree of hard carbon, which indicates the degree of completion of hard carbon combustion.

[0078] The hard carbon sintering effect value is calculated according to the operating parameters in the roller furnace and the hard carbon combustion degree value, specifically:

[0079]

[0080] Where: is the hard carbon sintering effect value corresponding to the i-th material flow image, is the hard carbon combustion degree value, is the combustion environment temperature in the furnace, is the standard temperature obtained under ideal laboratory conditions. It is expressed as the coefficient factor of the influence of temperature on combustion, which is a constant; is the material conveying speed, It is the standard material delivery rate obtained under ideal laboratory conditions.

[0081] The hard carbon sintering process effect data set is defined as follows:

[0082]

[0083] Where: is the hard carbon sintering process effect data set, is the combustion environment temperature in the first group of furnaces, is the conveying speed of the first group of materials, is the hard carbon combustion degree value corresponding to the first material flow image, and n is the total number of hard carbon combustion degree values ​​corresponding to the material flow images.

[0084] The x-axis of the temperature-conveying rate-sintering effect surface is the combustion environment temperature in the furnace, the y-axis of the temperature-conveying rate-sintering effect surface is the material conveying speed, and the z-axis of the temperature-conveying rate-sintering effect surface is the hard carbon sintering effect value.

[0085] In order to ensure the practical feasibility of the curved surface, some constraints in actual operation need to be introduced, including the given maximum temperature and minimum temperature, the fastest conveying rate and the slowest conveying rate for cutting.

[0086] If the temperature is too high, it may cause excessive gasification or carbonization during the hard carbon sintering process, forming unstable carbon materials and even damaging battery performance. In addition, too high a temperature may also lead to energy waste and increase production costs. Too low a temperature will lead to incomplete sintering, and the pore structure, particle size distribution and other characteristics of the hard carbon will not reach the ideal level, affecting the electrochemical performance of the sodium ion battery.

[0087] If the material flow rate is too fast, the sintering time will be shortened, which may lead to incomplete sintering. The material may not be kept at the appropriate temperature for a sufficient time, affecting the structure and particle size distribution of the hard carbon. If the material flow rate is too slow, the material will be overexposed to the high temperature environment, resulting in excessive sintering or coking, which may affect the performance of the hard carbon and even cause excessive deterioration of some materials.

[0088] The constraint formula is as follows:

[0089]

[0090] Where: is the lower limit of temperature, The upper limit of temperature.

[0091]

[0092] Where: is the lower limit of material conveying speed, It is the upper limit of material conveying speed.

[0093] Cut off the x-axis and y-axis of the temperature-conveying rate-sintering effect surface that do not meet the above constraints, find the optimal furnace combustion environment temperature and optimal material conveying speed corresponding to the maximum hard carbon sintering effect value in the cut surface, and set the furnace combustion environment temperature and material conveying speed to the optimal furnace combustion environment temperature and optimal material conveying speed during the hard carbon combustion process.

[0094] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0095] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0096] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0097] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0098] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0099] Finally: 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, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A real-time monitoring method for a sodium ion battery hard carbon roller furnace based on machine vision, characterized in that: The following steps are involved: Acquire furnace combustion data under different combustion environments, wherein the furnace combustion data includes operating parameters in the roller furnace and a material flow image, wherein the operating parameters in the roller furnace include the furnace combustion environment temperature and the material conveying speed to the material flow image; In combination with image processing algorithms, the material flow characteristics are identified, wherein the material flow characteristics include material morphology edge characteristics and material particle distribution characteristics; A hard carbon combustion degree evaluation model is constructed by combining the operating parameters and material flow characteristics in the roller hearth furnace, and the hard carbon combustion degree value corresponding to each material flow image is obtained; According to the operating parameters in the roller furnace and the hard carbon combustion degree value, the hard carbon sintering effect value is calculated, and a hard carbon sintering process effect data group is established; A temperature-conveying rate-sintering effect surface is constructed according to the established hard carbon sintering process effect data set, and the optimal furnace combustion environment temperature and the optimal material conveying rate corresponding to the maximum hard carbon sintering effect value are found in the temperature-conveying rate-sintering effect surface.

2. The real-time monitoring method of a sodium ion battery hard carbon roller furnace based on machine vision as claimed in claim 1, characterized in that: The material morphology edge features are obtained as follows: The acquired material flow image is converted into grayscale and denoised using a Gaussian filter; Calculate the gradient value and direction of each pixel in the material flow image after grayscale conversion to obtain an edge strength-direction gradient map; Performing non-maximum gradient value suppression on the edge strength-direction gradient map to obtain a local maximum value in the edge; The gradient value of the local maximum value in the edge is compared with the preset double threshold value to obtain the material morphology edge feature.

3. The real-time monitoring method of a sodium ion battery hard carbon roller furnace based on machine vision as claimed in claim 2, characterized in that: The edge strength-direction gradient map is subjected to non-maximum gradient value suppression to obtain the local maximum value in the edge, specifically: Compare the pixel gradient value of each pixel with the pixel gradient values ​​of neighboring pixels in the gradient direction; A pixel with the largest pixel gradient value among neighboring pixels is obtained, the pixel is marked as a first edge pixel, and the pixel gradient values ​​of other compared pixels are suppressed to zero.

4. The real-time monitoring method of a sodium ion battery hard carbon roller furnace based on machine vision as claimed in claim 3, characterized in that: The gradient value of the local maximum value in the edge is compared with the preset double threshold value to obtain the edge feature of the material morphology, which is as follows: Classify pixels with gradient values ​​higher than a high threshold as strong edge pixels, and directly mark the strong edge pixels as second edge pixels; Classifying pixels whose gradient values ​​are between a high threshold and a low threshold as weak edge pixels, connecting the weak edge pixels with the second edge pixels, and if the connection is successful, marking the weak edge pixels as second edge pixels; if the connection fails, suppressing the pixel gradient value of the weak edge pixels to zero; Classify pixels with gradient values ​​lower than a low threshold as non-edge pixels, and suppress the pixel gradient values ​​of non-edge pixels to zero; The second edge pixel value obtained after comparison is used as the edge feature of the material shape.

5. The real-time monitoring method of a sodium ion battery hard carbon roller furnace based on machine vision as claimed in claim 4, characterized in that: The material particle distribution characteristics are obtained as follows: The acquired material flow image is converted into grayscale, and the converted material flow image is divided into equal areas, and the number of pixels in each area is calculated; Calculate the grayscale values ​​of several pixels one by one and establish a grayscale histogram; The global optimal threshold is preset to divide the material flow image of each area into two categories: foreground and background; The material flow image classified as foreground is binarized to obtain the material particle distribution characteristics.

6. The method for real-time monitoring of a sodium ion battery hard carbon roller furnace based on machine vision as claimed in claim 5, characterized in that: The material flow image classified as the foreground is binarized to obtain the material particle distribution characteristics, which are specifically: Compare the grayscale values ​​of pixels in the area one by one. If the grayscale value of a pixel is less than a preset global optimal threshold, set the grayscale value of the pixel to zero. If the gray value of a pixel is greater than the preset global optimal threshold, the gray value of the pixel is set to 1 and marked as the location of the particle; The ratio of the position of the particle to all positions in the area is calculated as the material particle distribution characteristic.

7. The real-time monitoring method of a sodium ion battery hard carbon roller furnace based on machine vision as claimed in claim 6, characterized in that: The hard carbon combustion degree evaluation model is specifically: Where: For the hard carbon combustion degree assessment model, is the combustion environment temperature in the furnace, It is expressed as the coefficient factor of the influence of temperature on combustion, which is a constant; is the edge feature of the material shape, is the particle distribution characteristic of the material, is the material conveying speed, is the error term.

8. The method for real-time monitoring of a sodium ion battery hard carbon roller furnace based on machine vision as claimed in claim 7, characterized in that: The hard carbon sintering effect value is calculated according to the operating parameters in the roller furnace and the hard carbon combustion degree value, specifically: Where: is the hard carbon sintering effect value corresponding to the i-th material flow image, is the hard carbon combustion degree value, is the combustion environment temperature in the furnace, is the standard temperature obtained under ideal laboratory conditions. It is expressed as the coefficient factor of the influence of temperature on combustion, which is a constant; is the material conveying speed, It is the standard material delivery rate obtained under ideal laboratory conditions.

9. The method for real-time monitoring of a sodium ion battery hard carbon roller furnace based on machine vision as claimed in claim 8, characterized in that: The temperature-conveying rate-sintering effect surface is constructed according to the established hard carbon sintering process effect data group. In the temperature-conveying rate-sintering effect surface, the optimal furnace combustion environment temperature and the optimal material conveying speed corresponding to the maximum hard carbon sintering effect value are found, specifically: Set the x-axis variable of the temperature-transport rate-sintering effect surface to the combustion environment temperature in the furnace; Set the y-axis variable of the temperature-conveying rate-sintering effect surface to the material conveying speed; Set the z-axis variable of the temperature-transport rate-sintering effect surface to the hard carbon sintering effect value; Preset a surface constraint condition, cut off the x-axis of the temperature-conveying rate-sintering effect surface and the y-axis of the temperature-conveying rate-sintering effect surface that do not meet the surface constraint condition, and obtain a cut temperature-conveying rate-sintering effect surface; The optimal furnace combustion environment temperature and the optimal material conveying speed corresponding to the maximum hard carbon sintering effect value are found in the cut surface, and the furnace combustion environment temperature and the material conveying speed are set to the optimal furnace combustion environment temperature and the optimal material conveying speed during the hard carbon combustion process.

10. The method for real-time monitoring of a sodium ion battery hard carbon roller furnace based on machine vision as claimed in claim 9, characterized in that: The surface constraint conditions are specifically: Where: is the lower limit of temperature, is the upper limit of temperature; Where: is the lower limit of material conveying speed, It is the upper limit of material conveying speed.

Images