Monitoring method and system for manufacturing process of sodium ion battery

Abstract

The invention discloses a monitoring method and system for a sodium ion battery manufacturing process, and relates to the technical field of battery manufacturing monitoring, and the system comprises a monitoring module which divides a wide area of a coating substrate into a plurality of column-direction monitoring areas according to the moving direction of the wide area, and each column corresponds to a group of independent monitoring paths; the projection module is used for projecting a latticed structured light pattern matched with the width of each column of monitoring area on each column of monitoring area; the image acquisition module is used for carrying out synchronous image acquisition on each monitoring area in the continuous moving process of the coating base material so as to obtain image data; and the image analysis module is used for processing and analyzing the image data in a spatial dimension and a time dimension. According to the invention, an image acquisition frame of column direction monitoring area + projection structured light + image enhancement analysis is constructed, the response sensitivity of the image to the fluctuation and deformation of the coating surface is improved through the latticed structured light pattern, and high-resolution and high-stability surface state acquisition is realized.

Description

Technical Field

[0001] This invention relates to the field of battery manufacturing monitoring technology, specifically to a monitoring method and system for the manufacturing process of sodium-ion batteries. Background Technology

[0002] With the development of battery manufacturing technology, coating and pressing are core front-end processes, and their quality control has a significant impact on battery performance. In the coating stage, the slurry needs to be evenly distributed on the substrate surface to ensure the consistency of the subsequent electrode structure; in the pressing stage, the slurry layer is compacted by applying precise pressure to achieve the required thickness and density. To ensure process quality, some manufacturing systems have adopted image processing technology to monitor the coated surface online, extracting information such as pattern grayscale, surface texture, or local thickness to determine if there are any abnormalities in the coating. Meanwhile, some systems also collect parameters such as surface brightness and pressure after the pressure rollers to evaluate the compaction effect. A search revealed a Chinese patent (publication number: CN117691051A) that discloses an intelligent processing system and method for lithium battery electrodes, including a processing module, a monitoring module, a verification module, and an alarm module. This invention enables real-time detection of abnormalities in processing equipment during the intelligent processing of lithium battery electrodes, providing timely feedback to management personnel to prevent further losses due to equipment malfunctions. It also allows real-time monitoring of various parameters of the processed lithium battery electrodes, analyzing whether the finished product meets standards, and identifying specific problems in the processing steps or equipment based on defective products, promptly notifying maintenance personnel for repairs and minimizing raw material losses.

[0003] In existing technologies, most process inspection methods treat the coating stage and the pressing stage as independent inspection processes, lacking a mechanism for corresponding quality data and anomaly transmission analysis between the two stages. This lack of cross-process data correlation limits the ability to determine whether coating anomalies truly affect the pressing effect, restricting the precise control and real-time adjustment capabilities of the manufacturing process. Therefore, this invention proposes a monitoring system for the sodium-ion battery manufacturing process. Summary of the Invention

[0004] The purpose of this invention is to provide a monitoring method and system for the manufacturing process of sodium-ion batteries, so as to solve the problems mentioned in the background art.

[0005] The present invention can be achieved through the following technical solution: a monitoring system for the manufacturing process of sodium-ion batteries, comprising: a monitoring module, a projection module, an image acquisition module, an image analysis module, a construction module, and a mapping module; The monitoring module is used to divide the wide area of ​​the coated substrate into multiple column-oriented monitoring areas according to its movement direction, and each column corresponds to an independent monitoring path, which facilitates parallel quality tracking and position positioning of the entire coated surface, and improves the spatial resolution and structural manageability of the monitoring. The projection module is used to project a grid-like structured light pattern matching its width onto each column of monitoring areas. The grid-like structured light pattern is a rectangular network structure, which is used to enhance the response sensitivity to changes in the undulation of the coating surface during image acquisition. The image acquisition module is used to synchronously acquire images of each monitoring area during the continuous movement of the coating substrate in order to obtain image data. The image analysis module is used to process and analyze image data in both spatial and temporal dimensions. In the spatial dimension, the deformation of the grid-like structured light pattern in different regions at the same time is analyzed to determine the undulation state of the coated surface; In the time dimension, the changing trend of gray values ​​in each grid area is analyzed to determine whether the slurry concentration is stable and whether the distribution is uniform. The construction module registers the multi-frame image data acquired by the image acquisition module according to the corresponding column, time sequence and the moving speed of the coating substrate to construct a "monitoring set" corresponding to each column of monitoring area. The monitoring set structure includes time tags, grayscale change sequence, spatial location information and image deformation features, which is the basic data source for pressure roller mapping analysis. The mapping module is used to collect the pressure value and surface brightness image of the pressure roller at the corresponding position of each monitoring area in the pressure roller stage after coating, and perform position mapping and feature correspondence analysis with the monitoring set formed in the coating stage. By comparing the coupling relationship between grayscale fluctuation characteristics and the brightness response of the pressure roller, it is possible to determine whether coating defects have been transmitted to the compaction stage. This enables cross-process quality linkage analysis and achieves a closed loop of quality data between the coating and pressure roller stages. This improves the stability of the manufacturing process and the real-time nature of defect handling. It solves the problem that in the battery manufacturing process, there is no effective quality data correlation mechanism between the coating and pressure roller processes, which leads to coating abnormalities not being identified and responded to in a timely manner by the subsequent pressure roller stage, resulting in undetected defect transmission and delayed adjustments in the manufacturing process.

[0006] A further technical improvement of the present invention is that: the image analysis module, based on the image grayscale change and structural deformation characteristics, and according to the preset judgment rules, judges whether there are abnormal surface undulations or abnormal slurry concentrations in the coating stage and generates anomaly markers, and the anomaly markers are bound to the corresponding image positions and time frames in the monitoring set.

[0007] A further technical improvement of the present invention is that the spatial dimension processing of image data by the image analysis module includes: A1. The image acquisition module acquires the entire image data in the same time frame, and the monitoring module divides the multiple column-oriented monitoring areas to obtain column-oriented sub-regions; The image analysis module extracts image segments from each column sub-region as the target image for analysis and performs preprocessing. A2. In the target image obtained in A1, extract the boundary lines and intersection nodes corresponding to the grid-like structured light pattern formed by the projection module to construct the pattern structure information; A3. Align the positions of the intersection nodes in the image with the preset standard pattern model to establish a one-to-one correspondence between the actual pattern structure and the ideal model. A4. Based on the offset of the intersection node position, the ratio of the horizontal and vertical spacing changes and the deviation of the intersection angle, calculate the deformation characteristics of the pattern structure and construct the deformation description of the column sub-region. A5. Based on the set thresholds, determine whether there are continuously distributed pattern structure anomalies in the column sub-regions. If the fluctuation anomaly judgment conditions are met, generate an anomaly label and write it into the monitoring set in the construction module, and bind it with the image time frame and the position of the coated substrate.

[0008] A further technical improvement of the present invention is that the image analysis module processes the temporal dimension of image data, including: B1. During the movement of the coating substrate, the image acquisition module acquires continuous image frames containing a grid-like structured light pattern at preset time intervals. The image analysis module extracts image segments of the same column sub-region in different time frames and performs registration to form a time series image set for tracking pattern grayscale changes; B2. After completing the image enhancement process, the image analysis module extracts the set of pixels around the intersection area of ​​the pattern lines and calculates its average gray value as the gray value observation value of the area in the current time frame. B3. The system constructs grayscale change curves based on the grayscale observations of each column sub-region and calculates the grayscale fluctuation amplitude and fluctuation frequency. B4. If the grayscale fluctuation amplitude exceeds the set upper limit, or the fluctuation frequency exceeds the threshold, the system marks the column sub-region as "uneven concentration" and generates an abnormal marker, which is bound to the corresponding image frame and physical location information and written into the monitoring set of the construction module.

[0009] A further technical improvement of the present invention is that, in step B2, in order to improve robustness, a sampling window of fixed size is set for each intersection region, and its local gray-scale average value is extracted as the gray-scale observation value under that time frame.

[0010] A further technical improvement of the present invention is that: the mapping module establishes a multi-dimensional mapping relationship table between the coating stage and the pressure roller stage based on the anomaly markers generated by the image analysis module and their corresponding column numbers, time frame numbers and physical positions on the coating substrate, including position indexes, anomaly categories, pressure values ​​and brightness parameters. By comparing the overlap between the location of the anomaly area and the pressure distribution of the pressure roller stage, the grayscale mean of the corresponding area in the brightness image after compaction and the image edge clarity, it determines whether the anomaly markers generate a response in the pressure roller stage. If any feature meets the corresponding response threshold condition, the system will mark the anomaly as "responded anomaly"; Otherwise, it is marked as "unresponsive abnormal" and a feedback control signal is generated to instruct the upstream process parameters to be adjusted.

[0011] A further technical improvement of the present invention lies in that: the image analysis module models the abnormal change process of each monitored area based on the trend of image grayscale value changes in each column sub-region in continuous time frames, divides it into multiple time periods including a steady state segment, a pre-disturbance segment, abrupt change point, and a continuous abnormal segment, and generates evolutionary labels based on the directionality, amplitude, and persistence of grayscale value changes. The evolution tag is bound to the corresponding monitoring set location, time frame, and anomaly type, and serves as the input parameter for risk classification analysis of the pressure roller stage.

[0012] A further technical improvement of the present invention is that: the image analysis module identifies whether there is a synchronous abnormal impact zone between columns based on the abnormal marking results of multiple column sub-regions within the same time window. If at least three or more adjacent or non-adjacent column sub-regions simultaneously show abnormal markings within a preset time window, the system marks the region corresponding to the time window as a "cross-column impact zone". The impact zone marker is used to indicate that the anomaly has a potential system-level disturbance cause, and serves as the input basis for subsequent risk classification judgment of the mapping module and upstream equipment status assessment.

[0013] This invention also discloses a monitoring method for the manufacturing process of sodium-ion batteries, comprising the following steps: S1. Divide the wide area of ​​the coated substrate into multiple column-oriented monitoring areas according to its movement direction. Each column-oriented monitoring area corresponds to a set of independent monitoring paths, which are used to perform parallel quality tracking and position positioning of the entire coated surface. S2. Project a grid-like structured light pattern matching its width onto the monitoring area in each column. The grid-like structured light pattern is a rectangular network structure, which is used to enhance the response sensitivity to changes in the undulation of the coating surface during image acquisition. S3. During the continuous movement of the coating substrate, synchronous image acquisition is performed on each column monitoring area to obtain an image data sequence containing the projection of the grid-like structured light pattern. S4. Perform spatial and temporal processing and analysis on the acquired image data, specifically including: In the spatial dimension, the deformation of the network structured light pattern in different regions under the same time frame is analyzed to determine the undulation state of the coated surface; In the time dimension, the gray value change trend of each grid area is analyzed to determine whether the slurry concentration is stable and whether the distribution is uniform. S5. Based on the multi-frame image data acquired by the image acquisition, the registration process is performed according to the column number, time sequence and the moving speed of the coating substrate to construct a monitoring set corresponding to each column monitoring area. The monitoring set includes time label, grayscale change sequence, spatial location information and image deformation features. S6. In the pressure roller stage after coating, collect the pressure roller pressure value and surface brightness image of each column at the corresponding position of the monitoring area, and perform position mapping and feature correspondence analysis with the monitoring set constructed in the coating stage. S7. Based on the mapping results, compare the coupling relationship between the grayscale fluctuation characteristics and the brightness response of the pressure roller to determine whether the abnormality in the coating stage is transmitted to the compaction stage, thereby realizing cross-process quality linkage analysis. S8. Based on the above analysis results, a closed loop of quality data between the coating and pressing roller processes is achieved, which improves the stability of the manufacturing process and the real-time nature of defect handling, and solves the problem of defect transmission and adjustment lag caused by the disconnection of process data.

[0014] Compared with the prior art, the present invention has the following beneficial effects: This invention constructs an image acquisition framework of "column-oriented monitoring area + projected structured light + image enhancement analysis". By using a grid-like structured light pattern, it improves the image's response sensitivity to surface undulations and deformations, achieving high-resolution and high-stability surface state acquisition. Furthermore, it proposes a "monitoring set" data structure model, which unifies the organization of image frame sequences, grayscale changes, position information, and pattern deformation features to form a continuous and complete coating monitoring dataset, providing standardized data support for subsequent pressure roller mapping and trend judgment. Meanwhile, this invention establishes an image analysis mechanism of "spatial dimension + time dimension". By extracting the structural changes of pattern nodes and the trend of grayscale fluctuations, it can accurately identify typical coating anomalies such as "uneven surface" and "uneven concentration". In addition, it designs structural identification mechanisms such as "evolution label" and "abnormal impact zone" to realize trend modeling of the development process of abnormal state and judgment of inter-column synchronization, which helps to identify systematic process disturbances. On the other hand, this invention realizes multi-dimensional mapping analysis of coating abnormalities and pressure roller response. By coupling judgment of grayscale fluctuations with pressure and brightness response, it establishes an abnormal causal chain across process stages, improves the traceability and adjustment timeliness of the manufacturing process, and also sets up a feedback control path. When the pressure roller stage does not respond to coating abnormalities, the system automatically generates a feedback signal to indicate the adjustment of the front-end coating parameters, forming a closed-loop optimization control path for the manufacturing process. In summary, this invention, through structured light pattern detection, image sequence modeling, anomaly evolution judgment, and cross-stage mapping analysis, breaks through the technical bottlenecks of "stage isolation" and "response lag" in existing processes, providing intelligent and interconnected process monitoring and control methods for battery manufacturing, and has significant engineering application value. Attached Figure Description

[0015] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0016] Figure 1 This is a schematic diagram of the system logic of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0018] Please see Figure 1 As shown, the present invention provides a monitoring system for the manufacturing process of sodium-ion batteries, including: a monitoring module, a projection module, an image acquisition module, an image analysis module, a construction module, and a mapping module; The monitoring module is used to divide the wide area of ​​the coated substrate into multiple column-oriented monitoring areas according to its movement direction, and each column corresponds to an independent monitoring path, which facilitates parallel quality tracking and position positioning of the entire coated surface, and improves the spatial resolution and structural manageability of the monitoring. The projection module is used to project a grid-like structured light pattern that matches the width of each column of monitoring area. The grid-like structured light pattern is a rectangular network structure, which is used to enhance the response sensitivity to changes in the undulation of the coating surface during image acquisition. In this embodiment, the network structured light can be implemented in practical applications through laser arrays, stripe light sources or digital light processing projection devices to ensure that the pattern has sufficient stability and resolution. The image acquisition module is used to synchronously acquire images of each monitoring area during the continuous movement of the coating substrate in order to obtain image data; In this embodiment, the image acquisition module consists of a high-speed industrial camera and a synchronous triggering device. The acquired content includes the projection image of the grid-like structured light pattern on the coated surface, and the image frame sequence is recorded by combining the timestamp and the moving distance for subsequent time series modeling. The image analysis module is used to process and analyze image data in both spatial and temporal dimensions. In the spatial dimension, the deformation of the mesh-like structured light pattern in different regions at the same time is analyzed to determine the undulation state of the coated surface, including: A1. Image cropping and preprocessing of the monitoring area: Within the same time frame, the entire image data acquired by the image acquisition module is divided into multiple column-oriented monitoring areas according to the monitoring module to obtain column-oriented sub-regions; The image analysis module extracts image segments from each column sub-region as the target image for analysis. Subsequently, the system performs image preprocessing on the target image, including grayscale normalization, edge sharpening, and background noise suppression, to improve the edge clarity and structural distinguishability of the network structured light pattern, providing basic support for subsequent structure recognition and deformation analysis. A2. Recognition of grid-like structured light patterns: In the analysis target image preprocessed in step A1, the image analysis module extracts the boundary lines and node composition of the grid-like structured light pattern formed by the projection module on the coating surface (the set of lines and nodes formed by the grid-like structured light pattern) in the image. Specifically, the process includes using an edge detection algorithm to extract rectangular grid lines formed by light projection in the image, and further extracting grid intersections, i.e., the set of node coordinates, through geometric topology analysis. The line and node data serve as the basic structural information for subsequent deformation judgment. A3. Standard pattern model alignment and structural mapping: The system has a preset geometric model of a standard grid-like structured light pattern, which represents the side length, node spacing, and angle information of each grid unit under an ideal flat coating condition; The image analysis module uses coordinate registration and pattern alignment algorithms to spatially align the positions of each intersection point of the grid-like structured light pattern acquired by the image acquisition module in the current image frame data with the preset intersection point reference positions in the standard model. The spatial coordinates of the intersection points are used to establish a one-to-one correspondence between the actual position of the structured light pattern and the ideal model, thereby providing a positional basis for subsequent line spacing change analysis and angle deviation judgment. A4. Calculation of mesh structure deformation characteristics: The image analysis module extracts the grid-like structured light pattern from the image frame and performs feature analysis on the spatial relationships between the lines to determine the minute deformations on the surface of the coated substrate. Specifically, this process includes the following three dimensions: Intersection position offset analysis: By comparing the distance difference between the actual position of the intersection point of each light pattern line in the image frame in the two-dimensional coordinate system and the preset position in the standard pattern model, the degree of disturbance of the coating surface on the projection position of the structured light pattern is reflected. Analysis of the variation in the distance between adjacent intersections: The line spacing between adjacent intersections in the horizontal and vertical directions is measured separately and compared with the corresponding spacing values ​​in the standard pattern to obtain the horizontal spacing change ratio and the vertical spacing change ratio, which are used to determine whether there is scaling or compression in the coating area. Analysis of the included angle deviation formed by the intersection: Calculate the intersection angle formed by adjacent lines at the intersection point and compare it with the orthogonal angle in the standard pattern (such as 90°) to determine whether there is angular distortion, shearing or structural instability caused by coating fluctuations. The image analysis module uses the above three types of geometric features as quantitative indicators to construct a description of the structured light pattern deformation corresponding to each column of sub-region in the current image frame, which is used for subsequent fluctuation anomaly determination. A5. Anomaly detection based on pattern features: To effectively identify abnormal regions, the system sets corresponding thresholds based on three types of features extracted in step four: intersection point position offset, adjacent spacing change ratio, and intersection angle deviation. These thresholds include: Maximum allowable intersection position offset; Allowable proportions for variations in horizontal / vertical line spacing; Deviation of the intersection angle from the tolerance range; If a column sub-region contains three or more adjacent regions, and at least one of the cross-point features exceeds the corresponding threshold, and these abnormal cross-points are continuously distributed in the image frame, then the image analysis module determines that the column sub-region has abnormal coating surface undulation. The result of the abnormal fluctuation will be written into the monitoring set of the construction module as tag information and bound to the acquisition time of the image frame and the current position of the coating substrate for subsequent mapping module to perform corresponding analysis and quality transmission judgment in the pressure roller stage. Over time, the trend of grayscale values ​​in each grid region is analyzed to determine whether the slurry concentration is stable and whether its distribution is uniform, including: B1. During the movement of the coating substrate, the image acquisition module acquires continuous image frame data at preset time intervals. Each image frame contains a grid-like structured light pattern formed on the coating surface by the projection module. The image analysis module extracts and registers corresponding image segments of the same column sub-region in different time frames based on the column sub-regions divided by the monitoring module, forming a time series image set, which is used to track the pattern grayscale information of the column region over time. B2. Region grayscale feature extraction and normalization: For each image frame segment in the above image frame sequence, the image analysis module extracts the set of pixels around the intersection area of ​​each pattern line based on pattern enhancement processing (grayscale normalization, noise suppression) and calculates the average grayscale value of the intersection area. Furthermore, to improve robustness, a fixed-size sampling window is set for each intersection region, and its local grayscale average value is extracted as the grayscale observation value in that time frame; Specifically, to enhance the robustness of the grayscale extraction process to image noise and local disturbances, the system sets a fixed-size rectangular sampling window for each intersection region in each image frame. This window expands outward from the intersection point coordinates to form a local region sub-image. During operation, the coordinates of each intersection point are set as follows: The width and height of the sampling window are respectively and (Unit: pixels, typical range: 5–15 pixels), then the image area covered by this sampling window is: , ; Extract the grayscale values ​​of all pixels within this region. And calculate the average gray level of the region as the gray level observation value of the intersection point in the current time frame; To avoid edge truncation, if the intersection point is close to the image edge and the sampling window cannot fully cover it, one of the following measures shall be taken: Perform symmetrical cropping on the window to ensure area alignment; Expand image boundaries using mirror fill method; Finally, the set of grayscale observations of all intersection points is used to construct the pattern grayscale description of each column sub-region in that time frame, providing input data for subsequent grayscale curve modeling and concentration fluctuation analysis; B3. Construction of grayscale variation curves and calculation of fluctuation characteristics: The system performs time series modeling on the continuous grayscale observations of each column sub-region to construct grayscale change curves; The image analysis module calculates two key metrics based on this curve: Gray-scale fluctuation amplitude: the maximum – minimum gray-scale difference in the statistical sequence; Gray-scale fluctuation frequency: Calculates the number of times the gray-scale change exceeds a set threshold per unit time; These indicators reflect the consistency and stability of concentration in the area during the coating process; the greater the concentration fluctuation, the more drastic the grayscale change. B4. Judgment of abnormal slurry concentration: The system determines whether there are any anomalies in the current column-oriented sub-region based on preset concentration stability judgment rules and threshold parameters: If the grayscale fluctuation exceeds the set upper limit for concentration stability; Or the grayscale fluctuation frequency exceeds the set frequency threshold; The image analysis module then marks the sub-region of the column as "uneven concentration" in the current time period, generates abnormal marker data, binds it to the image frame timestamp and coating substrate location information corresponding to the region, and writes it into the monitoring set of the construction module; The image analysis module models the abnormal change process of each monitored area based on the grayscale value change trend of each column sub-region in continuous time frames. It divides the area into multiple time periods, including a steady-state period, a pre-disturbance period, abrupt change points, and a continuous abnormal period. Based on the directionality, amplitude, and persistence of grayscale value changes, it generates evolution labels. Evolutionary tags are bound to the corresponding monitoring set location, time frame, and anomaly type, and used as input parameters for risk classification analysis of the pressure roller stage; Specifically, the system first performs time series extraction on the “monitoring set” corresponding to each column of the sub-region in the construction module. The extracted content includes: the gray value change sequence at the same spatial location in consecutive image frames, and calculates its first derivative (rate of change) and second derivative (acceleration trend). Subsequently, the system divides the time series into different stages according to the following rules: Steady-state segment: The rate of change of gray value is lower than the set steady-state fluctuation threshold, indicating that the concentration in this area is in a stable state; Early stage of disturbance: The gray value begins to fluctuate continuously but has not yet exceeded the abnormal judgment threshold, which may indicate an early sign of unstable slurry supply; Abrupt change point: When the rate of grayscale change reaches a local maximum and exceeds the abrupt change rate threshold, it indicates that the concentration or surface state in that region has undergone a drastic change. Persistent abnormal segment: After the mutation point, the gray level remains in the abnormal threshold range, indicating that there are persistent problems such as uneven concentration or abnormal distribution in this area; Based on the above stage division results, the system generates a set of structured "abnormal evolution labels" for each monitoring area. These labels must contain at least: Region number (corresponding column number and physical coordinates); Anomaly type (e.g., concentration fluctuations / disturbances); Evolutionary patterns (such as "mutation-persistence", "steady state-mutation-recovery"); The start and end time frame numbers for each stage; Key change indicators (such as maximum slope, number of frames of abnormal duration, and magnitude of change); The evolution tag structure is embedded in the monitoring center and subsequently used by the mapping module as a basis for risk trend classification in the pressure roller stage. It can be linked with the pressure fluctuation response and the change pattern of the brightness image after pressing to determine whether the anomaly is an isolated fluctuation, a periodic defect or a systemic anomaly. Through this mechanism, the system can not only statically identify anomalies, but also dynamically depict the entire process of anomaly development, thereby improving the intelligent analysis and proactive response capabilities of the quality monitoring system. The image analysis module identifies whether there is a synchronous abnormal impact zone between columns based on the abnormal marking results of multiple column sub-regions within the same time window. If at least three or more adjacent or non-adjacent column sub-regions show abnormal markings at the same time within a preset time window, the system marks the area corresponding to that time window as a "cross-column impact zone". The impact zone marker is used to indicate that the anomaly has a potential system-level disturbance cause, and serves as the input basis for subsequent risk classification judgment of the mapping module and upstream equipment status assessment; Specifically, after the system completes the anomaly identification and marking (such as "uneven surface" or "uneven concentration") of each column monitoring area in the image analysis module, it performs a global scan of the time window composed of the same time frame or multiple adjacent time frames to extract the anomaly distribution status of all column monitoring areas in each frame. The system identifies "cross-column impact bands" based on the following judgment rules: Time window setting: Uses N consecutive frames of image data as a single time analysis window; Inter-column anomaly statistics: Within this time window, count the column-direction monitoring area numbers marked as anomalies in each frame of the image; Impact threshold setting: If, within this time window, there are at least 3 columns in the monitoring area that are marked as abnormal in any frame at the same time, and the interval between these column numbers does not exceed the maximum column span threshold (e.g., 5 columns), then it is determined to be an "inter-column synchronous impact event". Impact zone marker generation: Write the time frame and its corresponding multiple column numbers, anomaly type, and impact intensity indicators (such as the number of abnormal columns and the maximum grayscale change value) into the "impact zone structure marker" and embed it into the monitoring set corresponding to the construction module; Application feedback: This impact band mark not only serves as a diagnostic basis for system-level disturbance events during the coating stage, but is also used for the analysis of abnormal clustering response trends in the mapping module during the pressure roller stage; If the same impact zone also exhibits high pressure deviation or abnormal brightness during the pressure roller stage, the system will indicate that there may be abnormal process parameter synchronization or equipment stability issues. Through this mechanism, the system expands from the traditional "single-column independent identification" mechanism to anomaly synchronous correlation reasoning based on the horizontal spatial dimension, forming a hierarchical identification logic chain of "single point → column → cross column → system level", which helps to discover common anomaly causes in the coating process and trigger control response in advance, enhancing the ability to understand and control multi-source anomalies in the manufacturing process. The construction module registers the multi-frame image data acquired by the image acquisition module according to the corresponding column, time sequence and the moving speed of the coating substrate, and constructs a "monitoring set" corresponding to each column of monitoring area. The monitoring set structure includes time label, grayscale change sequence, spatial location information and image deformation features, which is the basic data source for pressure roller mapping analysis. The mapping module is used to collect the pressure value and surface brightness image of each monitoring area at the corresponding position in the pressure roller stage after coating, and perform position mapping and feature correspondence analysis with the monitoring set formed in the coating stage. By comparing the coupling relationship between grayscale fluctuation characteristics and pressure roller brightness response, it is possible to determine whether coating defects have been transmitted to the compaction stage. This enables cross-process quality linkage analysis and achieves a closed loop of quality data between the coating and pressure roller stages. This improves the stability of the manufacturing process and the real-time performance of defect handling. It solves the problem that in battery manufacturing, there is no effective quality data correlation mechanism between the coating and pressure roller processes, which leads to coating abnormalities not being identified and responded to in a timely manner by the subsequent pressure roller stage, resulting in undetected defect transmission and delayed adjustments in the manufacturing process. Based on the anomaly markers generated by the image analysis module and their corresponding column numbers, time frame numbers, and physical locations on the coating substrate, the mapping module establishes a multi-dimensional mapping relationship table between the coating stage and the pressure roller stage, which includes location indexes, anomaly categories, pressure values, and brightness parameters. By comparing the overlap between the location of the anomaly area and the pressure distribution of the pressure roller stage, the grayscale mean of the corresponding area in the brightness image after compaction, and the image edge clarity, the mapping module determines whether the anomaly markers respond during the pressure roller stage. If any feature meets the corresponding response threshold condition, the system will mark the anomaly as "responded anomaly"; Otherwise, it is marked as "unresponsive exception" and a feedback control signal is generated to instruct the upstream process parameters to be adjusted, specifically: Specifically, the system constructs a multi-dimensional mapping table of space-time-label-response, binds the anomaly judgment results (including "uneven surface" or "uneven concentration") of each column of sub-region in each time frame to its actual physical coordinates on the coating substrate, and synchronously writes the response data such as the pressure value of the pressure roller stage and the brightness image information after pressing corresponding to the physical coordinates into the mapping table. During the mapping stage, the system uses a mapping table to determine the spatial overlap between the abnormal area and the pressure roller response area. It also combines parameters such as pressure distribution characteristics, grayscale mean and edge blur of the post-press brightness image to evaluate whether the abnormality exhibits significant response characteristics during the pressure roller stage. If the pressure or brightness change is statistically significant in the abnormal area, the system will upgrade the abnormality to "responded abnormality"; if no significant response change is observed, the system will record it as "unresponded abnormality" and feed it back to the upstream process control module to prompt for adjustment of coating pressure, slurry concentration or equipment parameters. The above response verification results are written into the monitoring set structure, forming a one-to-one correspondence with the corresponding time frame, column number and position coordinate, serving as the key input for subsequent abnormal trend judgment and closed-loop process control; Based on the extraction of image features, the image analysis module further judges whether there are any abnormalities in each monitoring area during the coating stage according to preset judgment rules and threshold conditions. Specifically, if the deformation amplitude of the projected pattern in a certain area exceeds the set fluctuation threshold at the same time point, the system marks the monitored area as "uneven surface". If the grayscale value fluctuation of a certain area exceeds the set concentration stability threshold within a continuous time period, the system will mark the monitored area as "uneven concentration". When a monitoring area is marked as "uneven surface" or "uneven surface", the corresponding monitoring area will be marked as abnormal.

[0019] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0020] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A monitoring system for a sodium-ion battery manufacturing process, characterized in that, The method comprises the following steps: A monitoring module divides the wide area of the coated substrate into multiple columnar monitoring areas according to the moving direction of the coated substrate, and each column corresponds to a group of independent monitoring paths; A projection module projects a grid-like structured light pattern matching the width of each columnar monitoring area on the monitoring area; An image acquisition module synchronously acquires images of each monitoring area to obtain image data during the continuous movement of the coated substrate; An image analysis module processes and analyzes the image data in spatial and temporal dimensions; In the spatial dimension, the deformation of the grid-like structured light pattern in different areas at the same time is analyzed to determine the undulating state of the coating surface; In the temporal dimension, the change trend of the gray value of each grid area is analyzed to determine whether the slurry concentration is stable and the distribution is uniform; A construction module aligns and processes multiple frames of image data obtained by the image acquisition module according to the corresponding column, time sequence and moving speed of the coated substrate to construct a "monitoring set" corresponding to each columnar monitoring area; A mapping module acquires the roller pressure value and surface brightness image of the position corresponding to each columnar monitoring area during the roller stage after coating, and performs position mapping and feature corresponding analysis with the monitoring set formed during the coating stage.

2. The monitoring system for sodium-ion battery manufacturing process according to claim 1, wherein, The image analysis module determines whether there is surface undulation abnormality or slurry concentration abnormality during the coating stage based on the image gray value change and structure deformation characteristics according to the preset judgment rule, and generates an abnormality mark, and the abnormality mark is bound with the corresponding image position and time frame in the monitoring set.

3. The monitoring system for sodium-ion battery manufacturing process according to claim 2, wherein, The spatial dimension processing of the image analysis module on the image data comprises: A1, the image acquisition module acquires the whole image data at the same time frame, divides the multiple columnar monitoring areas divided by the monitoring module, and obtains columnar sub-regions; The image analysis module extracts image segments in each columnar sub-region as analysis target images and performs preprocessing; A2, in the analysis target image obtained in A1, the boundary lines and intersection nodes of the grid-like structured light pattern formed by the projection module are extracted to construct pattern structure information; A3, the positions of the intersection nodes in the image are aligned with the preset standard pattern model to establish a one-to-one correspondence between the actual pattern structure and the ideal model; A4, based on the intersection node position offset, the horizontal and vertical distance change ratio and the intersection angle deviation, the pattern structure deformation characteristics are calculated to construct the deformation description of the columnar sub-region; A5, according to the set threshold values, it is judged whether there is continuously distributed pattern structure abnormality in the columnar sub-region, if the undulation abnormality judgment condition is met, an abnormality label is generated and written into the monitoring set in the construction module, which is bound with the image time frame and the position of the coated substrate.

4. The monitoring system for sodium-ion battery manufacturing process of claim 3, wherein, The temporal dimension processing of the image analysis module on the image data comprises: B1, during the movement of the coated substrate, the image acquisition module acquires continuous image frames containing the grid-like structured light pattern at a preset time interval; The image analysis module extracts image segments of the same columnar sub-region in different time frames and performs alignment to form a time sequence image set for tracking the gray value change of the pattern; B2, after the image enhancement processing is completed, the image analysis module extracts the pixel set around the intersection area of the pattern lines and calculates the average gray value as the gray observation value of the area at the current time frame; B3, the system constructs a gray change curve based on the gray observation values of the columnar sub-regions and calculates the gray fluctuation amplitude and frequency; B4, if the gray fluctuation amplitude exceeds the set upper limit or the fluctuation frequency exceeds the threshold value, the system marks the columnar sub-region as "concentration unevenness" and generates an abnormality mark, which is bound to the corresponding image frame and physical location information and written into the monitoring set of the construction module.

5. The monitoring system for sodium-ion battery manufacturing process of claim 1, wherein, The mapping module establishes a multi-dimensional mapping relationship table containing position index, abnormality category, pressure value and brightness parameter between the coating stage and the pressure roller stage based on the abnormality mark generated by the image analysis module and its corresponding columnar number, time frame number and physical position on the coated substrate, and judges whether the abnormality mark produces a response in the pressure roller stage by comparing the abnormal region position and the pressure intensity distribution coincidence degree in the pressure roller stage, the gray mean value of the corresponding area in the brightness image after compaction and the image edge definition; If any feature meets the corresponding response threshold condition, the system determines that the abnormality mark is an "responded abnormality"; Otherwise, it is marked as "non-responded abnormality" and a feedback control signal is generated to indicate that the process parameters in the previous stage are adjusted.

6. The monitoring system for sodium-ion battery manufacturing process of claim 1, wherein, The image analysis module models the abnormal change process of each monitoring region based on the image gray value change trend of each columnar sub-region in consecutive time frames, divides it into multiple time periods including steady state section, pre-disturbance section, mutation point and continuous abnormality section, and generates an evolution label according to the directionality, amplitude and persistence of the gray value change; The evolution label is bound to the corresponding monitoring set position, time frame and abnormality type and used as an input parameter for risk classification analysis in the pressure roller stage.

7. The monitoring system for sodium-ion battery manufacturing process of claim 1, wherein, The image analysis module identifies whether there is an inter-column synchronous abnormal impact zone based on the abnormality mark results of multiple columnar sub-regions within the same time window. If at least three and more adjacent or non-adjacent columnar sub-regions simultaneously appear abnormality marks within a preset time window, the system marks the region corresponding to the time window as "cross-column impact zone"; The impact zone mark is used to indicate that the abnormality has a potential system-level disturbance cause and is used as an input basis for subsequent mapping module risk classification judgment and upstream equipment state evaluation.

8. A monitoring method for a sodium-ion battery manufacturing process, characterized in that, The method uses the monitoring system of any one of claims 1-7, comprising: S1, dividing the wide area of the coated substrate into multiple columnar monitoring regions according to its moving direction, each columnar monitoring region corresponding to a group of independent monitoring paths; S2, projecting a rectangular grid structure light pattern matching the width of each columnar monitoring region on it; S3, synchronously collecting images of each columnar monitoring region during the continuous movement of the coated substrate to obtain a sequence of image data containing the projected structure light pattern; S4, processing and analyzing the image data in spatial and temporal dimensions, wherein: In the spatial dimension, the deformation of the structure light pattern in different regions at the same time frame is analyzed; In the time dimension, the change trend of the gray value of each grid region is analyzed; S5, according to column direction number, time sequence and coating substrate moving speed, the image data is registered to construct a monitoring set of each column direction monitoring area, the monitoring set includes time label, gray change sequence, spatial position information and image deformation characteristics; S6, in the pressure roller stage, the pressure roller pressure value and the post-pressing brightness image of each column direction monitoring area position are collected, and position mapping and characteristic corresponding analysis are performed on the monitoring set; S7, according to the mapping result, the coupling relationship between the gray fluctuation characteristics and the pressure roller brightness response is compared, whether the coating abnormality is transmitted to the compaction stage is judged; S8, the analysis result is used for data linkage between the coating and the pressure roller process, and a monitoring closed loop is realized.

Images