Cloud edge collaboration based industrial process visualized early warning method

Abstract

The application discloses an industrial process visualized early warning method based on cloud edge cooperation, relates to the technical field of process visualization early warning, and takes the lithium battery pole piece coating and drying production line as an application scene, associates production formula data, equipment action data, material position data, process response data and quality inspection result data according to material sections, judges possible defect types of specific material sections at future workstations in advance through cloud end defect shadow template and edge end real-time matching, and displays risk sources, propagation paths, expected appearance workstations and residual intervention distances in the form of risk development belts. The method can complete visualized early warning and intervention verification before defects are formed, and reduces the risk of missed inspection and batch scrapping.

Description

Technical Field

[0001] This invention relates to the field of process visualization and early warning technology, and in particular to an industrial process visualization and early warning method based on cloud-edge collaboration. Background Technology

[0002] In continuous industrial processes such as coating and drying of lithium battery electrodes, the electrodes, in roll form, sequentially pass through the coating station, temperature zones 1 to 6, the cooling station, the correction station, and the winding station. In actual production, process defects often exhibit a significant delayed manifestation. For example, a drop in localized air pressure at the nozzle in temperature zone 3, fluctuations in exhaust volume, or abnormal tension compensation may not immediately result in visible defects at that station. Instead, these defects may manifest as insufficient edge drying, edge powdering, dark lines, or edge warping only after the material continues to run to temperature zone 5, the cooling station, or the winding process.

[0003] Current industrial early warning methods typically rely on equipment parameter thresholds, single-point sensor alarms, or backend machine vision inspection. Equipment parameter threshold alarms can only determine whether the current temperature, air pressure, tension, or linear speed exceeds preset ranges, making it difficult to identify the delayed impact of a minor process action on the quality of subsequent material segments. Machine vision inspection can usually only detect existing surface defects and cannot provide a window for intervention before the defects become apparent. While cloud-based quality prediction can utilize historical data for analysis, it is difficult to meet the millisecond-level early warning and on-site intervention requirements of continuous production lines without real-time binding of material segment location, process actions, and downstream responses at the edge.

[0004] Especially in the lithium battery electrode coating and drying scenario, whether an abnormal operation of the same equipment will cause defects depends on the product model, slurry batch, target coating thickness, oven temperature zone setting, target linear velocity, target tension, and the response changes of the material segment after passing through downstream stations. Existing technologies typically monitor or analyze equipment status, process response, and quality inspection results as independent data, making it difficult to establish a continuous correlation between upstream process actions, material segment propagation paths, downstream response changes, and future defect manifestation locations. Therefore, an industrial process visualization and early warning method is needed that can generate historical defect shadow templates in the cloud and identify in real time whether the current material segment carries the risk of future defects. Summary of the Invention

[0005] The technical problem solved by this invention is that existing technologies are unable to establish real-time correlation and visual early warning between upstream weak process anomalies and downstream material segment responses, future defect manifestation stations, and remaining intervention distances in continuous roll material industrial processes.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: The cloud-edge collaborative industrial process visualization and early warning method includes the following steps: Step S1: Collect production formula data, equipment action data and material location data at the edge end to generate material segment process phase segments; Step S2: Collect process response data at the edge and generate response afterimage data; Step S3: The cloud platform establishes a defect shadow template based on historical material segment process phase fragments, historical response afterimage data, and quality inspection result data; Step S4: At the edge end, development warning data is generated based on the process phase segment of the material segment, the response afterimage data, and the defect shadow template. Step S5: Generate a future workstation visualization warning layer based on the development warning data; Step S6: Perform intervention based on the development warning data and generate intervention rollback data.

[0007] Preferably, step S1 includes the following sub-steps: Step S101: Collect production formula data at the edge end. The production formula data includes product model, slurry batch, target coating thickness, oven temperature zone setting, target linear velocity, and target tension. Step S102: Collect equipment action data at the edge end. The equipment action data includes coating valve opening, die head pressure, nozzle air pressure, exhaust volume, tension compensation amount, and correction execution amount. Step S103: Collect material position data at the edge end. The material position data includes the roll encoder pulse, material segment number, current position of the material segment, and expected arrival station of the material segment. Step S104: When the equipment action data changes beyond the action trigger threshold, the production formula data, equipment action data, and material location data are bound together to generate a material segment process phase segment.

[0008] Preferably, step S2 includes the following sub-steps: Step S201: The edge end tracks the process of the material segment passing through the downstream station according to the material segment number; Step S202: Collect process response data, which includes edge temperature difference, surface reflection grayscale, solvent concentration change value, tension micro-amplitude, wet film thickness fluctuation value, and edge warping height. Step S203: Write the process response data into the process phase segment of the material segment in the order of material segment number, trigger station and downstream station to generate response afterimage data.

[0009] Preferably, step S3 includes the following sub-steps: Step S301: Receive historical material segment process phase fragments, historical response afterimage data, and quality inspection result data from the cloud. Step S302: Group the products according to product model, slurry batch, target coating thickness, and oven temperature zone setting. Step S303: Within the same group, find the material segment whose quality inspection result data is abnormal, and extract the corresponding historical material segment process phase fragment and historical response afterimage data; Step S304: Combine the trigger station, the direction of change of equipment action data, the sequence of downstream stations, the direction of change of process response data, and the type of quality inspection defect into a defect shadow template.

[0010] Preferably, step S4 includes the following sub-steps: Step S401: Obtain the current material segment process phase fragment and the current response afterimage data at the edge end; Step S402: Locate the defect shadow template corresponding to the current production formula data; Step S403: Perform a first match between the trigger station, the direction of change of equipment action data, and the expected arrival station of the material segment in the current material segment process phase segment and the defect shadow template; Step S404: Perform a second match between the process response data change direction and downstream station sequence in the current response afterimage data and the defect shadow template; Step S405: When both the first match and the second match meet the preset matching conditions, development warning data is generated. The development warning data includes the risk material segment number, the risk source station, the expected development station, the expected development distance, the defect type, and the risk level.

[0011] Preferably, step S5 includes the following sub-steps: Step S501: Read the risk material segment number and current location of the material segment based on the development warning data; Step S502: Generate a risk development zone in the production line station diagram based on the current position of the material segment, the expected development station, and the expected development distance. Step S503: Mark the risk source station, expected development station, defect type, risk level, and remaining intervention distance on the risk development zone; Step S504: Output the risk imaging tape to the control terminal display interface.

[0012] Preferably, step S6 includes the following sub-steps: Step S601: The edge end selects an intervention action based on the development warning data. The intervention action includes adjusting the nozzle pressure, adjusting the exhaust volume, adjusting the tension compensation amount, adjusting the correction execution amount, and adjusting the target linear velocity. Step S602: After performing the intervention action, continue to collect process response data; Step S603: Compare the process response data after intervention with the response afterimage data before intervention to generate intervention rollback data; In step S604, if several pre-retrograde data indicate a decrease in risk level, the risk imaging band will be shortened or canceled; if the risk level does not decrease, the risk material segment number will be sent to the downstream quality inspection end.

[0013] Preferably, the logic for establishing the defect shadow template is as follows: The cloud uses the defect type in the quality inspection result data as the target result, traces back the historical response afterimage data of the same material segment, and then traces back the corresponding historical material segment process phase segment; If multiple abnormal material segments have the same trigger station, the same direction of change in equipment action data, and the same direction of change in process response data, they are recorded as the same defect shadow template.

[0014] Preferably, the logic for generating the risk level is as follows: At the edge, the number of triggering consistent items between the current material segment process phase fragment and the defect shadow template, and the number of response consistent items between the current response afterimage data and the defect shadow template are counted respectively. When both the number of triggering consensus items and the number of responding consensus items reach the high-risk threshold, the risk level is high-risk. If only one of the risk factors reaches the high-risk threshold, the risk level is medium risk; if none of them reach the threshold, the risk level is low risk.

[0015] Preferably, the display logic of the future workstation visualization early warning layer is as follows: The control terminal uses the material movement direction as the display axis, displays the risk source station at the current station position, displays the expected manifestation station at the downstream station position, and covers the risk manifestation band at the material segment position corresponding to the risk material segment number, so that operators can obtain a visual early warning result consisting of the source of the anomaly, the propagation path, the expected manifestation position, and the intervention result.

[0016] The beneficial effects of this invention are as follows: Taking the lithium battery electrode coating and drying production line as an application scenario, this invention establishes a correlation between production formula data, equipment operation data, material location data, process response data, and quality inspection result data according to material segments. Through cloud-based defect shadow templates and real-time matching at the edge, it can predict in advance the types of defects that may appear in specific material segments at future workstations. The risk development belt displays the source of risk, propagation path, expected display workstation, and remaining intervention distance. This method can complete visual early warning and intervention verification before defects are formed, reducing the risk of missed detections and batch scrapping. Attached Figure Description

[0017] Figure 1A flowchart illustrating the steps of an industrial process visualization and early warning method based on cloud-edge collaboration, as provided in an embodiment of the present invention. Detailed Implementation

[0018] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] Example, refer to Figure 1 Using a lithium-ion battery electrode coating and drying production line as a specific application scenario, this paper provides an industrial process visualization and early warning method based on cloud-edge collaboration. The production line includes a coating station, temperature zones 1 to 6, a cooling station, a deviation correction station, and a winding station. The electrodes are transported in continuous rolls, and anomalies typically do not form immediately at the anomaly source station but rather manifest in downstream temperature zones or after winding. Therefore, the core of this embodiment is: The edge terminal binds the current process action to the material segment it is passing through, establishes a defect shadow template in the cloud, predicts the future display station of the defect before it is formed, and generates a risk imaging band in the production line visualization interface.

[0020] Step S1: Collect production formula data, equipment action data and material location data at the edge end to generate material segment process phase segments.

[0021] Step S2: Collect process response data at the edge and generate response afterimage data.

[0022] Step S3: The cloud platform establishes a defect shadow template based on historical material segment process phase fragments, historical response afterimage data, and quality inspection result data.

[0023] Step S4: At the edge end, development warning data is generated based on the process phase segment of the material segment, the response afterimage data, and the defect shadow template.

[0024] Step S5: Generate a future workstation visualization warning layer based on the development warning data.

[0025] Step S6: Perform intervention based on the development warning data and generate intervention rollback data.

[0026] Step S1 includes the following sub-steps: In step S101, the edge end collects production formula data at the start of production for each batch of electrodes. The production formula data includes product model, slurry batch, target coating thickness, oven temperature zone settings, target linear velocity, and target tension. For example, a batch might have product model P621, slurry batch J20260528, a target coating thickness of 86 μm, oven temperature zone settings for zones 1 to 6 of 78℃, 86℃, 94℃, 102℃, 96℃, and 88℃ respectively, a target linear velocity of 42 m / min, and a target tension of 28 N. This production formula data is not merely for display purposes but is used to subsequently select defect shadow templates, because the defect risk associated with the same nozzle pressure drop varies depending on the slurry batch, coating thickness, and temperature zone settings.

[0027] Step S102: Acquire equipment motion data at the edge. This data includes coating valve opening, die pressure, nozzle pressure, exhaust volume, tension compensation, and correction execution. Each piece of equipment motion data is recorded according to the acquisition time and workstation number. For example, in zone 3, the pressure at the second nozzle on the right side decreases from 1.82 kPa to 1.54 kPa within 8 seconds; the tension compensation changes from 0.4 N to 1.2 N as the same material passes through zone 4; and the correction execution changes from 0.6 mm to 1.7 mm at the cooling station inlet. This equipment motion data is used to determine whether a process phase trigger has occurred, rather than simply serving as general equipment status data.

[0028] Step S103: Acquire material position data at the edge. The material position data includes the roll encoder pulse, material segment number, current position of the material segment, and the expected arrival station of the material segment. Divide the continuous electrode sheets into fixed material segment lengths, for example, generating one material segment number every 2 meters. Let the material segment number be... The cumulative pulses of the roll encoder are The electrode running length corresponding to a single pulse is The current position of the material segment is: ; in, Material segment At the time of collection The current position of the material segment. Material segment The corresponding roll encoder accumulated pulses, This indicates the electrode travel length corresponding to a single pulse. It is based on the fixed coordinates of each workstation on the production line. Material section Arrive at workstation The estimated remaining distance is: ; in, Material segment Arrive at workstation The estimated remaining distance, Indicates workstation Fixed coordinates within the production line. When A value greater than 0 indicates that the material segment has not yet reached the workstation. When the value is not greater than 0, it indicates that the material segment has already passed through this station.

[0029] Step S104: The edge end generates a material segment process phase segment based on whether the equipment action data exceeds the action trigger threshold. Specifically, the action change value is first calculated for each type of equipment action data. For example, the action change value of the nozzle air pressure is: ; in, Material segment At the time of collection Corresponding to the The change value of the action data of the device type, This indicates the current collected value of the action data for this type of device. This represents the average value of the action data for this type of equipment in the normal historical data segment. This represents the standard deviation of the action data for this type of equipment within a normal historical data segment. When Exceeding the corresponding action trigger threshold And the duration reaches the continuous triggering threshold. At that time, the current production formula data, current equipment action data, and current material location data are bound together as a material segment process phase fragment. The material segment process phase fragment includes the material segment number, product model, slurry batch, target coating thickness, oven temperature zone setting, target linear velocity, target tension, trigger station, trigger action type, direction of equipment action data change, action change value, action duration, current position of the material segment, and expected arrival station of the material segment.

[0030] Step S2 includes the following sub-steps: Step S201: The edge end tracks the process of the material segment passing through the downstream station based on the material segment number. For example, material segment... When the air pressure at the right nozzle in zone 3 decreases, a process phase segment of the material segment is generated. The edge end then continues to track the process of this material segment passing through zone 4, zone 5, zone 6, the cooling station, the correction station, and the pre-winding inspection station. This tracking process relies on material position data, not on the time alignment between multiple sensor data, thus enabling direct determination of which electrode segment is affected by a particular process action.

[0031] Step S202: Acquire process response data at the edge. This data includes edge temperature difference, surface reflection grayscale, solvent concentration variation, tension micro-amplitude, wet film thickness fluctuation, and edge warpage height. Specifically, the edge temperature difference is acquired by a thermal imaging sensor to characterize uneven drying at the left and right edges of the electrode; the surface reflection grayscale is acquired by a line scan camera to characterize changes in wet film surface gloss; the solvent concentration variation is acquired by an exhaust sensor to characterize the evaporation process; the tension micro-amplitude is acquired by a tension roller sensor to characterize electrode micro-vibration; the wet film thickness fluctuation is acquired by an online thickness gauge to characterize coating stability; and the edge warpage height is acquired by a laser displacement sensor to characterize edge deformation.

[0032] In step S203, the edge device writes the process response data into the process phase segment of the material segment according to the material segment number, trigger station, and downstream station, generating response afterimage data. To enable unified comparison of data with different dimensions, the edge device calculates the response deviation value for each type of process response data: ; in, Material segment At workstation Corresponding to the Response deviation value of process response data, This represents the currently collected process response data. This indicates that the normal historical material section is at the workstation. Corresponding to the The mean of the process response data, This indicates that the normal historical material section is at the workstation. Corresponding to the The standard deviation of process response data. Response afterimage data includes material segment number, trigger station, downstream station sequence, edge temperature difference response deviation, surface reflection grayscale response deviation, solvent concentration change response deviation, tension micro-amplitude response deviation, wet film thickness fluctuation response deviation, and edge warp height response deviation. Therefore, response afterimage data represents the continuous response of a material segment propagating along the production line after being affected by the actions of upstream equipment, rather than a single moment's state value.

[0033] Step S3 includes the following sub-steps: In step S301, the cloud receives historical material segment process phase fragments, historical response afterimage data, and quality inspection result data. The quality inspection result data is obtained through post-winding sampling, visual inspection, and post-rolling re-inspection, and includes the material segment number, quality inspection defect type, defect location, and defect severity. Quality inspection defect types may include insufficient edge drying, edge chalking, dark lines, abnormal coating thickness, and edge warping. Since the quality inspection result data also includes the material segment number, the cloud can trace back to the corresponding historical material segment process phase fragments and historical response afterimage data during the production process for the same material segment.

[0034] In step S302, the cloud platform groups products according to product model, slurry batch, target coating thickness, and oven temperature zone settings. This grouping aims to avoid lumping anomalies from different process formulations together. For example, a drop in air pressure at the nozzle in zone 3 might manifest as insufficient edge drying for thick-coated products, while for thin-coated products it might only show a slight decrease in surface reflectivity. The cloud platform creates a defect shadow template within each group to ensure consistency between the template and the specific production scenario.

[0035] The logic for creating the defect shadow template is as follows: The cloud-based system uses the defect type in the quality inspection results data as the target result, traces back the historical response afterimage data of the same material segment, and then traces back the corresponding historical material segment process phase fragment.

[0036] If multiple abnormal material segments have the same trigger station, the same direction of change in equipment action data, and the same direction of change in process response data, they are recorded as the same defect shadow template.

[0037] Step S303: The cloud searches for material segments with abnormal quality inspection results within the same group, and extracts the corresponding historical material segment process phase fragments and historical response afterimage data. Let the set of abnormal material segments within a certain group be denoted as . ,in This indicates the type of quality inspection defect. The cloud-based system extracts motion feature vectors and response feature vectors for each abnormal material segment. The motion feature vector consists of the direction and value of changes in equipment motion data, while the response feature vector consists of the downstream station sequence and the response deviation value of each station. To measure the ability of a particular feature to distinguish defects, the feature discrimination index is calculated: ; in, Indicates the first Feature discriminant of each feature This represents the median of this feature within the anomalous material segment. This represents the median of this feature in a normal historical data segment. This represents the standard deviation of this feature within a normal historical data segment. Template weights are then generated based on the feature's discriminative power. ; in, Indicates the first Template weights for each feature, Indicates the number of features involved in template creation. Indicates the first The feature discrimination of each feature. In this way, data such as surface reflection grayscale, edge temperature difference, nozzle wind pressure, and tension micro-amplitude are all included in the template weight calculation, and features with higher discrimination ability have a higher proportion in subsequent matching.

[0038] In step S304, the cloud platform combines the trigger station, the direction of equipment action data change, the sequence of downstream stations, the direction of process response data change, and the quality inspection defect type into a defect shadow template. The defect shadow template includes template number, product model, slurry batch, target coating thickness range, oven temperature zone setting range, trigger station, direction of equipment action data change, action trigger threshold, downstream station sequence, direction of process response data change, response trigger threshold, template weight, and quality inspection defect type. For example, defect shadow template T3 records: Product model P621, target coating thickness 84μm to 88μm, air pressure decreases at the right nozzle of temperature zone 3, followed by an increase in edge temperature difference in temperature zone 4, a decrease in surface reflective grayscale in temperature zone 5, an increase in tension micro-amplitude at the cooling station, and the final quality inspection defect type is edge chalking.

[0039] Step S4 includes the following sub-steps: Step S401: The edge acquires the current material segment's process phase fragment and the current response afterimage data. The current material segment's process phase fragment comes from step S1, and the current response afterimage data comes from step S2; both are associated through the material segment number. For example, the material segment... If a decrease in nozzle air pressure is recorded in the third temperature zone, an increase in edge temperature difference is formed in the fourth temperature zone, and a decrease in surface reflective grayscale is formed in the fifth temperature zone, then the material segment has a complete data chain for development matching.

[0040] Step S402: The edge processor searches for a defect shadow template corresponding to the current production formula data. The edge processor filters the current product model, slurry batch, target coating thickness, and oven temperature zone setting against the defect shadow templates sent from the cloud, retaining only templates with the same product model, the same slurry batch range, a target coating thickness range covering the current target coating thickness, and an oven temperature zone setting range covering the current oven temperature zone setting. This step avoids incorrectly calling templates for other products or other process formulas.

[0041] Step S403: The edge end performs a first match on the current material segment's process phase segment. The first match is used to determine whether the anomaly source is consistent with the defect shadow template. Specifically, the trigger matching value is calculated: ; in, Material segment The trigger matching value, This indicates the number of device action data points involved in the first match. Indicates the first Template weights corresponding to the action data of each device. Indicates the current material segment corresponding to the first The action change value of each device's action data. Indicates the first The action trigger threshold for individual device action data. Indicates the direction of change in the current device action data. This indicates the direction of change in the equipment motion data recorded in the defect shadow template. This indicates that the value is 1 if the condition is met and 0 if the condition is not met.

[0042] Step S404: The edge performs a second match on the current response afterimage data. This second match is used to determine whether the downstream response propagation process is consistent with the defect shadow template. Specifically, the response matching value is calculated: ; in, Material segment The response matching value, This indicates the number of process response data points participating in the second matching. Indicates the first Template weights corresponding to each process response data point Indicates the current material segment corresponding to the first The response deviation value of each process response data. Indicates the first The response trigger threshold for each process response data Indicates the direction of change in the current process response data. This indicates the direction of change in the process response data recorded in the defect shadow template.

[0043] Step S405: When both the first and second matches meet the preset matching conditions, the edge terminal generates development warning data. The edge terminal first calculates the comprehensive development risk value: ; in, Material segment The overall development risk value, This indicates the trigger value. Indicates the response matching value. This indicates the formula matching value between the production formula data and the defect shadow template. , and These represent the weights of the trigger matching value, response matching value, and recipe matching value, respectively. In the formula... This indicates a weighted composition operation, specifically accumulating the product of the corresponding weight and the corresponding matching value. If If the low-risk threshold is reached, a warning data for imaging is generated; if If the risk level reaches the medium-risk threshold, the risk level is medium-risk; if If the high-risk threshold is reached, the risk level is high. Development warning data includes the risk material segment number, the risk source station, the expected development station, the expected development distance, the defect type, the risk level, and the remaining intervention distance.

[0044] Step S5 includes the following sub-steps: Step S501: The edge end reads the risk material segment number and the current position of the material segment based on the development warning data. For example, the risk material segment number recorded in the development warning data is... The risk originates at the right-side air nozzle in zone 3, and the expected development station is the exit of zone 5, with an expected development distance of 18m. The defect type is edge powdering, and the risk level is high. The edge end continues to update the current position of the material segment based on the material location data, ensuring the risk development zone moves synchronously with the material segment.

[0045] Step S502: The edge end generates a risk imaging band in the production line station diagram based on the current position of the material segment, the expected imaging station, and the expected imaging distance. The risk imaging band is not an equipment alarm light, but rather a visible area covering the region between the current position of the material segment and the expected imaging station. If the current position of the material segment is... The expected display workstation coordinates are The length of the risk imaging band is: ; in, Indicates the risk imaging band at the time of collection. The display length, This indicates the expected fixed coordinates of the display station within the production line. As the electrode moves, getting closer The length of the risk imaging zone is shortened accordingly, so that operators can visually see that the distance that can be intervened is decreasing.

[0046] Step S503: The control unit marks the risk source station, expected development station, defect type, risk level, and remaining intervention distance on the risk development zone. The remaining intervention distance is calculated directly from the material location data, using the following formula: ; in, Material segment The remaining intervention distance to reach the latest intervention station. This indicates the coordinates of the workstation where intervention is scheduled to occur latest. If the remaining intervention distance is less than the distance warning threshold, the control unit will bold the boundary of the risk warning zone and prompt the operator to take immediate action.

[0047] Step S504: The control unit outputs the risk development zone to the display interface. The display interface includes at least a production line workstation diagram, a material section operation diagram, and a development warning data list. The production line workstation diagram displays the workstations where the risk originates and the workstations expected to develop; the material section operation diagram displays the risk material section number; and the development warning data list displays the defect type, risk level, and remaining intervention distance. Through this layer, the operator can directly see that "the anomaly source is in the 3rd temperature zone, and the risk material section is..." "It is expected that edge powdering risk will appear at the exit of the 5th temperature zone, with a remaining intervention distance of 12m."

[0048] Step S6 includes the following sub-steps: Step S601: The edge end selects intervention actions based on the development warning data. Intervention actions include adjusting the nozzle air pressure, adjusting the exhaust volume, adjusting the tension compensation amount, adjusting the correction execution amount, and adjusting the target linear velocity. The edge end selects the priority intervention action based on the risk source station and defect type in the defect shadow template. For example, when the development warning data indicates that the decrease in the air pressure of the right nozzle in the 3rd temperature zone leads to the risk of edge chalking, the priority intervention action is to adjust the nozzle air pressure; when the development warning data indicates that the increase in the tension micro-amplitude leads to the risk of edge warping, the priority action is to adjust the tension compensation amount.

[0049] Step S602: After executing the intervention, continue collecting process response data. The intervention is not immediately considered successful at the edge; instead, the data collection continues at the downstream station for the affected material segment, including edge temperature difference, surface reflection grayscale, solvent concentration change, tension micro-amplitude, wet film thickness fluctuation, and edge warpage height. The response deviation value is then recalculated. Therefore, the intervention effect is verified by actual process response data, rather than by whether the actuator completes its action.

[0050] Step S603: The edge device compares the process response data after intervention with the response afterimage data before intervention to generate intervention rollback data. Specifically, the intervention rollback value is calculated: ; in, Material segment Intervention rollback value, This indicates the overall imaging risk value before intervention. This indicates the overall imaging risk value after intervention. If... Reaching the rollback threshold indicates that the risk has been mitigated; if If the regression threshold is not reached, it indicates insufficient intervention.

[0051] In step S604, if several pre-retrograde data indicate a decrease in risk level, the control end shortens or cancels the risk imaging band; if the risk level does not decrease, the risk material segment number is sent to the downstream quality inspection end. For example, if the material segment... The overall development risk value before intervention is 0.82, and the overall development risk value after intervention is 0.46. The risk level has been reduced from high risk to medium risk, so the risk development zone is shortened. If the overall development risk value after several prognoses is still 0.78, then the material segment is marked as a mandatory re-inspection segment, and the material segment number, expected development station and defect type are sent to the downstream quality inspection end.

[0052] The display logic for the future workstation visualization early warning layer is as follows: The control terminal uses the material movement direction as the display axis, displays the risk source station at the current station position, displays the expected manifestation station at the downstream station position, and covers the risk manifestation band at the material segment position corresponding to the risk material segment number, so that operators can obtain a visual early warning result consisting of the source of the anomaly, the propagation path, the expected manifestation position, and the intervention result.

[0053] This invention collects production formula data, equipment action data, and material location data at the edge. When the equipment action data exceeds the action trigger threshold, it binds the process action to the material segment passing through the corresponding station, forming a process phase segment for the material segment. This shifts the warning focus from whether the equipment is abnormal to whether a material segment already carries the risk of future defects. This method avoids false alarms or missed alarms caused by alarms based solely on single-point equipment parameter exceeding limits, enabling the system to accurately track continuously operating electrode material segments.

[0054] This invention continuously collects data on the edge temperature difference, surface reflection grayscale, solvent concentration change, tension micro-amplitude, wet film thickness fluctuation, and edge warping height of the same material segment at downstream stations, generating response afterimage data. This allows the impact of upstream process actions on subsequent process states to be continuously recorded on a material segment basis. As a result, the system can identify delayed response chains such as the increase in edge temperature difference in the fourth temperature zone, the decrease in surface reflection grayscale in the fifth temperature zone, and the increase in tension micro-amplitude at the cooling station after the air pressure at the nozzle in the third temperature zone drops, rather than issuing isolated alarms only when a single sensor value exceeds its limit.

[0055] This invention receives historical material segment process phase fragments, historical response afterimage data, and quality inspection result data from the cloud. It then establishes defect shadow templates based on product model, slurry batch, target coating thickness, and oven temperature zone settings. This allows defects already confirmed in historical quality inspections, such as edge chalking, insufficient edge drying, dark lines, or edge warping, to be back-linked to upstream action characteristics and downstream response characteristics prior to their occurrence. This approach transforms the cloud model from an abstract quality prediction model into a reusable early warning rule system with clearly defined trigger stations, response stations, defect types, and template weights, improving the interpretability and reproducibility of the model building process.

[0056] This invention performs a first and second match between the current material segment's process phase fragment and the current response afterimage data and a defect shadow template at the edge end, calculates a comprehensive development risk value, and generates development warning data including the risky material segment number, the risk source station, the expected development station, the expected development distance, the defect type, and the risk level. This method can predict in advance which electrode segment might exhibit what type of defect at which station in the future, before the defect is identified by the visual inspection equipment, thus providing on-site operators with an earlier intervention window.

[0057] This invention generates a future workstation visualization warning layer based on development warning data, forming a risk development zone on the production line workstation diagram and material section operation diagram. It displays the workstation where the risk originates, the workstation expected to become visible, the defect type, the risk level, and the remaining intervention distance. This transforms the warning result from traditional text or audible / visual alarms into a visualized risk propagation map unfolding along the material movement direction. Operators can intuitively see that the source of the anomaly is located in the 3rd temperature zone, but the risk will appear at the exit of the 5th temperature zone, and the specific material section number will be affected, thereby improving the accuracy of on-site judgment and handling.

[0058] This invention, after adjusting nozzle pressure, exhaust volume, tension compensation, correction, or target linear velocity, continues to collect post-intervention process response data and compares it with pre-intervention response residual data to generate intervention rollback data. This allows the system to verify whether the intervention action truly reduced the overall development risk value. If the risk level decreases, the risk development zone is shortened or eliminated; if the risk level does not decrease, the risk material segment number is sent to the downstream quality inspection end. Therefore, this invention not only provides early warning but also establishes a control process that identifies, displays, implements, verifies, and re-inspects risks.

[0059] In summary, this invention enables material segment-level tracking, future workstation-level prediction, defect type-level judgment, and intervention distance-level display in continuous industrial processes such as lithium battery electrode coating and drying. This significantly improves the early warning lead time and on-site handling accuracy of latent defects, reduces the scrapping of large sections of material due to defects discovered at the back end, and enhances the stability and quality control capabilities of continuous production processes.

[0060] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0061] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.

Claims

1. A cloud-edge collaborative industrial process visualization and early warning method, characterized in that, Includes the following steps: Step S1: Collect production formula data, equipment action data and material location data at the edge end to generate material segment process phase segments; Step S2: Collect process response data at the edge and generate response afterimage data; Step S3: The cloud platform establishes a defect shadow template based on historical material segment process phase fragments, historical response afterimage data, and quality inspection result data; Step S4: At the edge end, development warning data is generated based on the process phase segment of the material segment, the response afterimage data, and the defect shadow template. Step S5: Generate a future workstation visualization warning layer based on the development warning data; Step S6: Perform intervention based on the development warning data and generate intervention rollback data.

2. The industrial process visualization and early warning method based on cloud-edge collaboration as described in claim 1, characterized in that, Step S1 includes the following sub-steps: Step S101: Collect production formula data at the edge end. The production formula data includes product model, slurry batch, target coating thickness, oven temperature zone setting, target linear velocity, and target tension. Step S102: Collect equipment action data at the edge end. The equipment action data includes coating valve opening, die head pressure, nozzle air pressure, exhaust volume, tension compensation amount, and correction execution amount. Step S103: Collect material position data at the edge end. The material position data includes the roll encoder pulse, material segment number, current position of the material segment, and expected arrival station of the material segment. Step S104: When the equipment action data changes beyond the action trigger threshold, the production formula data, equipment action data, and material location data are bound together to generate a material segment process phase segment.

3. The industrial process visualization and early warning method based on cloud-edge collaboration as described in claim 2, characterized in that, Step S2 includes the following sub-steps: Step S201: The edge end tracks the process of the material segment passing through the downstream station according to the material segment number; Step S202: Collect process response data, which includes edge temperature difference, surface reflection grayscale, solvent concentration change value, tension micro-amplitude, wet film thickness fluctuation value, and edge warping height. Step S203: Write the process response data into the process phase segment of the material segment in the order of material segment number, trigger station and downstream station to generate response afterimage data.

4. The industrial process visualization and early warning method based on cloud-edge collaboration as described in claim 3, characterized in that, Step S3 includes the following sub-steps: Step S301: Receive historical material segment process phase fragments, historical response afterimage data, and quality inspection result data from the cloud. Step S302: Group the products according to product model, slurry batch, target coating thickness, and oven temperature zone setting. Step S303: Within the same group, find the material segment whose quality inspection result data is abnormal, and extract the corresponding historical material segment process phase fragment and historical response afterimage data; Step S304: Combine the trigger station, the direction of change of equipment action data, the sequence of downstream stations, the direction of change of process response data, and the type of quality inspection defect into a defect shadow template.

5. The industrial process visualization and early warning method based on cloud-edge collaboration as described in claim 4, characterized in that, Step S4 includes the following sub-steps: Step S401: Obtain the current material segment process phase fragment and the current response afterimage data at the edge end; Step S402: Locate the defect shadow template corresponding to the current production formula data; Step S403: Perform a first match between the trigger station, the direction of change of equipment action data, and the expected arrival station of the material segment in the current material segment process phase segment and the defect shadow template; Step S404: Perform a second match between the process response data change direction and downstream station sequence in the current response afterimage data and the defect shadow template; Step S405: When both the first match and the second match meet the preset matching conditions, development warning data is generated. The development warning data includes the risk material segment number, the risk source station, the expected development station, the expected development distance, the defect type, and the risk level.

6. The industrial process visualization and early warning method based on cloud-edge collaboration as described in claim 5, characterized in that, Step S5 includes the following sub-steps: Step S501: Read the risk material segment number and current location of the material segment based on the development warning data; Step S502: Generate a risk development zone in the production line station diagram based on the current position of the material segment, the expected development station, and the expected development distance. Step S503: Mark the risk source station, expected development station, defect type, risk level, and remaining intervention distance on the risk development zone; Step S504: Output the risk imaging tape to the control terminal display interface.

7. The industrial process visualization and early warning method based on cloud-edge collaboration as described in claim 6, characterized in that, Step S6 includes the following sub-steps: Step S601: The edge end selects an intervention action based on the development warning data. The intervention action includes adjusting the nozzle pressure, adjusting the exhaust volume, adjusting the tension compensation amount, adjusting the correction execution amount, and adjusting the target linear velocity. Step S602: After performing the intervention action, continue to collect process response data; Step S603: Compare the process response data after intervention with the response afterimage data before intervention to generate intervention rollback data; Step S604: If several pre-retrograde data indicate a decrease in risk level, then shorten or cancel the risk imaging band. If the risk level does not decrease, the risk material segment number will be sent to the downstream quality inspection end.

8. The industrial process visualization and early warning method based on cloud-edge collaboration as described in claim 4, characterized in that, The logic for establishing the defect shadow template is as follows: The cloud uses the defect type in the quality inspection result data as the target result, traces back the historical response afterimage data of the same material segment, and then traces back the corresponding historical material segment process phase segment; If multiple abnormal material segments have the same trigger station, the same direction of change in equipment action data, and the same direction of change in process response data, they are recorded as the same defect shadow template.

9. The industrial process visualization and early warning method based on cloud-edge collaboration as described in claim 5, characterized in that, The logic for generating the risk level is as follows: At the edge, the number of triggering consistent items between the current material segment process phase fragment and the defect shadow template, and the number of response consistent items between the current response afterimage data and the defect shadow template are counted respectively. When both the number of triggering consensus items and the number of responding consensus items reach the high-risk threshold, the risk level is high-risk. If only one of the risk factors reaches the high-risk threshold, the risk level is medium risk; if none of them reach the threshold, the risk level is low risk.

10. The industrial process visualization and early warning method based on cloud-edge collaboration as described in claim 6, characterized in that, The display logic of the future workstation visualization early warning layer is as follows: The control terminal uses the material movement direction as the display axis, displays the risk source station at the current station position, displays the expected manifestation station at the downstream station position, and covers the risk manifestation band at the material segment position corresponding to the risk material segment number, so that operators can obtain a visual early warning result consisting of the source of the anomaly, the propagation path, the expected manifestation position, and the intervention result.

Images