An ai data hub based remote production control method and system

Abstract

The application discloses a remote production control method based on an AI data hub, which comprises the following steps: an AI data hub imports customer orders from an ERP platform, determines a production task, and sends a production instruction to a laser cutting machine, a slitting machine and a welding machine according to the production task; a correction machine adjusts a correction parameter within a correction range determined by a historical correction database according to material detection data after welding, a modulus value of a cutting adjustment vector, a modulus value of a slitting adjustment vector and a modulus value of the welding adjustment vector; after each correction is completed, the AI data hub updates the customer order and feeds back to the ERP platform and a remote client. Through deep integration of AI decision, equipment control and remote collaboration, the application realizes efficient, accurate and self-adaptive intelligent manufacturing remote control.

Description

Technical Field

[0001] This invention relates to the field of industrial production control, and in particular to a remote production control method and system based on an AI data hub. Background Technology

[0002] As the manufacturing industry rapidly moves towards intelligent and digital transformation, enterprises face multiple challenges, including fragmented orders, customized products, and flexible production. In the metal processing sector, particularly in high-precision manufacturing scenarios such as automotive parts and aerospace structural components, collaborative production involving multiple processes like laser cutting, slitting, and welding has become the industry standard. However, existing production control systems still have significant shortcomings in achieving efficient, precise, and intelligent production management, specifically in the following aspects: First, existing technologies typically only enable one-way transmission between "orders" and "work orders," failing to provide real-time feedback of production site data such as equipment operating parameters and quality inspection results to the ERP system. This leads to delays in order execution status updates, making it impossible for companies to accurately estimate delivery times and resulting in a decline in customer experience. Second, traditional production forecasting methods are mostly based on simple historical averages or linear extrapolations, considering only order quantity and standard working hours while neglecting crucial factors such as equipment parameter adjustments and historical production status. This results in low accuracy in production forecasting, preventing companies from rationally scheduling delivery. Summary of the Invention

[0003] This invention provides a remote production control method and system based on an AI data hub, which deeply integrates AI decision-making, equipment control and remote collaboration to achieve efficient, accurate and adaptive intelligent manufacturing control.

[0004] To achieve the above objectives, a first aspect of this application provides a remote production control method based on an AI data hub, comprising: The AI ​​data hub imports customer orders from the ERP platform, determines production tasks, and sends production instructions to the laser cutting machine, slitting machine, and welding machine based on the production tasks. The laser cutting machine adjusts the cutting parameters according to the production instructions and control instructions from a remote location, and generates a cutting adjustment vector based on the adjustment amount of the cutting parameters. The slitting machine adjusts the slitting parameters according to the production instructions and the control instructions from the remote control, and generates a slitting adjustment vector based on the adjustment amount of the slitting parameters; The welding machine adjusts the welding parameters according to the production instructions and the control instructions from the remote control, and generates a welding adjustment vector based on the adjustment amount of the welding parameters; The calibration machine adjusts the calibration parameters within the calibration range determined by the historical calibration database based on the post-weld material inspection data, the modulus of the cutting adjustment vector, the modulus of the slitting adjustment vector, and the modulus of the welding adjustment vector. After each correction is completed, the AI ​​data hub updates the customer order and feeds it back to the ERP platform and remote client.

[0005] In one possible implementation of the first aspect, before the AI ​​data hub updates the customer order and feeds it back to the ERP platform and the remote client, it specifically includes: The current production state vector is obtained based on the cutting parameters, the slitting parameters, the welding parameters, and the correction parameters; From the historical production record database, several production state vectors similar to the current production state vector are determined; each production state vector consists of daily average cutting parameters, daily average slitting parameters, daily average welding parameters, and daily average correction parameters, and each production state vector is mapped to a production result, and this mapping relationship is stored in the historical production record database; The production state vector is input into a preset machine learning model to obtain the first product output; Calculate the standard deviation of the production results corresponding to all the similar production state vectors; The output of the first product is adjusted based on the standard deviation to obtain an estimated future output of the product.

[0006] In one possible implementation of the first aspect, determining a plurality of production state vectors similar to the current production state vector specifically includes: In the historical production record database, if the yield rate of the production result is lower than the preset yield threshold or the production result has more than the preset number of defects, the production status vector corresponding to the production result will be excluded. Perform an approximate nearest neighbor search and return a preset number of similar production state vectors.

[0007] In one possible implementation of the first aspect, the laser cutting machine adjusts cutting parameters according to the production instructions and remotely controlled instructions, and generates a cutting adjustment vector based on the adjustment amount of the cutting parameters, specifically including: The values ​​of each cutting parameter are obtained from the production instruction, and the cutting parameters of the laser cutting machine are updated. The values ​​or correction values ​​of each cutting parameter are obtained from remote control commands, and the cutting parameters of the laser cutting machine are updated accordingly. A cutting adjustment vector is generated based on the adjustment amount of the cutting parameters and uploaded to the AI ​​data center.

[0008] In one possible implementation of the first aspect, the slitting machine adjusts slitting parameters according to the production instruction and the control instruction from a remote location, and generates a slitting adjustment vector based on the adjustment amount of the slitting parameters, specifically including: The values ​​of each slitting parameter are obtained from the production instruction, and the slitting parameters of the slitting machine are updated. The values ​​or correction values ​​of each slitting parameter are obtained from the control commands from the remote location, and the slitting parameters of the slitting machine are updated. Based on the adjustment amount of the striping parameters, a striping adjustment vector is generated and uploaded to the AI ​​data hub.

[0009] In one possible implementation of the first aspect, the welding machine adjusts welding parameters and generates a welding adjustment vector based on the adjustment amount of the welding parameters, according to the production instruction and the control instruction from a remote location, specifically including: The values ​​of each welding parameter are obtained from the production instruction, and the welding parameters of the welding machine are updated. The values ​​or correction values ​​of each welding parameter are obtained from the control commands received from a remote location, and the welding parameters of the welding machine are updated accordingly. A welding adjustment vector is generated based on the adjustment amount of the welding parameters and uploaded to the AI ​​data center.

[0010] In one possible implementation of the first aspect, the calibration machine adjusts calibration parameters within a calibration range determined by a historical calibration database based on post-weld material inspection data, the magnitude of the cutting adjustment vector, the magnitude of the slitting adjustment vector, and the magnitude of the welding adjustment vector, specifically including: The calibration machine obtains the dimensional deviation value based on the material inspection data after welding; If the magnitude of the cutting adjustment vector is less than or equal to the first magnitude threshold, and the magnitude of the slitting adjustment vector is less than or equal to the second magnitude threshold, and the magnitude of the welding adjustment vector is less than or equal to the third magnitude threshold, the table correction parameter is looked up in a preset lookup table according to the dimensional deviation value, and the correction parameter of the correction machine is adjusted according to the table correction parameter; if the table correction parameter is outside the correction range determined by the historical correction database, a request for manual correction is sent to the remote control terminal; the preset lookup table is a lookup table mapping the relationship between the dimensional deviation value and the table correction parameter; If the magnitude of the cutting adjustment vector is greater than the first magnitude threshold, the value of the correction parameter in each table of the preset lookup table is reduced according to the magnitude of the cutting adjustment vector. If the modulus of the strip adjustment vector is greater than the second modulus threshold, the value of the correction parameter of each table in the preset lookup table is reduced according to the modulus of the strip adjustment vector. If the magnitude of the welding adjustment vector is greater than the third magnitude threshold, the value of the correction parameter in each table of the preset lookup table is reduced according to the magnitude of the welding adjustment vector. The calibration parameters are found in the preset lookup table according to the dimensional deviation value. The calibration parameters of the calibration machine are adjusted according to the calibration parameters and the values ​​of each calibration parameter in the preset lookup table are reset. If the calibration parameters are outside the calibration range determined by the historical calibration database, a request for manual calibration is sent to the remote control terminal.

[0011] In one possible implementation of the first aspect, the specific confirmation process for the correction range determined by the historical correction database is as follows: The magnitudes of the cutting adjustment vector, the strip adjustment vector, and the welding adjustment vector are weighted and summed to obtain the comprehensive disturbance index. Select successful calibration cases from the historical calibration database that match the current operating conditions; In the selected calibration cases, the minimum value of the calibration parameter is chosen as the lower limit of the calibration range; In the selected calibration case, the maximum value of the calibration parameter is chosen as the upper limit of the calibration range; The correction range is adjusted based on the comprehensive disturbance index.

[0012] In one possible implementation of the first aspect, a weighted summation is performed on the magnitudes of the cutting adjustment vector, the striping adjustment vector, and the welding adjustment vector to obtain a comprehensive disturbance index, specifically including: The current calibration state vector is obtained based on the cutting parameters, the slitting parameters, and the welding parameters; The current calibration state vector is reduced in dimensionality by PCA in both the striping parameter dimension and the welding parameter dimension to obtain the first dimensionality-reduced state vector. The current calibration state vector is reduced in dimensionality by PCA in both the cutting parameter dimension and the welding parameter dimension to obtain a second dimensionality-reduced state vector. The current calibration state vector is reduced in dimensionality by PCA in both the cutting parameter dimension and the striping parameter dimension to obtain a third dimensionality-reduced state vector. The cosine similarity between the cutting adjustment vector and the first dimensionality-reduced state vector is taken as the cutting weighting coefficient; The cosine similarity between the strip adjustment vector and the second dimensionality-reduced state vector is taken as the strip weighting coefficient; The cosine similarity between the welding adjustment vector and the third dimensionality-reduced state vector is taken as the welding weighting coefficient; The comprehensive disturbance index is obtained by summing the products of the cutting weighting coefficient and the magnitude of the cutting adjustment vector, the product of the strip weighting coefficient and the magnitude of the strip adjustment vector, and the product of the welding weighting coefficient and the magnitude of the welding adjustment vector.

[0013] A second aspect of this application provides a remote production control system based on an AI data hub, including an AI data hub, an ERP platform, a production workshop, and a remote control terminal; the production workshop includes a laser cutting machine, a slitting machine, a welding machine, and a straightening machine; The AI ​​data hub imports customer orders from the ERP platform, determines production tasks, and sends production instructions to the laser cutting machine, the slitting machine, and the welding machine according to the production tasks. The laser cutting machine adjusts the cutting parameters according to the production instructions and control instructions from a remote location, and generates a cutting adjustment vector based on the adjustment amount of the cutting parameters. The slitting machine adjusts the slitting parameters according to the production instructions and the control instructions from the remote control, and generates a slitting adjustment vector based on the adjustment amount of the slitting parameters; The welding machine adjusts the welding parameters according to the production instructions and the control instructions from the remote control, and generates a welding adjustment vector based on the adjustment amount of the welding parameters; The calibration machine adjusts the calibration parameters within the calibration range determined by the historical calibration database based on the post-weld material inspection data, the modulus of the cutting adjustment vector, the modulus of the slitting adjustment vector, and the modulus of the welding adjustment vector. After each correction is completed, the AI ​​data hub updates the customer order and feeds it back to the ERP platform and remote client.

[0014] Compared to existing technologies, this embodiment provides a remote production control method and system based on an AI data hub. This system directly connects to the ERP platform via the AI ​​data hub, enabling automatic parsing and issuance of production instructions from customer orders without manual intervention, significantly reducing order response time. Simultaneously, it supports real-time issuance of control instructions from remote clients, allowing the production line to quickly adapt to order changes, process adjustments, or abnormal interventions, significantly improving the flexibility and agility of the manufacturing system.

[0015] Under a unified AI central control, key equipment such as laser cutting machines, slitting machines, and welding machines not only execute basic production tasks but also dynamically adjust process parameters based on remote commands, generating structured adjustment vectors (such as cutting adjustment vectors, slitting adjustment vectors, and welding adjustment vectors). These vectors quantify the deviations and adjustment ranges of each process, providing data support for subsequent calibration and enabling cross-process collaborative optimization. The calibration machine integrates the actual inspection data of the welded material with the modulus of the adjustment vectors of each process (reflecting the degree of process fluctuation) and intelligently adjusts the calibration parameters within a reasonable range defined by the historical calibration database, avoiding over-calibration or under-calibration. This mechanism effectively compensates for accumulated errors in previous processes, significantly improving the dimensional accuracy, deformation control, and overall quality consistency of the final product.

[0016] In summary, this method deeply integrates AI decision-making, equipment control, and remote collaboration, achieving efficient, precise, and adaptive intelligent manufacturing control, and has significant technological advancements and industrial application value. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a remote production control method based on an AI data hub, according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of a remote production control system based on an AI data hub, according to an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] To resolve the above issues, please refer to [link / reference]. Figure 1 An embodiment of the present invention provides a remote production control method based on an AI data hub, comprising: S10, the AI ​​data hub imports customer orders from the ERP platform, determines production tasks, and sends production instructions to the laser cutting machine, slitting machine, and welding machine according to the production tasks.

[0020] S11. The laser cutting machine adjusts the cutting parameters according to the production instructions and control instructions from a remote location, and generates a cutting adjustment vector based on the adjustment amount of the cutting parameters.

[0021] S12. The slitting machine adjusts the slitting parameters according to the production instructions and the control instructions from the remote control, and generates a slitting adjustment vector according to the adjustment amount of the slitting parameters.

[0022] S13. The welding machine adjusts the welding parameters according to the production instructions and the control instructions from the remote control, and generates a welding adjustment vector according to the adjustment amount of the welding parameters.

[0023] S14. The calibration machine adjusts the calibration parameters within the calibration range determined by the historical calibration database based on the post-weld material inspection data, the modulus of the cutting adjustment vector, the modulus of the slitting adjustment vector, and the modulus of the welding adjustment vector.

[0024] S15. After each correction is completed, the AI ​​data center updates the customer order and feeds it back to the ERP platform and remote client.

[0025] The S10 achieves seamless automatic conversion from order information to production execution, eliminating delays and errors in traditional manual scheduling and instruction transmission. The AI ​​data hub can intelligently analyze elements such as product specifications, quantity, and delivery date in orders, automatically generate suitable production tasks, and accurately distribute them to corresponding equipment, significantly improving the efficiency and accuracy of production plan execution and laying a data foundation for subsequent intelligent control of the entire process.

[0026] S11 achieves dynamic and precise cutting process control by integrating local production instructions with remote expert / system intervention instructions. The generated cutting adjustment vector quantifies the deviation between actual parameters and benchmark values ​​(such as laser power, cutting speed, focal position, etc.) in the form of structured data. This not only facilitates real-time monitoring of cutting quality but also provides key inputs for subsequent processes (such as calibration), enhancing the adaptability and traceability of the cutting process.

[0027] During the material slitting process, parameters such as slitting width, speed, and pressure can be dynamically optimized based on batch differences or real-time operating conditions (such as tension fluctuations and tool wear). The adjustment range and direction of the slitting process can be quantified through slitting adjustment vectors. S12 effectively avoids dimensional deviations or edge burrs caused by slitting errors, while providing data support for multi-process error fusion analysis and enhancing the overall consistency control capability of the production line.

[0028] In S13, welding, as a critical process prone to thermal deformation, is where even minor changes in its parameters (such as current, voltage, wire feed speed, and shielding gas flow rate) directly affect weld quality and structural deformation. This step achieves refined closed-loop management of the welding process through remote collaborative control and the generation of welding adjustment vectors. This ensures both weld strength and appearance quality, while also making the deformation trend caused by welding explicit in vector form, providing a preliminary predictive basis for subsequent corrections.

[0029] S14 breaks through the limitations of traditional calibration methods that rely solely on final test results. It innovatively integrates process adjustment information from preceding processes (characterizing the intensity of fluctuations in each process through vector magnitude) with actual test data to construct a multi-source information-driven intelligent calibration decision model. By combining this with a reasonable range defined by a historical calibration database, it avoids secondary damage or over-processing caused by blind calibration. While ensuring calibration effectiveness, it improves material utilization and equipment safety, significantly increasing the finished product qualification rate.

[0030] S15 establishes an end-to-end closed-loop feedback mechanism for production status, enabling key information such as order execution progress, process adjustment records, and quality correction results to be synchronized in real time to the enterprise management system (ERP) and remote monitoring terminals. On the one hand, it supports production managers in dynamically monitoring order fulfillment status and optimizing resource scheduling; on the other hand, it provides decision-making basis for remote experts, realizing a remote operation and maintenance closed loop of "monitoring-intervention-verification," thereby improving the transparency, collaborative efficiency, and customer satisfaction of the overall manufacturing system.

[0031] The above steps together constitute a complete intelligent production control closed loop of "order-driven—equipment collaboration—parameter self-adjustment—vector quantization—intelligent correction—status feedback." This not only enables remote and precise control of individual equipment but also significantly improves the flexibility, stability, and intelligence level of the production line through the fusion analysis and decision optimization of multi-process data via an AI data hub. It effectively reduces reliance on manual labor, decreases scrap rates, and shortens delivery cycles, providing reliable technical support for high-end manufacturing, flexible customization, and unmanned factories.

[0032] For example, before the AI ​​data hub updates the customer order and feeds it back to the ERP platform and remote client, it specifically includes: The current production state vector is obtained based on the cutting parameters, the slitting parameters, the welding parameters, and the correction parameters.

[0033] From the historical production record database, several production state vectors similar to the current production state vector are determined; each production state vector consists of daily average cutting parameters, daily average slitting parameters, daily average welding parameters, and daily average correction parameters, and each production state vector is mapped to a production result, and this mapping relationship is stored in the historical production record database.

[0034] The production state vector is input into a preset machine learning model to obtain the first product output.

[0035] Calculate the standard deviation of the production results corresponding to all the similar production state vectors.

[0036] The output of the first product is adjusted based on the standard deviation to obtain an estimated future output of the product.

[0037] First, based on the cutting parameters, slitting parameters, welding parameters, and correction parameters actually executed during the current production process, a multi-dimensional vector is constructed as the current production state vector. This vector is used to characterize the overall process configuration state of the current production line.

[0038] Secondly, several production state vectors similar to the current production state vector are retrieved from the historical production record database. Each production state vector stored in the historical production record database consists of the daily average cutting parameters, daily average slitting parameters, daily average welding parameters, and daily average correction parameters for the corresponding historical production cycle. Furthermore, each production state vector is mapped one-to-one with an actual production result (e.g., daily yield, pass rate, or output per unit time), and this mapping is pre-stored in the historical production record database.

[0039] Furthermore, the current production state vector is input into a pre-trained machine learning model, which outputs a preliminary product output prediction, denoted as the first product output. The machine learning model can be a regression model, an ensemble learning model, or a neural network model, and its training objective is to predict the corresponding production result based on the input process parameter vector.

[0040] Subsequently, for the several similar production state vectors obtained from the above retrieval, their corresponding production results are extracted, and the standard deviation of these production results is calculated. This standard deviation is used to quantify the degree of fluctuation of historical production results under similar process conditions, reflecting the level of uncertainty in output forecasting under current operating conditions.

[0041] Finally, the output of the first product is dynamically adjusted based on the standard deviation to obtain the final estimated output of the future product. For example, a large standard deviation indicates high dispersion and poor stability in production results under similar historical conditions, in which case a conservative correction (such as downward adjustment) can be applied to the output of the first product; conversely, a small standard deviation keeps the output of the first product basically unchanged or makes only minor adjustments. Through this mechanism, the output estimation not only relies on model predictions but also incorporates the statistical reliability of historical data, significantly improving the usability and robustness of the prediction results in actual production scheduling.

[0042] The estimated future product output will serve as a key production performance indicator, used to update the expected completion volume, delivery time, and resource utilization status of customer orders. This data will be synchronously fed back to the ERP platform and remote clients by the AI ​​data hub, enabling closed-loop management of order-production-collaboration.

[0043] For example, determining a plurality of production state vectors similar to the current production state vector specifically includes: In the historical production record database, if the yield rate of the production result is lower than the preset yield threshold or the production result has more than the preset number of defects, the production status vector corresponding to the production result will be excluded. Perform an approximate nearest neighbor search and return a preset number of similar production state vectors.

[0044] For example, the laser cutting machine adjusts the cutting parameters according to the production command and the remotely controlled command, and generates a cutting adjustment vector based on the adjustment amount of the cutting parameters, specifically including: The values ​​of each cutting parameter are obtained from the production instruction, and the cutting parameters of the laser cutting machine are updated. The values ​​or correction values ​​of each cutting parameter are obtained from remote control commands, and the cutting parameters of the laser cutting machine are updated accordingly. A cutting adjustment vector is generated based on the adjustment amount of the cutting parameters and uploaded to the AI ​​data center.

[0045] First, a quality screening is performed on all historical records in the historical production record database: if the yield rate of a certain historical record is lower than the preset yield threshold (e.g., 95%), or the number of defects is greater than the preset defect number threshold (e.g., 3 defects / batch), then the production status vector associated with that production result is excluded from the candidate set and will not participate in the subsequent similarity calculation. Subsequently, based on the remaining high-quality historical production state vectors, an Approximate Nearest Neighbor (ANN) search algorithm (e.g., using LSH, HNSW, or FAISS index structures) is executed. Using the current production state vector as the query vector, the algorithm retrieves and returns a preset number (e.g., k=10) of the production state vectors with the smallest Euclidean or cosine distance that meet the similarity threshold requirement from the historical database, serving as the final similar sample set. This screening mechanism ensures that the historical data used for subsequent output prediction and uncertainty assessment originates from stable and reliable production processes, effectively avoiding prediction bias caused by interference from poor-quality data.

[0046] Next, the production instruction is parsed to extract the target values ​​of each cutting parameter, including but not limited to laser power, cutting speed, focal position, auxiliary gas pressure, and nozzle diameter. Based on these values, the local process parameter configuration of the laser cutting machine is updated. A control instruction from a remote control terminal is received and parsed. This control instruction may contain the complete values ​​of the cutting parameters or incremental corrections (e.g., "cutting speed +5%)". According to the instruction content, the current cutting parameters of the laser cutting machine are updated a second time. Finally, the adjustment amount (i.e. change amount) of each cutting parameter relative to the previous value is calculated during this parameter update process, and all adjustment amounts are combined into a vector in a preset order, which is denoted as the cutting adjustment vector; the dimension of the vector is consistent with the number of cutting parameters, and its elements can be absolute change values ​​or relative percentages; the laser cutting machine uploads the cutting adjustment vector to the AI ​​data center in real time for subsequent disturbance assessment, correction decision and production status modeling.

[0047] For example, the slitting machine adjusts slitting parameters and generates a slitting adjustment vector based on the adjustment amount of the slitting parameters according to the production instruction and the control instruction from a remote location, specifically including: The values ​​of each slitting parameter are obtained from the production instruction, and the slitting parameters of the slitting machine are updated. The values ​​or correction values ​​of each slitting parameter are obtained from the control commands from the remote location, and the slitting parameters of the slitting machine are updated. Based on the adjustment amount of the striping parameters, a striping adjustment vector is generated and uploaded to the AI ​​data hub.

[0048] First, the production instruction is parsed to obtain the target values ​​for each slitting parameter, including but not limited to slitting width, slitting speed, tool clearance, tension setting, and winding pressure. Based on these values, the local process parameter configuration of the slitting machine is updated. This ensures that the slitting machine can automatically configure basic process parameters according to the production instruction automatically generated by the ERP order at the start of each batch of production, avoiding errors or delays caused by manual parameter setting.

[0049] Secondly, the system receives and parses control commands from the remote control terminal. These commands contain the complete values ​​or incremental corrections for each slitting parameter (e.g., "slitting width −0.1 mm" or "tension setting value +3%)". Based on the command content, the current slitting parameters of the slitting machine are updated a second time. This two-layer control architecture of "basic commands + remote correction" balances automated execution and anomaly response capabilities, making it suitable for flexible manufacturing scenarios with multiple product types and small batches. It supports both remotely issuing complete parameter sets (for large adjustments) and incremental corrections (for fine optimization), improving human-machine collaboration efficiency.

[0050] Finally, the adjustment amount of each strip parameter relative to its original value during this parameter update process is calculated, and all adjustments are combined into a vector in a preset order, denoted as the strip adjustment vector. The dimension of this vector is consistent with the number of strip parameters, and its elements can represent absolute change values ​​or relative change percentages. The striping machine uploads the strip adjustment vector to the AI ​​data center in real time for subsequent production disturbance assessment, correction decisions, and overall state modeling. Transforming parameter changes into a structured vector, whose modulus directly reflects the "disturbance intensity" of this adjustment, provides key input for the subsequent correction machine (such as determining whether conservative correction is needed).

[0051] For example, the welding machine adjusts welding parameters according to the production instruction and the control instruction from a remote location, and generates a welding adjustment vector based on the adjustment amount of the welding parameters, specifically including: The values ​​of each welding parameter are obtained from the production instruction, and the welding parameters of the welding machine are updated. The values ​​or correction values ​​of each welding parameter are obtained from the control commands received from a remote location, and the welding parameters of the welding machine are updated accordingly. A welding adjustment vector is generated based on the adjustment amount of the welding parameters and uploaded to the AI ​​data center.

[0052] The welding machine analyzes the production instructions (originating from ERP orders) issued by the AI ​​data center and automatically loads preset welding process parameters (such as current, voltage, speed, etc.) as the initial process benchmark for this task. This achieves task-driven automated initialization, avoiding errors from manual parameter setting; it also ensures the standardization and consistency of process execution across different orders, improving product quality stability.

[0053] During the welding process, the welding machine can receive control commands from remote experts, AI optimization modules, or online quality inspection feedback. These commands can be a complete set of parameters or incremental corrections (such as "current +5A"), and the welding machine dynamically adjusts the current parameters accordingly. Because it supports real-time online optimization, it can effectively cope with disturbances such as material fluctuations, equipment aging, or environmental changes; it also facilitates compatibility with various collaborative scenarios (expert remote guidance, AI adaptive optimization, etc.) and improves anomaly response efficiency.

[0054] The changes in each dimension of this parameter adjustment are calculated and combined into a structured vector (e.g., [ΔI, ΔV, Δv, ...]), and uploaded to the AI ​​data center in real time as a quantitative representation of production disturbances. This allows for the quantification of the intensity of process disturbances, using the modulus to assess the impact of this adjustment on overall production stability; furthermore, it supports multi-device collaborative decision-making: by fusing with cutting and slitting adjustment vectors, it can participate in AI inference such as correction strategy generation and output prediction.

[0055] Through the above three steps, the welding process is transformed from a "static execution unit" into an "intelligent sensing and feedback node," which not only improves welding quality and yield, but also provides key data support for the adaptive control, remote collaboration and AI-driven optimization of the entire production line, significantly enhancing the accuracy, efficiency and reliability of the intelligent manufacturing system.

[0056] For example, the calibration machine adjusts the calibration parameters within a calibration range determined by a historical calibration database based on post-weld material inspection data, the modulus of the cutting adjustment vector, the modulus of the slitting adjustment vector, and the modulus of the welding adjustment vector. Specifically, this includes: The calibration machine obtains the dimensional deviation value based on the material inspection data after welding; If the magnitude of the cutting adjustment vector is less than or equal to the first magnitude threshold, and the magnitude of the slitting adjustment vector is less than or equal to the second magnitude threshold, and the magnitude of the welding adjustment vector is less than or equal to the third magnitude threshold, the table correction parameter is looked up in a preset lookup table according to the dimensional deviation value, and the correction parameter of the correction machine is adjusted according to the table correction parameter; if the table correction parameter is outside the correction range determined by the historical correction database, a request for manual correction is sent to the remote control terminal; the preset lookup table is a lookup table mapping the relationship between the dimensional deviation value and the table correction parameter; If the magnitude of the cutting adjustment vector is greater than the first magnitude threshold, the value of the correction parameter in each table of the preset lookup table is reduced according to the magnitude of the cutting adjustment vector. If the modulus of the strip adjustment vector is greater than the second modulus threshold, the value of the correction parameter of each table in the preset lookup table is reduced according to the modulus of the strip adjustment vector. If the magnitude of the welding adjustment vector is greater than the third magnitude threshold, the value of the correction parameter in each table of the preset lookup table is reduced according to the magnitude of the welding adjustment vector. The calibration parameters are found in the preset lookup table according to the dimensional deviation value. The calibration parameters of the calibration machine are adjusted according to the calibration parameters and the values ​​of each calibration parameter in the preset lookup table are reset. If the calibration parameters are outside the calibration range determined by the historical calibration database, a request for manual calibration is sent to the remote control terminal.

[0057] The magnitudes of the cutting adjustment vector, slitting adjustment vector, and welding adjustment vector are obtained respectively, and compared with the preset first, second, and third magnitude thresholds. If all three are less than or equal to the corresponding thresholds, it indicates that the current production process is less disturbed and the state is stable; otherwise, it indicates that at least one process has significant parameter adjustments and the system is in a state of high disturbance.

[0058] When the disturbance level is low, the calibration parameters are obtained by directly querying the preset reference table (which stores the mapping relationship between the size deviation value and the recommended calibration parameters) based on the size deviation value, and the calibration parameters of the calibration machine are adjusted accordingly. If the calibration parameters exceed the calibration range determined by the historical calibration database (i.e., the reasonable parameter range in historical successful cases), a request for manual calibration is sent to the remote control terminal to prevent out-of-bounds operation.

[0059] If the magnitude of any adjustment vector exceeds its corresponding threshold (high disturbance condition), all table correction parameters in the preset lookup table are dynamically attenuated: cutting magnitude exceeding the limit → all correction parameters are reduced proportionally; slitting magnitude exceeding the limit → also reduced; welding magnitude exceeding the limit → also reduced; (linear attenuation, exponential attenuation, or weighted scaling based on the magnitude can be used). Based on the attenuated preset lookup table, the correction parameters are searched again according to the dimensional deviation value to adjust the correction machine; after adjustment, the parameter values ​​in the preset lookup table are reset to their original state to ensure that the next correction is not affected by this attenuation; if the correction parameters found at this time still exceed the historical correction range, a manual correction request is triggered.

[0060] In summary, the embodiments of the present invention set physical / process boundaries by historical correction ranges and combined with manual intervention mechanisms to eliminate dangerous operations; they mainly rely on table lookup and supplemented by attenuation, taking into account both real-time performance and stability, making them suitable for industrial field deployment; since the historical correction database can be continuously expanded with the accumulation of successful cases, the correction range and lookup table can be updated online.

[0061] For example, the specific confirmation process for the correction range determined by the historical correction database is as follows: The magnitudes of the cutting adjustment vector, the strip adjustment vector, and the welding adjustment vector are weighted and summed to obtain the comprehensive disturbance index. Select successful calibration cases from the historical calibration database that match the current operating conditions; In the selected calibration cases, the minimum value of the calibration parameter is chosen as the lower limit of the calibration range; In the selected calibration case, the maximum value of the calibration parameter is chosen as the upper limit of the calibration range; The correction range is adjusted based on the comprehensive disturbance index.

[0062] The Comprehensive Disturbance Index (D) is a quantitative indicator used to measure the overall disturbance intensity caused by multiple upstream processes (cutting, slitting, welding) to the quality of the final product during the current production process. Simply put, it answers the question: "How far has this batch of products deviated from its normal state due to parameter adjustments in the preceding processes?" A higher index indicates a more unstable system, requiring more careful subsequent corrections; a lower index indicates stable production, allowing for bolder corrections.

[0063] For example, the magnitudes of the cutting adjustment vector, the striping adjustment vector, and the welding adjustment vector are weighted and summed to obtain a comprehensive disturbance index, specifically including: The current calibration state vector is obtained based on the cutting parameters, the slitting parameters, and the welding parameters; The current calibration state vector is reduced in dimensionality by PCA in both the striping parameter dimension and the welding parameter dimension to obtain the first dimensionality-reduced state vector. The current calibration state vector is reduced in dimensionality by PCA in both the cutting parameter dimension and the welding parameter dimension to obtain a second dimensionality-reduced state vector. The current calibration state vector is reduced in dimensionality by PCA in both the cutting parameter dimension and the striping parameter dimension to obtain a third dimensionality-reduced state vector. The cosine similarity between the cutting adjustment vector and the first dimensionality-reduced state vector is taken as the cutting weighting coefficient; The cosine similarity between the strip adjustment vector and the second dimensionality-reduced state vector is taken as the strip weighting coefficient; The cosine similarity between the welding adjustment vector and the third dimensionality-reduced state vector is taken as the welding weighting coefficient; The comprehensive disturbance index is obtained by summing the products of the cutting weighting coefficient and the magnitude of the cutting adjustment vector, the product of the strip weighting coefficient and the magnitude of the strip adjustment vector, and the product of the welding weighting coefficient and the magnitude of the welding adjustment vector.

[0064] Comprehensive Disturbance Index The calculation formula is:

[0065] in, For the cutting adjustment vector, To adjust the vector in strips, For the welding adjustment vector, w1, w2, w3 This represents the weighting coefficient for the corresponding process.

[0066] w1, w2, w3It is not fixed, but dynamically calculated based on the current operating conditions: Construct the current calibration state vector, and then perform multi-view PCA dimensionality reduction: Perform PCA on the (slitting, welding) plane → obtain the first principal component (cutting) direction u1; Perform PCA on the (cutting, welding) plane → obtain the second principal component (strip) direction u2; Perform PCA on the (cutting, slitting) plane → obtain the third principal component (welding) direction u3; Then, cosine similarity is calculated as the weight. If "slitting and welding" are highly coupled under the current working condition (e.g., thin plate welding is easily affected by slitting tension), then PCA will extract strongly correlated directions on this plane. In this case, if the cutting adjustment vector is not similar to this direction (small cosine value), it means that the cutting change is an "independent disturbance" and should be given a higher weight. Conversely, if the adjustment of a certain process is consistent with the current main coupling direction, it may be a "coordinated adjustment", with a smaller disturbance, and the weight can be reduced.

[0067] The weights can adaptively reflect the disturbance contribution of each process under the current working conditions, which is more intelligent and accurate than manually setting fixed weights (such as w1=0.4, w2=0.3, w3=0.3).

[0068] After calculating D, it is used to dynamically adjust the correction range (as mentioned earlier): a large D → a narrower correction range → more conservative correction; a small D → a wider correction range → more aggressive correction. If D > Dmax (the historical maximum disturbance), the system can directly issue an alarm: "The current state exceeds historical experience; manual intervention is recommended."

[0069] Suppose that during a certain batch of production: the cutting machine significantly increases the laser power due to material reflection, resulting in... The slitting machine fine-tunes the width, making The welding machine current increased slightly, making .

[0070] If, under the current operating conditions, cutting has the greatest impact on the final dimensions (e.g., thermal deformation dominates the cutting of high-strength steel), the system uses PCA to determine that the cutting dimensions are highly independent → and calculates: w 1 = 0.7 w 2 = 0.1, w 3 = 0.2, then: D=0.7×8.0+0.1×0.5+0.2×1.2=5.6+0.05+0.24=5.89. This value of D is much higher than usual (usually D<2), and the system judges it as "severe disturbance". Therefore, the correction range is narrowed from [−0.5mm, +0.5mm] to [−0.2mm, +0.2mm]. Even if a size deviation of +0.4mm is detected, only +0.2mm correction is allowed to avoid overshoot. At the same time, the event is recorded for subsequent model optimization.

[0071] This comprehensive disturbance index calculation method represents an advanced application of data-driven decision-making in smart manufacturing. It not only considers "what has been adjusted" but also makes different adjustments in different contexts, thereby making a more accurate risk assessment.

[0072] Compared to existing technologies, this embodiment provides a remote production control method based on an AI data hub. This method directly connects to the ERP platform via the AI ​​data hub, enabling automatic parsing and issuance of production instructions from customer orders without manual intervention, significantly reducing order response time. Simultaneously, it supports real-time issuance of control instructions from remote clients, allowing the production line to quickly adapt to order changes, process adjustments, or abnormal interventions, significantly improving the flexibility and agility of the manufacturing system.

[0073] Under a unified AI central control, key equipment such as laser cutting machines, slitting machines, and welding machines not only execute basic production tasks but also dynamically adjust process parameters based on remote commands, generating structured adjustment vectors (such as cutting adjustment vectors, slitting adjustment vectors, and welding adjustment vectors). These vectors quantify the deviations and adjustment ranges of each process, providing data support for subsequent calibration and enabling cross-process collaborative optimization. The calibration machine integrates the actual inspection data of the welded material with the modulus of the adjustment vectors of each process (reflecting the degree of process fluctuation) and intelligently adjusts the calibration parameters within a reasonable range defined by the historical calibration database, avoiding over-calibration or under-calibration. This mechanism effectively compensates for accumulated errors in previous processes, significantly improving the dimensional accuracy, deformation control, and overall quality consistency of the final product.

[0074] In summary, this method deeply integrates AI decision-making, equipment control, and remote collaboration, achieving efficient, precise, and adaptive intelligent manufacturing control, and has significant technological advancements and industrial application value.

[0075] Please see Figure 2 One embodiment of this application provides a remote production control system based on an AI data hub, including an AI data hub 1, an ERP platform 2, a production workshop 3, and a remote control terminal 4; the production workshop 3 includes a laser cutting machine 31, a slitting machine 32, a welding machine 33, and a straightening machine 34.

[0076] The AI ​​data hub 1 imports customer orders from the ERP platform 2, determines production tasks, and sends production instructions to the laser cutting machine 31, the slitting machine 32, and the welding machine 33 according to the production tasks.

[0077] The laser cutting machine 31 adjusts the cutting parameters according to the production instructions and control instructions from a remote location, and generates a cutting adjustment vector based on the adjustment amount of the cutting parameters. The slitting machine 32 adjusts the slitting parameters according to the production instructions and the control instructions from the remote control, and generates a slitting adjustment vector based on the adjustment amount of the slitting parameters; The welding machine 33 adjusts the welding parameters according to the production instructions and the control instructions from the remote control, and generates a welding adjustment vector according to the adjustment amount of the welding parameters; The calibration machine 34 adjusts the calibration parameters within the calibration range determined by the historical calibration database based on the post-weld material inspection data, the modulus of the cutting adjustment vector, the modulus of the slitting adjustment vector, and the modulus of the welding adjustment vector. After each correction is completed, the AI ​​data hub 1 updates the customer order and feeds it back to the ERP platform 2 and the remote client 4.

[0078] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the remote production control system based on the AI ​​data hub described above can be referred to the corresponding process in the aforementioned method embodiments, and will not be elaborated further here.

[0079] Compared to existing technologies, this embodiment provides a remote production control system based on an AI data hub. This system directly connects to the ERP platform via the AI ​​data hub, enabling automatic parsing and issuance of production instructions from customer orders without manual intervention, significantly reducing order response time. Simultaneously, it supports real-time control command issuance from remote clients, allowing the production line to quickly adapt to order changes, process adjustments, or abnormal interventions, significantly improving the flexibility and agility of the manufacturing system.

[0080] Under a unified AI central control, key equipment such as laser cutting machines, slitting machines, and welding machines not only execute basic production tasks but also dynamically adjust process parameters based on remote commands, generating structured adjustment vectors (such as cutting adjustment vectors, slitting adjustment vectors, and welding adjustment vectors). These vectors quantify the deviations and adjustment ranges of each process, providing data support for subsequent calibration and enabling cross-process collaborative optimization. The calibration machine integrates the actual inspection data of the welded material with the modulus of the adjustment vectors of each process (reflecting the degree of process fluctuation) and intelligently adjusts the calibration parameters within a reasonable range defined by the historical calibration database, avoiding over-calibration or under-calibration. This mechanism effectively compensates for accumulated errors in previous processes, significantly improving the dimensional accuracy, deformation control, and overall quality consistency of the final product.

[0081] In summary, this method deeply integrates AI decision-making, equipment control, and remote collaboration, achieving efficient, precise, and adaptive intelligent manufacturing control, and has significant technological advancements and industrial application value.

[0082] One embodiment of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the remote production control method based on an AI data hub as described above.

[0083] The computer device may be a smartphone, tablet, desktop computer, or cloud server, among other computing devices. This computer device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the figures are merely examples of computer devices and do not constitute a limitation on the computer device. It may include more or fewer components than illustrated, or a combination of certain components, or different components, such as input / output devices, network access devices, etc.

[0084] The processor referred to can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0085] In some embodiments, the memory may be an internal storage unit of the computer device, such as a hard drive or RAM. In other embodiments, the memory may be an external storage device of the computer device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory may include both internal and external storage units of the computer device. The memory is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory can also be used to temporarily store data that has been output or will be output.

[0086] This application provides a computer program product that, when run on a computer device, enables the computer device to execute the steps described in the various method embodiments above.

[0087] In the several embodiments provided in this application, it will be understood that each block in the flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the figures. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved.

[0088] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0089] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A remote production control method based on an AI data hub, characterized in that, include: The AI ​​data hub imports customer orders from the ERP platform, determines production tasks, and sends production instructions to the laser cutting machine, slitting machine, and welding machine based on the production tasks. The laser cutting machine adjusts the cutting parameters according to the production instructions and control instructions from a remote location, and generates a cutting adjustment vector based on the adjustment amount of the cutting parameters. The slitting machine adjusts the slitting parameters according to the production instructions and the control instructions from the remote control, and generates a slitting adjustment vector based on the adjustment amount of the slitting parameters; The welding machine adjusts the welding parameters according to the production instructions and the control instructions from the remote control, and generates a welding adjustment vector based on the adjustment amount of the welding parameters; The calibration machine adjusts calibration parameters within a calibration range determined by a historical calibration database based on post-weld material inspection data, the magnitude of the cutting adjustment vector, the magnitude of the slitting adjustment vector, and the magnitude of the welding adjustment vector. Specifically, this includes: the calibration machine obtaining dimensional deviation values ​​based on post-weld material inspection data; if the magnitude of the cutting adjustment vector is less than or equal to a first magnitude threshold, and the magnitude of the slitting adjustment vector is less than or equal to a second magnitude threshold, and the magnitude of the welding adjustment vector is less than or equal to a third magnitude threshold, the calibration machine searches for table calibration parameters in a preset lookup table based on the dimensional deviation values, and adjusts the calibration parameters of the calibration machine according to the table calibration parameters; if the table calibration parameters are within a calibration range determined by a historical calibration database... Outside the correction range determined by the correction database, a request for manual correction is sent to the remote control terminal; the preset lookup table is a lookup table mapping the relationship between the dimensional deviation value and the table correction parameters; if the magnitude of the cutting adjustment vector is greater than the first magnitude threshold, the value of each table correction parameter in the preset lookup table is reduced according to the magnitude of the cutting adjustment vector; if the magnitude of the strip adjustment vector is greater than the second magnitude threshold, the value of each table correction parameter in the preset lookup table is reduced according to the magnitude of the strip adjustment vector; if the magnitude of the welding adjustment vector is greater than the third magnitude threshold, the value of each table correction parameter in the preset lookup table is reduced according to the magnitude of the welding adjustment vector. The value is determined; the table correction parameter is found in the preset lookup table according to the size deviation value; the correction parameters of the calibration machine are adjusted according to the table correction parameter and the values ​​of each table correction parameter in the preset lookup table are reset; if the table correction parameter is outside the correction range determined by the historical calibration database, a request for manual correction is sent to the remote control terminal; the correction range determined by the historical calibration database is specifically confirmed by: weighting and summing the magnitude of the cutting adjustment vector, the magnitude of the slitting adjustment vector and the magnitude of the welding adjustment vector to obtain the comprehensive disturbance index, specifically including: obtaining the current calibration state vector according to the cutting parameter, the slitting parameter and the welding parameter; The current calibration state vector is subjected to PCA dimensionality reduction in both the striping parameter dimension and the welding parameter dimension to obtain a first dimensionality-reduced state vector; the current calibration state vector is subjected to PCA dimensionality reduction in both the striping parameter dimension and the welding parameter dimension to obtain a second dimensionality-reduced state vector; the current calibration state vector is subjected to PCA dimensionality reduction in both the striping parameter dimension and the striping parameter dimension to obtain a third dimensionality-reduced state vector; the cosine similarity between the striping adjustment vector and the first dimensionality-reduced state vector is taken as the striping weighting coefficient; the cosine similarity between the striping adjustment vector and the second dimensionality-reduced state vector is taken as the striping weighting coefficient; the cosine similarity between the welding adjustment vector and the third dimensionality-reduced state vector is taken as the welding weighting coefficient;The comprehensive disturbance index is obtained by summing the products of the cutting weighting coefficient and the modulus of the cutting adjustment vector, the product of the strip weighting coefficient and the modulus of the strip adjustment vector, and the product of the welding weighting coefficient and the modulus of the welding adjustment vector. From the historical correction database, successful correction cases that match the current operating conditions are selected. Among the selected correction cases, the minimum value of the correction parameter is chosen as the lower limit of the correction range; among the selected correction cases, the maximum value of the correction parameter is chosen as the upper limit of the correction range. The correction range is then adjusted according to the comprehensive disturbance index. After each correction is completed, the AI ​​data hub updates the customer order and feeds it back to the ERP platform and remote client.

2. The remote production control method based on an AI data hub as described in claim 1, characterized in that, Before the AI ​​data hub updates the customer order and feeds it back to the ERP platform and remote client, it specifically includes: The current production state vector is obtained based on the cutting parameters, the slitting parameters, the welding parameters, and the correction parameters; From the historical production record database, several production state vectors similar to the current production state vector are determined; each production state vector consists of daily average cutting parameters, daily average slitting parameters, daily average welding parameters, and daily average correction parameters, and each production state vector is mapped to a production result, and this mapping relationship is stored in the historical production record database; The production state vector is input into a preset machine learning model to obtain the first product output; Calculate the standard deviation of the production results corresponding to all the similar production state vectors; The output of the first product is adjusted based on the standard deviation to obtain an estimated future output of the product.

3. The remote production control method based on an AI data hub as described in claim 2, characterized in that, The determination of several production state vectors similar to the current production state vector specifically includes: In the historical production record database, if the yield rate of the production result is lower than the preset yield threshold or the production result has more than the preset number of defects, the production status vector corresponding to the production result will be excluded. Perform an approximate nearest neighbor search and return a preset number of similar production state vectors.

4. The remote production control method based on an AI data hub as described in claim 1, characterized in that, The laser cutting machine adjusts cutting parameters according to the production instructions and remotely controlled instructions, and generates a cutting adjustment vector based on the adjustment amount of the cutting parameters, specifically including: The values ​​of each cutting parameter are obtained from the production instruction, and the cutting parameters of the laser cutting machine are updated. The values ​​or correction values ​​of each cutting parameter are obtained from remote control commands, and the cutting parameters of the laser cutting machine are updated accordingly. A cutting adjustment vector is generated based on the adjustment amount of the cutting parameters and uploaded to the AI ​​data center.

5. The remote production control method based on an AI data hub as described in claim 1, characterized in that, The slitting machine adjusts slitting parameters according to the production instructions and the control instructions from a remote location, and generates a slitting adjustment vector based on the adjustment amount of the slitting parameters, specifically including: The values ​​of each slitting parameter are obtained from the production instruction, and the slitting parameters of the slitting machine are updated. The values ​​or correction values ​​of each slitting parameter are obtained from the control commands from the remote location, and the slitting parameters of the slitting machine are updated. Based on the adjustment amount of the striping parameters, a striping adjustment vector is generated and uploaded to the AI ​​data hub.

6. The remote production control method based on an AI data hub as described in claim 1, characterized in that, The welding machine adjusts welding parameters according to the production instructions and the control instructions from a remote location, and generates a welding adjustment vector based on the adjustment amount of the welding parameters, specifically including: The values ​​of each welding parameter are obtained from the production instruction, and the welding parameters of the welding machine are updated. The values ​​or correction values ​​of each welding parameter are obtained from the control commands received from a remote location, and the welding parameters of the welding machine are updated accordingly. A welding adjustment vector is generated based on the adjustment amount of the welding parameters and uploaded to the AI ​​data center.

7. A remote production control system based on an AI data hub, characterized in that, It includes an AI data hub, an ERP platform, a production workshop, and a remote control terminal; the production workshop includes laser cutting machines, slitting machines, welding machines, and straightening machines; The AI ​​data hub imports customer orders from the ERP platform, determines production tasks, and sends production instructions to the laser cutting machine, the slitting machine, and the welding machine according to the production tasks. The laser cutting machine adjusts the cutting parameters according to the production instructions and control instructions from a remote location, and generates a cutting adjustment vector based on the adjustment amount of the cutting parameters. The slitting machine adjusts the slitting parameters according to the production instructions and the control instructions from the remote control, and generates a slitting adjustment vector based on the adjustment amount of the slitting parameters; The welding machine adjusts the welding parameters according to the production instructions and the control instructions from the remote control, and generates a welding adjustment vector based on the adjustment amount of the welding parameters; The calibration machine adjusts calibration parameters within a calibration range determined by a historical calibration database based on post-weld material inspection data, the modulus of the cutting adjustment vector, the modulus of the slitting adjustment vector, and the modulus of the welding adjustment vector. Specifically, this includes: the calibration machine obtaining dimensional deviation values ​​based on post-weld material inspection data; if the modulus of the cutting adjustment vector is less than or equal to a first modulus threshold, and the modulus of the slitting adjustment vector is less than or equal to a second modulus threshold, and the modulus of the welding adjustment vector is less than or equal to a third modulus threshold, the calibration machine searches for table calibration parameters in a preset lookup table based on the dimensional deviation values, and adjusts the calibration parameters of the calibration machine according to the table calibration parameters; if the table calibration parameters are within a range determined by a historical calibration database... Outside the correction range determined by the historical correction database, a request for manual correction is sent to the remote control terminal; the preset lookup table is a lookup table mapping the relationship between the dimensional deviation value and the table correction parameters; if the magnitude of the cutting adjustment vector is greater than the first magnitude threshold, the value of each table correction parameter in the preset lookup table is reduced according to the magnitude of the cutting adjustment vector; if the magnitude of the strip adjustment vector is greater than the second magnitude threshold, the value of each table correction parameter in the preset lookup table is reduced according to the magnitude of the strip adjustment vector; if the magnitude of the welding adjustment vector is greater than the third magnitude threshold, the value of each table correction parameter in the preset lookup table is reduced according to the magnitude of the welding adjustment vector. The value of the deviation is determined; the table correction parameter is found in the preset reference table according to the size deviation value; the correction parameter of the calibration machine is adjusted according to the table correction parameter and the value of each table correction parameter in the preset reference table is reset; if the table correction parameter is outside the correction range determined by the historical calibration database, a request for manual calibration is sent to the remote control terminal; the correction range determined by the historical calibration database is specifically confirmed by: weighting and summing the magnitude of the cutting adjustment vector, the magnitude of the slitting adjustment vector and the magnitude of the welding adjustment vector to obtain the comprehensive disturbance index, specifically including: obtaining the current calibration state vector according to the cutting parameter, the slitting parameter and the welding parameter; The current calibration state vector is subjected to PCA dimensionality reduction in both the striping parameter dimension and the welding parameter dimension to obtain a first dimensionality-reduced state vector; the current calibration state vector is subjected to PCA dimensionality reduction in both the striping parameter dimension and the welding parameter dimension to obtain a second dimensionality-reduced state vector; the current calibration state vector is subjected to PCA dimensionality reduction in both the striping parameter dimension and the striping parameter dimension to obtain a third dimensionality-reduced state vector; the cosine similarity between the striping adjustment vector and the first dimensionality-reduced state vector is taken as the striping weighting coefficient; the cosine similarity between the striping adjustment vector and the second dimensionality-reduced state vector is taken as the striping weighting coefficient; the cosine similarity between the welding adjustment vector and the third dimensionality-reduced state vector is taken as the welding weighting coefficient;The comprehensive disturbance index is obtained by summing the products of the cutting weighting coefficient and the modulus of the cutting adjustment vector, the product of the strip weighting coefficient and the modulus of the strip adjustment vector, and the product of the welding weighting coefficient and the modulus of the welding adjustment vector. From the historical correction database, successful correction cases that match the current operating conditions are selected. Among the selected correction cases, the minimum value of the correction parameter is chosen as the lower limit of the correction range; among the selected correction cases, the maximum value of the correction parameter is chosen as the upper limit of the correction range. The correction range is then adjusted according to the comprehensive disturbance index. After each correction is completed, the AI ​​data hub updates the customer order and feeds it back to the ERP platform and remote client.

Images