Cloud edge-end collaborative real-time leather quality detection lightweight model dynamic deployment method

Abstract

The invention relates to the technical field of cloud edge end collaborative computing, and discloses a cloud edge end collaborative real-time leather quality detection lightweight model dynamic deployment method. The method comprises the following steps: establishing a leather data acquisition module, a leather data calculation module, a cloud edge end collaborative model training module, a cloud edge end collaborative model optimization module, a cloud edge end collaborative model operation module, a cloud edge end collaborative model evaluation module and a cloud edge end collaborative model feedback module, the leather data calculation module is used for establishing a leather quality detection lightweight model and calculating data, the cloud edge end collaborative model training module is used for carrying out model training, and the cloud edge end collaborative model optimization module is used for carrying out model optimization; the cloud edge end collaborative model operation module deploys the optimized leather quality detection lightweight model to edge equipment of a leather production line, the cloud edge end collaborative model evaluation module evaluates the whole process, and the cloud edge end collaborative model feedback module automatically triggers corresponding improvement measures of the model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cloud edge-end collaborative computing, in particular to a cloud edge-end collaborative real-time leather quality detection lightweight model dynamic deployment method. BACKGROUND

[0002] In today's leather manufacturing industry, with the increasing market competition and the continuous improvement of consumers' quality requirements for leather products, ensuring the high quality of leather in the production process has become a crucial link. Traditional leather quality detection methods often rely on manual experience and simple physical detection means, which have many limitations. For example, manual detection is inefficient, difficult to analyze and process a large amount of production data in real time; the detection accuracy is easily affected by subjective factors, and it is difficult to accurately identify minor quality defects; and it lacks comprehensive evaluation ability for complex quality problems, making it difficult to analyze the root cause of the problem from multiple dimensions.

[0003] At the same time, although some existing automatic detection equipment can improve detection efficiency to some extent, it can only detect specific quality indicators and lacks systematic monitoring and comprehensive evaluation of the entire production process. For example, some equipment can only detect the thickness or surface flaws of leather, and cannot comprehensively consider the influence of strength, softness, and environmental factors in the production process on leather quality. In addition, traditional detection methods are difficult to adjust detection strategies and standards in the face of complex and variable production environment and process parameters, resulting in a disconnection between detection results and actual production needs, and cannot provide effective guidance and feedback for production.

[0004] With the continuous development of Internet of Things, big data and artificial intelligence technologies, although some attempts have been made to apply these new technologies to the field of leather quality detection, these applications are often in the early stages and have problems such as difficulty in data integration, complex model training and inapplicability to actual production environment. For example, some machine learning-based quality prediction models fail to fully consider the process characteristics of leather production and the particularity of actual production data, resulting in poor generalization ability and practicality of the model, and failing to play an effective role in actual production. SUMMARY

[0005] (I) Technical problems solved

[0006] In view of the deficiencies of the prior art, the present application provides a cloud edge-end collaborative real-time leather quality detection lightweight model dynamic deployment method, which has the advantages of efficiency, real-time, accuracy and dynamic adaptability, and solves the problems of low efficiency, insufficient accuracy, lack of systematic monitoring and comprehensive evaluation ability of traditional leather quality detection methods, and the single function of existing automatic detection equipment and the inability to dynamically adjust detection strategies.

[0007] (II) Technical solutions

[0008] To achieve the above object, the present application provides the following technical solution: a cloud-edge-end collaborative real-time leather quality detection lightweight model dynamic deployment method, comprising the following steps:

[0009] Step one, establish modules: leather data acquisition module, leather data calculation module, cloud-edge-end collaborative model training module, cloud-edge-end collaborative model optimization module, cloud-edge-end collaborative model running module, cloud-edge-end collaborative model evaluation module and cloud-edge-end collaborative model feedback module;

[0010] Step two, leather data acquisition module: deploy sensors, cameras and online monitors on the leather production line, responsible for collecting images, audio or equipment operation data during the leather production process, and transmitting the collected data to the leather data calculation module in real time;

[0011] Step three, leather data calculation module: establish a leather quality detection lightweight model according to the data collected by the leather data acquisition module, use the calculation formula in the leather quality detection lightweight model to perform real-time calculation and analysis on the collected data, and quickly judge whether the product has defects;

[0012] Step four, cloud-edge-end collaborative model training module: the cloud server inside the module receives the data uploaded by the leather data calculation module, uses the uploaded data to train the model, and transmits the distilled knowledge of the cloud-edge-end collaborative model training module to the leather quality detection lightweight model through the knowledge distillation technology;

[0013] Step five, cloud-edge-end collaborative model optimization module: use model pruning technology to optimize the leather quality detection lightweight model, remove redundant parameters, reduce the model size, and at the same time, the cloud adjusts the detection parameters according to the leather quality feedback information of the leather production line, and sends the updated parameters to the leather data calculation module;

[0014] Step six, cloud-edge-end collaborative model running module: deploy the optimized leather quality detection lightweight model to the edge device of the leather production line, and run the model in real time to quickly detect and analyze the collected leather production data;

[0015] Step seven, cloud-edge-end collaborative model evaluation module: according to the detection results of the leather data calculation module and the cloud-edge-end collaborative model running module, evaluate the entire leather production process;

[0016] Step eight, cloud-edge-end collaborative model feedback module: receive the evaluation information of the cloud-edge-end collaborative model evaluation module, when problems are found, the module will automatically trigger the corresponding improvement measures of the model.

[0017] Preferably, the leather data acquisition module comprises a leather production process data acquisition unit, a leather production environment impact data unit and a leather equipment operation data acquisition unit.

[0018] Preferably, the leather production process data acquisition unit acquires leather production process data in real time through high-precision sensors installed on production equipment, and the leather production process data includes leather thickness, leather strength and leather softness.

[0019] Preferably, the leather production environment impact data unit acquires leather production environment impact data through environmental monitoring sensors and monitoring systems of production equipment, and the leather production environment impact data includes the influence degree of temperature factors on leather surface defects, the influence degree of humidity factors on leather surface defects, the influence degree of ventilation factors on leather surface defects and the influence degree of other environmental pollutants on leather surface defects.

[0020] Preferably, the leather equipment operation data acquisition unit acquires leather equipment operation data through sensors and monitoring devices built in the equipment, and the leather equipment operation data includes actual values of different leather equipment operation parameters.

[0021] Preferably, the leather data calculation module comprises a leather comprehensive quality analysis unit, a leather surface defect analysis unit and a leather equipment operation parameter analysis unit.

[0022] Preferably, the leather comprehensive quality analysis unit calculates a leather quality comprehensive index Xe according to leather production process data, and the calculation formula is:

[0023] Xe = h * β1 + s * β2 + r * β3

[0024] In the formula, Xe represents the leather quality comprehensive index, h represents the leather thickness, s represents the leather strength, r represents the leather softness, β1, β2 and β3 represent weight coefficients corresponding to the leather thickness, strength and softness respectively, and β1 + β2 + β3 = 1.

[0025] Preferably, the leather surface defect analysis unit calculates an environmental impact leather surface defect coefficient Cl according to leather production environment impact data, and the calculation formula is:

[0026]

[0027] In the formula, Cl represents the environmental impact leather surface defect coefficient, w represents the influence degree of temperature factors on leather surface defects, t represents the influence degree of humidity factors on leather surface defects, v represents the influence degree of ventilation factors on leather surface defects, h represents the influence degree of other environmental pollutants on leather surface defects, G represents the overall mean value of the environmental impact leather surface defect coefficient, b represents the overall standard deviation, and g represents the leather surface defect coefficient of the leather production environment impact data unit.w , g t , g v , g h respectively represent the mean value of temperature, humidity, ventilation and other environmental pollution factors leading to leather surface defects, b w , b t , b v , b h respectively represent the standard deviation of temperature, humidity, ventilation and other environmental pollution factors leading to leather surface defects.

[0028] Preferably, the leather equipment operation parameter analysis unit calculates the leather quality and process parameter adaptation degree Zy according to the leather equipment operation data, and the calculation formula is:

[0029]

[0030] In the formula, Zy represents the leather quality and process parameter adaptation degree, f1, f2, f3, … f n respectively represent the actual value of different leather equipment operation parameters, μ1, μ2, μ3, … μ n respectively represent the standard value corresponding to different leather equipment operation parameters, α1, α2, α3, … α n respectively represent the weight coefficient of the leather equipment operation parameter, n represents the total number of equipment operation parameters, which can reflect the influence degree of different leather equipment operation parameters on the leather quality, |f i -μ i | represents the absolute difference between the actual value and the standard value, represents the relative deviation, represents the degree of process parameter close to the ideal state, and finally the weighted sum is obtained to obtain the adaptation degree.

[0031] Preferably, the cloud edge end collaborative model evaluation module evaluates the results according to the leather quality comprehensive index Xe, the environmental influence leather surface defect coefficient Cl and the leather quality and process parameter adaptation degree Zy.

[0032] Compared with the prior art, the application provides a cloud edge end collaborative real-time leather quality detection lightweight model dynamic deployment method, which has the following beneficial effects:

[0033] 1、The present application calculates the leather quality comprehensive index Xe through the formula, which considers the thickness, strength and softness of the leather to evaluate the quality of the leather. When the leather quality comprehensive index Xe is lower than the threshold set in the model, the cloud-edge collaborative model feedback module calls the pictures and video data collected by the sensors, cameras and online monitors deployed on the leather production line to check the details of the leather production line one by one and find the possible quality problem sources, such as equipment failure, raw material defects or improper operation, so as to solve the problems.

[0034] 2、The present application calculates the environmental impact on leather surface defect coefficient Cl to obtain a comprehensive and quantitative index to measure the influence of various factors in the production environment on the leather surface defects, rather than considering a single environmental factor, which can more comprehensively reflect the actual situation. When the environmental impact on leather surface defect coefficient Cl is lower than the threshold set in the model, the cloud-edge collaborative model feedback module automatically adjusts the system for controlling temperature, humidity, ventilation and other environmental pollutants to adapt to the current production environment, for example, adjusts the temperature setting value of the air conditioning system, the wind speed of the ventilation system or the parameters of the humidity control equipment, so as to optimize the production environment and reduce the influence of environmental factors on the quality of the leather.

[0035] 3、The present application calculates the leather quality and process parameter adaptation degree Zy to obtain a comprehensive and quantitative index to measure the matching degree between the actual production process parameters and the ideal standard process parameters, which helps to comprehensively understand whether the production process meets the requirements of high-quality production, rather than relying on a single parameter to judge. When the leather quality and process parameter adaptation degree Zy are lower than the threshold set in the model, the cloud-edge collaborative model feedback module feeds back the evaluation results to the production line control system to adjust the production process parameters in time, for example, adjusts the tanning time, dyeing temperature and coating thickness, reduces the quality problems caused by insufficient process parameter adaptation, and ensures the stability of the leather production process and the reliability of the product quality. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 The method flowchart of the present application. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0038] Please refer to Figure 1, a cloud edge collaborative real-time leather quality detection lightweight model dynamic deployment method, comprising the following steps:

[0039] Step one, establish modules: leather data acquisition module, leather data calculation module, cloud edge collaborative model training module, cloud edge collaborative model optimization module, cloud edge collaborative model running module, cloud edge collaborative model evaluation module and cloud edge collaborative model feedback module;

[0040] Step two, leather data acquisition module: deploy sensors, cameras and online monitors on the leather production line, responsible for collecting images, audio or equipment operation data during the leather production process, and transmitting the collected data to the leather data calculation module in real time;

[0041] Step three, leather data calculation module: establish a leather quality detection lightweight model according to the data collected by the leather data acquisition module, use the calculation formula in the leather quality detection lightweight model to perform real-time calculation and analysis on the collected data, quickly judge whether the product has defects, and upload part of the data to the cloud for model training for further model optimization;

[0042] Step four, cloud edge collaborative model training module: the cloud server inside the module receives the data uploaded by the leather data calculation module, trains the model using the uploaded data, and transmits the distilled knowledge of the cloud edge collaborative model training module to the leather quality detection lightweight model through the knowledge distillation technology, thereby improving the detection performance of the leather quality detection lightweight model;

[0043] Step five, cloud edge collaborative model optimization module: use model pruning technology to optimize the leather quality detection lightweight model, remove redundant parameters and reduce model size to better adapt to the resource constraints of the leather data calculation module, while the cloud adjusts the detection parameters according to the leather quality feedback information of the leather production line, and sends the updated parameters to the leather data calculation module;

[0044] Step six, cloud edge collaborative model running module: deploy the optimized leather quality detection lightweight model to the edge device of the leather production line, and run the model in real time to quickly detect and analyze the collected leather production data. During module operation, it will output detection results in real time to determine whether the leather has defects, and feedback the detection results to the production control system for timely adjustment of the production process or quality control. At the same time, the running module also synchronizes the key data of the detection results to the cloud to provide data support for subsequent model evaluation and feedback;

[0045] Step seven, cloud edge collaborative model evaluation module: according to the detection results of the leather data calculation module and the cloud edge collaborative model running module, evaluate the entire process of the entire leather production;

[0046] Step eight, cloud-edge-end collaborative model feedback module: receive the evaluation information of the cloud-edge-end collaborative model evaluation module, when a problem is found (such as a certain indicator is lower than the threshold value), the module will automatically trigger the corresponding improvement measures of the model, and in this process, new data is continuously collected to improve the detection ability of the model, for example, by continuously accumulating data under different quality problem conditions, the model can more accurately evaluate and predict the occurrence of quality problems, while optimizing the adjustment strategy in the feedback mechanism, making the entire real-time leather quality detection system of cloud-edge-end collaboration more efficient and reliable.

[0047] The leather data acquisition module includes a leather production process data acquisition unit, a leather production environment impact data unit, and a leather equipment operation data acquisition unit.

[0048] The leather production process data acquisition unit acquires real-time leather production process data through high-precision sensors installed on production equipment. The leather production process data includes leather thickness, leather strength, and leather softness.

[0049] The leather production environment impact data unit obtains leather production environment impact data through environmental monitoring sensors and monitoring systems built into production equipment. The leather production environment impact data includes the impact of temperature factors on leather surface defects, the impact of humidity factors on leather surface defects, the impact of ventilation factors on leather surface defects, and the impact of other environmental pollutants (such as dust and chemical gases) on leather surface defects.

[0050] The leather equipment operation data acquisition unit obtains leather equipment operation data through sensors and monitoring devices built into the equipment. The leather equipment operation data includes actual values of different leather equipment operation parameters.

[0051] The leather data calculation module includes a leather comprehensive quality analysis unit, a leather surface defect analysis unit, and a leather equipment operation parameter analysis unit.

[0052] The leather comprehensive quality analysis unit calculates the leather quality comprehensive index Xe based on the leather production process data. The calculation formula is:

[0053] Xe = h * β1 + s * β2 + r * β3

[0054] In the formula, Xe represents the leather quality comprehensive index, h represents the leather thickness, s represents the leather strength, r represents the leather softness, β1, β2, and β3 represent the weight coefficients corresponding to the leather thickness, strength, and softness, respectively, and β1 + β2 + β3 = 1;

[0055] The advantages are: the leather quality comprehensive index Xe is calculated using a formula that comprehensively considers the thickness, strength, and softness of the leather to calculate a comprehensive index to evaluate the quality of the leather. When the leather quality comprehensive index Xe is lower than the threshold set inside the model, the cloud-edge-device collaborative model feedback module will call the image and video data collected by the sensors, cameras, and online monitoring instruments deployed on the leather production line to check the details of the leather production line one by one, find the root cause of possible quality problems, such as equipment failure, raw material defects, or improper operation, so as to solve the problems in a targeted manner.

[0056] The leather surface defect analysis unit calculates the environmental impact coefficient Cl of leather surface defects based on the environmental impact data of leather production. The calculation formula is as follows:

[0057]

[0058] In the formula, Cl represents the environmental impact coefficient of leather surface defects; w represents the degree of influence of temperature on leather surface defects (excessively high or low temperatures may lead to defects such as drying and deformation of the leather, quantified as a specific numerical value); t represents the degree of influence of humidity on leather surface defects (humidity changes affect the moisture content and softness of the leather, thus affecting surface quality); v represents the degree of influence of ventilation on leather surface defects (good ventilation helps the leather dry evenly and chemicals volatilize, while poor ventilation may lead to odors and mold problems, expressed numerically); h represents the degree of influence of other environmental pollutants (such as dust and chemical gases) on leather surface defects (the corresponding value is determined according to the type and concentration of pollutants); G represents the overall mean of the environmental impact coefficient of leather surface defects, representing the average degree of influence under normal production conditions; b represents the overall standard deviation, reflecting the dispersion of the environmental impact coefficient; g w g t g v g h These represent the average values ​​of surface defects in leather caused by temperature, humidity, ventilation, and other environmental contaminants, respectively. w b t b v b h These represent the standard deviations of the defects on the leather surface caused by temperature, humidity, ventilation, and other environmental pollutants, respectively. By standardizing each environmental factor (i.e., subtracting the mean and then dividing by the standard deviation), they are transformed into comparable standard normal distribution variables, and then the weighted sum is used to obtain the final environmental impact coefficient.

[0059] The advantages are: by calculating the environmental impact on the leather surface defect coefficient Cl, a comprehensive and quantitative index can be obtained to measure the influence degree of various factors in the production environment on the leather surface defects, instead of considering a single environmental factor, which can more comprehensively reflect the actual situation. When the environmental impact on the leather surface defect coefficient Cl is lower than the threshold set in the model, the cloud edge end collaborative model feedback module automatically adjusts the system of controlling temperature, humidity, ventilation and other environmental pollutants according to the evaluation results to adapt to the current production environment, for example, adjusting the temperature setting value of the air conditioning system, the wind speed of the ventilation system or the parameters of the humidity control equipment, so as to optimize the production environment and reduce the influence of environmental factors on the quality of leather.

[0060] The leather equipment operation parameter analysis unit calculates the leather quality and process parameter adaptation degree Zy according to the leather equipment operation data, and the calculation formula is:

[0061]

[0062] In the formula, Zy represents the leather quality and process parameter adaptation degree, f1, f2, f3, … fn represent the actual values of different leather equipment operation parameters, μ1, μ2, μ3, … μn represent the standard values corresponding to different leather equipment operation parameters, α1, α2, α3, … αn represent the weight coefficients of the leather equipment operation parameters, and n represents the total number of the equipment operation parameters. n n n i i

[0063] The advantages are: by calculating the leather quality and process parameter adaptation degree Zy, a comprehensive and quantitative index can be obtained to measure the matching degree between the actual production process parameters and the ideal standard process parameters, which is helpful to comprehensively understand whether the production process meets the requirements of high-quality production, instead of relying on a single parameter for judgment. When the leather quality and process parameter adaptation degree Zy is lower than the threshold set in the model, the cloud edge end collaborative model feedback module will feed back the evaluation results to the production line control system, so as to adjust the production process parameters in time, for example, adjusting the tanning time, dyeing temperature and coating thickness, reducing the quality problems caused by insufficient process parameter adaptation, so as to ensure the stability of the leather production process and the reliability of the product quality.

[0064] ​​​​​​​The cloud-edge-end collaborative model evaluation module evaluates the results according to the leather quality comprehensive index Xe, the environment affecting leather surface defect coefficient Cl, and the leather quality and process parameter adaptation degree Zy.

[0065] The advantage is that by comprehensively evaluating the results of the three dimensions, the overall production capacity of the production line can be more accurately judged, for example, when the leather quality comprehensive index Xe is high, but the environment affecting leather surface defect coefficient Cl is also high and the process parameter adaptation degree Zy is low, it indicates that although the current produced leather itself may be of good quality, there may be a risk of surface defects caused by environmental factors in the production process and the process parameters need to be adjusted. Such comprehensive evaluation helps to fully understand the advantages and disadvantages of the production system, and can also accurately identify the key factors affecting leather production according to the evaluation results of different dimensions, for example, when the environment affecting leather surface defect coefficient Cl exceeds the threshold, it can be determined that factors such as temperature, humidity, dust in the environment have had an adverse effect on the leather surface, so that targeted measures can be taken to improve the environment. It can also timely discover potential quality problems or production hazards, for example, although the leather quality comprehensive index Xe has not decreased significantly in the current production, the abnormal changes of the environment affecting leather surface defect coefficient Cl and the process parameter adaptation degree Zy can indicate that the future leather quality has a downward trend, which enables production personnel to take measures to adjust before the problem really affects product quality, improves the controllability of production, and solves the problems of low efficiency, insufficient precision, lack of systematic monitoring and comprehensive evaluation capability of traditional leather quality detection methods, and the problems of single function and inability to dynamically adjust detection strategies of existing automatic detection equipment.

[0066] Although embodiments of the present application have been shown and described, it is to be understood that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the present application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A dynamic deployment method for a lightweight model of real-time leather quality inspection with cloud-edge-device collaboration, characterized in that, Includes the following steps: Step 1: Establish modules: Leather data acquisition module, leather data calculation module, cloud-edge-device collaborative model training module, cloud-edge-device collaborative model optimization module, cloud-edge-device collaborative model operation module, cloud-edge-device collaborative model evaluation module, and cloud-edge-device collaborative model feedback module; Step 2, Leather Data Acquisition Module: Sensors, cameras, and online monitoring instruments are deployed on the leather production line to collect images, audio, or equipment operation data during the leather production process and transmit the collected data to the leather data calculation module in real time. Step 3, Leather Data Calculation Module: Based on the data collected by the leather data acquisition module, a lightweight leather quality inspection model is established. The calculation formula in the lightweight leather quality inspection model is used to perform real-time calculation and analysis on the collected data to quickly determine whether there are defects in the product. Step 4: Cloud-Edge-Device Collaborative Model Training Module: The cloud server inside the module receives the data uploaded by the leather data calculation module, uses the uploaded data for model training, and transfers the distilled knowledge from the cloud-edge-device collaborative model training module to the lightweight leather quality inspection model through knowledge distillation technology. Step 5, Cloud-Edge Collaborative Model Optimization Module: The lightweight model for leather quality inspection is optimized using model pruning technology to remove redundant parameters and reduce the model size. At the same time, the cloud adaptively adjusts the inspection parameters based on the leather quality feedback information from the leather production line and sends the updated parameters to the leather data calculation module. Step Six: Cloud-Edge-Device Collaborative Model Execution Module: Deploy the optimized lightweight leather quality inspection model to the edge devices of the leather production line and run the model in real time to quickly detect and analyze the collected leather production data; Step 7: Cloud-Edge-Device Collaborative Model Evaluation Module: Based on the detection results of the leather data calculation module and the cloud-edge-device collaborative model operation module, evaluate the entire leather production process; Step 8: Cloud-Edge-Device Collaborative Model Feedback Module: Receives evaluation information from the Cloud-Edge-Device Collaborative Model Evaluation Module. When a problem is detected, the module will automatically trigger corresponding improvement measures for the model.

2. The dynamic deployment method for a lightweight real-time leather quality inspection model with cloud-edge-device collaboration according to claim 1, characterized in that: The leather data acquisition module includes a leather production process data acquisition unit, a leather production environmental impact data unit, and a leather equipment operation data acquisition unit.

3. The dynamic deployment method for a lightweight real-time leather quality inspection model with cloud-edge-device collaboration according to claim 2, characterized in that: The leather production process data acquisition unit collects leather production process data in real time through high-precision sensors installed on the production equipment. The leather production process data includes leather thickness, leather strength, and leather softness.

4. The dynamic deployment method for a lightweight real-time leather quality inspection model with cloud-edge-device collaboration according to claim 2, characterized in that: The leather production environmental impact data unit acquires leather production environmental impact data through environmental monitoring sensors and the monitoring system built into the production equipment. The leather production environmental impact data includes the degree of influence of temperature factors on leather surface defects, the degree of influence of humidity factors on leather surface defects, the degree of influence of ventilation factors on leather surface defects, and the degree of influence of other environmental pollutants on leather surface defects.

5. The dynamic deployment method for a lightweight real-time leather quality inspection model with cloud-edge-device collaboration according to claim 2, characterized in that: The leather equipment operation data acquisition unit acquires leather equipment operation data through built-in sensors and monitoring devices. The leather equipment operation data includes the actual values ​​of different leather equipment operation parameters.

6. The dynamic deployment method for a lightweight real-time leather quality inspection model with cloud-edge-device collaboration according to claim 1, characterized in that: The leather data calculation module includes a leather comprehensive quality analysis unit, a leather surface defect analysis unit, and a leather equipment operating parameter analysis unit.

7. The dynamic deployment method for a lightweight real-time leather quality inspection model with cloud-edge-device collaboration according to claim 6, characterized in that: The comprehensive leather quality analysis unit calculates the comprehensive leather quality index Xe based on leather production process data. The calculation formula is as follows: Xe=h*β1+s*β2+r*β3 In the formula, Xe represents the comprehensive leather quality index, h represents leather thickness, s represents leather strength, r represents leather softness, and β1, β2, and β3 represent the weighting coefficients corresponding to leather thickness, strength, and softness, respectively, and β1+β2+β3=1.

8. The dynamic deployment method for a lightweight real-time leather quality inspection model with cloud-edge-device collaboration according to claim 6, characterized in that: The leather surface defect analysis unit calculates the environmental impact coefficient Cl of leather surface defects based on the environmental impact data of leather production. The calculation formula is as follows: In the formula, Cl represents the environmental impact coefficient of leather surface defects, w represents the influence of temperature on leather surface defects, t represents the influence of humidity on leather surface defects, v represents the influence of ventilation on leather surface defects, h represents the influence of other environmental pollutants on leather surface defects, G represents the population mean of the environmental impact coefficient of leather surface defects, b represents the population standard deviation, and g w g t g v g h b represents the average value of leather surface defects caused by temperature, humidity, ventilation, and other environmental pollution factors, respectively. w b t b v b h These represent the standard deviations corresponding to defects on the leather surface caused by temperature, humidity, ventilation, and other environmental pollution factors.

9. The dynamic deployment method for a lightweight real-time leather quality inspection model with cloud-edge-device collaboration according to claim 6, characterized in that: The leather equipment operation parameter analysis unit calculates the leather quality and process parameter compatibility Zy based on the leather equipment operation data. The calculation formula is as follows: In the formula, Zy represents the fit between leather quality and process parameters, and f1, f2, f3, ... f n These represent the actual values ​​of different leather processing equipment operating parameters, μ1, μ2, μ3, ... μ n These represent the standard values ​​of the operating parameters for different leather processing equipment, α1, α2, α3, ... α n These represent the weighting coefficients of the leather equipment operating parameters, where n represents the total number of equipment operating parameters. This reflects the degree of influence of different leather equipment operating parameters on leather quality. |f i -μ i | represents the absolute difference between the actual value and the standard value. Indicates relative deviation. This indicates the degree to which the process parameters approach the ideal state, and the final weighted sum is the fit degree.

10. The dynamic deployment method for a lightweight real-time leather quality inspection model with cloud-edge-device collaboration according to claim 9, characterized in that: The cloud-edge-device collaborative model evaluation module evaluates the results based on the comprehensive leather quality index Xe, the environmental impact coefficient of leather surface defects Cl, and the leather quality and process parameter fit Zy.

Images