Gold foil processing production control method and system based on artificial intelligence

Abstract

The present invention application discloses a gold foil processing production control method and system based on artificial intelligence, which belongs to the field of industrial automation technology. The present invention application constructs a comprehensive control decision data set by comprehensively considering the equipment status monitoring data and the production environment sampling data. The control decision data set can reflect the comprehensive operating status of the target processing production line under different feature recognition weights, and provide accurate guidance for working condition adjustment. At the same time, through the analysis of the data set, the present invention application can also obtain working condition adjustment discrimination information, realize accurate control of the intelligent robot, thereby comprehensively improving the intelligence level and overall performance of the production line.

Description

Technical Field

[0001] The present invention belongs to the field of industrial automation technology, and specifically relates to a gold foil processing production control method and system based on artificial intelligence. Background Art

[0002] In the current industrial production environment, equipment status monitoring and environmental control of processing production lines are key links to ensure production efficiency and product quality. However, traditional methods often only focus on the operating status of the equipment itself, ignoring the important impact of the production environment on equipment performance and product quality. At the same time, there is a lack of comprehensive analysis and linkage consideration of the operating status of the equipment under different feature recognition weights, resulting in a lack of scientific basis for working condition adjustment decisions, making it difficult to achieve the optimal effect.

[0003] Especially in complex and ever-changing production scenarios, the interaction between equipment status and environmental factors is more complex, and single-dimensional monitoring and adjustment can no longer meet actual needs. In addition, with the widespread application of intelligent robots in production lines, how to achieve precise control of these robots to adapt to changing production conditions has also become an urgent problem to be solved. Summary of the invention

[0004] The present invention application provides a gold foil processing production control method and system based on artificial intelligence, which can solve or partially solve the technical problems involved in the above-mentioned background technology.

[0005] The present invention provides a gold foil processing production control method based on artificial intelligence, which is applied to a gold foil processing production control system. The method comprises: obtaining equipment status monitoring data and production environment sampling data of a target processing production line; the equipment status monitoring data comprises an equipment operation state vector of the target processing production line under a first feature recognition weight and an equipment operation state vector of the target processing production line under a second feature recognition weight; obtaining a control decision data set of the target processing production line according to the equipment status monitoring data and the production environment sampling data; the control decision data set comprises an equipment operation state vector of the target processing production line under a first feature recognition weight and a linkage production line description vector of the target processing production line under X feature recognition weights, wherein the linkage production line description vector of the target processing production line under any feature recognition weight is obtained according to the equipment operation state vector of the target processing production line under the second feature recognition weight and the production environment sampling data, where X is a positive integer; performing a working condition adjustment analysis on the target processing production line according to the control decision data set to obtain working condition adjustment discrimination information of the target processing production line; the working condition adjustment discrimination information is used to instruct at least one intelligent robot of the target processing production line to perform control processing.

[0006] The present invention application provides a gold foil processing production control system, comprising at least one processor and a memory; the memory stores computer execution instructions; the at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the above method.

[0007] The present invention application provides a readable storage medium, on which a program or instruction is stored, and the program or instruction implements the steps of the above method when executed by a processor.

[0008] The present invention application constructs a comprehensive control decision data set by comprehensively considering the equipment status monitoring data and the production environment sampling data. The control decision data set can reflect the comprehensive operating status of the target processing production line under different feature recognition weights, and provide accurate guidance for working condition adjustment. At the same time, through the analysis of the data set, the present invention application can also obtain working condition adjustment discrimination information, realize accurate control of the intelligent robot, thereby comprehensively improving the intelligence level and overall performance of the production line. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 The present invention provides a flowchart of a gold foil processing and production control method based on artificial intelligence.

[0010] Figure 2 This is a structural schematic diagram of a gold foil processing and production control system provided in the present invention application. DETAILED DESCRIPTION

[0011] The following will be combined with the drawings in the present invention application to clearly and completely describe the technical solutions in the present invention application. Obviously, the described embodiments are part of the embodiments of the present invention application, rather than all the embodiments. Based on the embodiments in the present invention application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention application.

[0012] The terms "first", "second", etc. in the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the present application represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0013] Figure 1A gold foil processing production control method based on artificial intelligence is shown, which is applied to a gold foil processing production control system. The method includes the following steps 110 to 130.

[0014] In modern precision manufacturing, gold foil is a high-end material, and its processing and production process is highly sensitive to equipment status and production environment. In order to ensure the quality and production efficiency of gold foil products, an efficient and intelligent gold foil processing and production control system is particularly important. The gold foil processing and production control system proposed in the present invention realizes the comprehensive optimization of the gold foil processing production line through precise data collection, intelligent analysis and decision-making, and flexible working condition adjustment.

[0015] Step 110: The gold foil processing production control system obtains the equipment status monitoring data and production environment sampling data of the target processing production line.

[0016] The equipment status monitoring data includes an equipment operation status vector of the target processing production line under a first feature recognition weight and an equipment operation status vector of the target processing production line under a second feature recognition weight.

[0017] The primary task of the gold foil processing production control system is to obtain comprehensive data of the target processing production line, which is the basis for subsequent analysis and decision-making. Specifically, the gold foil processing production control system needs to collect two types of data: equipment status monitoring data and production environment sampling data.

[0018] 1) Equipment status monitoring data directly reflects the operating status of each device on the production line. In order to capture the equipment status more comprehensively, the gold foil processing production control system uses multiple feature recognition weights for monitoring. Under the first feature recognition weight, the gold foil processing production control system records a set of equipment operating status vectors, which can include key parameters such as the equipment's operating speed, temperature, and vibration. At the same time, under the second feature recognition weight, the gold foil processing production control system also obtains another set of equipment operating status vectors, which can focus on aspects such as the equipment's energy consumption, load, and maintenance requirements. Through the comprehensive consideration of these two weights, the gold foil processing production control system can more accurately evaluate the overall performance and potential risks of the equipment.

[0019] 2) Production environment sampling data: In addition to the equipment status, the impact of the production environment on gold foil processing cannot be ignored. The gold foil processing production control system needs to monitor and record key indicators such as temperature, humidity, and air cleanliness of the production environment in real time. These production environment sampling data are of great significance for subsequent analysis of potential problems in the production process and optimization of production conditions.

[0020] Step 120: The gold foil processing production control system obtains a control decision data set of the target processing production line based on the equipment status monitoring data and the production environment sampling data.

[0021] Among them, the control decision data set includes the equipment operation state vector of the target processing production line under the first feature recognition weight and the linkage production line description vector of the target processing production line under X feature recognition weights, and the linkage production line description vector of the target processing production line under any feature recognition weight is obtained based on the equipment operation state vector of the target processing production line under the second feature recognition weight and the production environment sampling data, and X is a positive integer.

[0022] After obtaining comprehensive data, the gold foil processing production control system began to enter the data analysis stage. The goal of this stage is to generate a control decision data set to guide production decisions based on the collected data.

[0023] 1) Composition of the control decision data set: The control decision data set is the basis of the system's intelligent decision-making. It contains the equipment operation state vector and the linkage production line description vector under multiple feature recognition weights. Under the first feature recognition weight, the gold foil processing production control system continues to retain the equipment operation state vector so that the direct operation state of the equipment can be considered when making decisions. Under X feature recognition weights (X is a positive integer, indicating that the gold foil processing production control system considers multiple different weights or dimensions), the gold foil processing production control system generates a linkage production line description vector. These vectors not only take into account the status of a single device, but also comprehensively consider the operation status and production environment of the entire production line and even the related linkage production line.

[0024] 2) Generation of linkage production line description vectors: It is particularly worth mentioning that the generation of linkage production line description vectors is the key to the intelligent decision-making of the gold foil processing production control system. The gold foil processing production control system first conducts a comprehensive evaluation of the equipment's potential failures, production efficiency, energy consumption, etc. based on the equipment operation state vector and production environment sampling data under the second feature recognition weight. Then, the gold foil processing production control system combines these evaluation results with the data under the first feature recognition weight to generate linkage production line description vectors under different feature recognition weights. These vectors describe the optimal operating status and possible adjustment direction of the entire production line and the associated linkage production line under different conditions.

[0025] Step 130: The gold foil processing production control system performs a working condition adjustment analysis on the target processing production line according to the control decision data set to obtain working condition adjustment discrimination information of the target processing production line.

[0026] The working condition adjustment discrimination information is used to instruct control processing of at least one intelligent robot of the target processing production line.

[0027] After the control decision data set is generated, the gold foil processing production control system begins to enter the working condition adjustment analysis stage. The goal of this stage is to conduct an in-depth analysis of the working conditions of the target processing production line based on the control decision data set and obtain working condition adjustment judgment information.

[0028] 1) The significance of the working condition adjustment discrimination information: The working condition adjustment discrimination information is the basis for the gold foil processing production control system to actually control the production line. It indicates how the gold foil processing production control system should control at least one intelligent robot on the production line to achieve the purpose of optimizing production. These control processes may include adjusting the robot's operating speed, changing processing parameters, and performing equipment maintenance.

[0029] 2) Generation of working condition adjustment discrimination information: The gold foil processing production control system first conducts a comprehensive assessment of the current working condition of the production line based on the equipment operation state vector and linkage production line description vector in the control decision data set. Then, the gold foil processing production control system generates working condition adjustment discrimination information based on the actual needs of the production line (such as output, quality, energy consumption, etc.). This information can include specific adjustment suggestions, adjustment timing, adjustment range, etc. to ensure that the production line operates in the optimal state.

[0030] 3) Control processing of intelligent robots: After obtaining the working condition adjustment and discrimination information, the gold foil processing production control system begins to control the intelligent robots on the production line. These robots may be responsible for key links such as cutting, grinding, and packaging of gold foil. The gold foil processing production control system sends control instructions to the robots to adjust their operating status and processing parameters to ensure the quality and production efficiency of gold foil products. At the same time, the gold foil processing production control system also monitors the operating status and processing effect of the robots in real time to adjust the control strategy in time.

[0031] In an exemplary application scenario, a gold foil processing production line consists of four main devices: a cutting machine, a grinding machine, a coating machine, and a packaging machine. The gold foil processing production control system needs to monitor the operating status and production environment of these four devices in real time, and generate a control decision data set based on the data, and then adjust the working conditions to optimize production.

[0032] 1. Equipment status monitoring data

[0033] 1.1 Cutting machine

[0034] Cutting speed: The initial setting is 15m / s, but real-time monitoring data shows that the current speed is 13m / s; Motor temperature: The normal range is 40-60℃, and the current temperature is 55℃; Vibration amplitude: The normal range is 0-0.2mm, and the current vibration amplitude is 0.15mm.

[0035] 1.2. Grinding machine

[0036] Grinding speed: The initial setting is 8m / s, and real-time monitoring data shows that the current speed is 7.5m / s; Motor temperature: The normal range is 35-55℃, and the current temperature is 45℃; Vibration amplitude: The normal range is 0-0.1mm, and the current vibration amplitude is 0.08mm; Abrasive consumption rate: The normal consumption is 1kg per hour, and the current consumption rate is 1.2kg / h.

[0037] 1.3 Coating machine

[0038] Coating speed: Initial setting is 6m / s, real-time monitoring data shows that the current speed is 5.8m / s; coating thickness: target thickness is 0.01mm, current coating thickness is 0.009mm; vacuum degree: normal range is 10 -3 -10 -4 Pa, the current vacuum degree is 10 -3.5 Pa.

[0039] 1.4 Packaging machine

[0040] Packaging speed: The initial setting is 10m / s, and real-time monitoring data shows that the current speed is 9.5m / s; Packaging material consumption rate: Normal consumption is 2000m per hour, and the current consumption rate is 2100m / h; Motor temperature: The normal range is 30-50℃, and the current temperature is 40℃.

[0041] 2. Production environment sampling data

[0042] Temperature: The target temperature of the production environment is 25°C, and the current temperature is 24.8°C; Humidity: The target humidity of the production environment is 50%, and the current humidity is 48%; Air cleanliness: The production environment requires air cleanliness of level 100, and the current cleanliness is level 98.

[0043] 3. Control decision data set generation

[0044] The gold foil processing production control system generates the following control decision data set based on the equipment status monitoring data and production environment sampling data.

[0045] Cutting machine: The cutting speed is low, it is recommended to adjust it to 14m / s to improve efficiency; the motor temperature and vibration amplitude are both within the normal range and no adjustment is required.

[0046] Grinder: The grinding speed is low and the abrasive consumption rate is high. It is recommended to adjust the grinding parameters to the optimal state, such as adjusting the grinding pressure or replacing more wear-resistant abrasives to improve grinding efficiency and reduce consumption; the motor temperature and vibration amplitude are within the normal range and no adjustment is required.

[0047] Coating machine: The coating speed is too low and the coating thickness has not reached the target value. It is recommended to adjust the coating parameters, such as increasing the coating time or adjusting the coating material, to ensure the coating quality; the vacuum degree is within the normal range and no adjustment is required.

[0048] Packaging machine: The packaging speed is low and the packaging material consumption rate is high. It is recommended to optimize the packaging process, such as adjusting the tension of the packaging material or improving the design of the packaging machine to improve packaging efficiency and reduce consumption; the motor temperature is within the normal range and no adjustment is required.

[0049] At the same time, the gold foil processing production control system also takes into account the actual conditions of the production environment. It is recommended to adjust the temperature and humidity of the production environment to the target values ​​and increase the investment in air purification equipment to improve the air cleanliness to level 100.

[0050] 4. Working condition adjustment and judgment information

[0051] The gold foil processing production control system generates the following working condition adjustment judgment information based on the analysis results of the control decision data set.

[0052] Cutting machine: Adjust the cutting speed to 14m / s; monitor the motor temperature and vibration amplitude to ensure they are within the normal range.

[0053] Grinding machine: Adjust the grinding parameters to the best state, such as grinding pressure is XN (the specific value needs to be determined based on experiments), and replace more wear-resistant abrasives. Monitor the motor temperature and vibration amplitude to ensure that they are within the normal range.

[0054] Coating machine: adjust coating parameters, such as increasing coating time by Y seconds (the specific value needs to be determined based on experiments) or adjusting coating materials. Monitor the vacuum degree to ensure it is within the normal range.

[0055] Packaging machine: Optimize the packaging process, such as adjusting the tension of the packaging material to ZN / m (the specific value needs to be determined based on experiments) or improving the design of the packaging machine. Monitor the motor temperature to ensure it is within the normal range.

[0056] Production environment: Adjust the temperature of the production environment to 25°C and the humidity to 50%; increase the investment in air purification equipment and improve the air cleanliness to Class 100.

[0057] Furthermore, after implementing the above-mentioned working condition adjustment judgment information, the gold foil processing production control system once again monitored the equipment status and production environment data: the cutting speed of the cutting machine was increased to 14m / s, and the motor temperature and vibration amplitude remained within the normal range; the grinding speed of the grinder was increased to the optimized level, the abrasive consumption rate was reduced to the normal level, and the motor temperature and vibration amplitude remained within the normal range; the coating speed of the coating machine was increased to the optimized level, the coating thickness reached the target value, and the vacuum degree remained within the normal range; the packaging speed of the packaging machine was increased to the optimized level, the packaging material consumption rate was reduced to the normal level, and the motor temperature remained within the normal range; the temperature, humidity and air cleanliness of the production environment were all adjusted to the target values.

[0058] In the present invention, the gold foil processing production control system realizes intelligent management of the production process through a highly integrated data processing and analysis mechanism based on step 120. The core of this step is that the gold foil processing production control system can generate a comprehensive control decision data set based on the collected equipment status monitoring data and production environment sampling data to guide the efficient operation of the target processing production line.

[0059] In detail, equipment status monitoring data refers to the operating parameters collected in real time from various equipment on the production line, such as temperature, pressure, speed, etc. These data reflect the immediate working status of the equipment; production environment sampling data covers various indicators of the environment around the production line, such as humidity, temperature, air quality, etc., which have a direct or indirect impact on product quality. The feature recognition weight is a set of coefficients used by the gold foil processing production control system to analyze and evaluate the importance of these data. Different weights represent the emphasis of the gold foil processing production control system on different factors when making decisions.

[0060] In the process of constructing the control decision data set, the gold foil processing production control system first calculates the equipment operation state vector of the target processing production line under the first feature recognition weight. This vector is the result of weighted processing of all equipment status monitoring data, and it comprehensively reflects the operation state of the entire production line under the current weight. For example, if the first feature recognition weight focuses on the stability and efficiency of the equipment, then the operation state vector will reflect more parameter information that affects stability and efficiency.

[0061] Next, the gold foil processing production control system will further consider the impact of the production environment sampling data, and combine the equipment operation state vector of the target processing production line under the second feature recognition weight to generate a linkage production line description vector. The "linkage" applied in the present invention means that the gold foil processing production control system not only focuses on the state of a single device or a single weight, but also conducts a comprehensive analysis of the equipment state with environmental factors and states under different weights, so as to obtain a more comprehensive and accurate description of the production line. For example, the second feature recognition weight may focus on product quality and energy consumption. The gold foil processing production control system will combine the equipment state and environmental data under this weight to analyze their specific impact on product quality and energy consumption.

[0062] Finally, the gold foil processing production control system will repeat the above process and generate linkage production line description vectors for the target processing production line under these weights for X feature recognition weights (X is a positive integer representing the number of different decision angles considered by the gold foil processing production control system). These vectors together constitute the control decision data set, providing a scientific basis for real-time adjustment and optimization of the production line.

[0063] It can be seen that step 120 generates a comprehensive and accurate control decision data set by comprehensively using equipment status monitoring data and production environment sampling data, combined with multiple feature recognition weights for comprehensive analysis. The control decision data set not only reflects the current status of the production line, but also predicts possible future change trends, providing strong support for the efficient and stable operation of gold foil processing and production.

[0064] In other possible application scenarios, the gold foil processing production control system first collects a set of equipment status monitoring data and production environment sampling data.

[0065] Equipment status monitoring data (taking a key equipment as an example): temperature: 75°C; pressure: 100Psi; speed: 1500RPM.

[0066] Production environment sampling data: humidity: 50%; temperature: 25°C; air quality index (AQI): 50 (excellent).

[0067] Next, the gold foil processing production control system performs an analysis based on the preset feature recognition weights. There are three feature recognition weights set, focusing on stability, efficiency and product quality respectively.

[0068] The first feature recognition weight (stability): The gold foil processing production control system performs weighted processing on the equipment status monitoring data to obtain the equipment operation state vector. For example, if the weights of temperature, pressure and speed in terms of stability are 0.3, 0.4 and 0.3 respectively, an exemplary expression of the operation state vector can be: Stability operation state vector = 0.3×75+0.4×100+0.3×1500=497.5 (this invention application is only an example, and the actual calculation may be more complicated). Note that 497.5 in this invention application is a comprehensive score used to represent the stability state of the equipment under the current weight.

[0069] Second feature identification weight (efficiency): Similarly, the gold foil processing production control system weights the equipment status monitoring data according to the weight of efficiency to obtain another equipment operation state vector. At the same time, the gold foil processing production control system also considers the impact of production environment sampling data on efficiency. For example, if the weights of humidity and temperature on efficiency are 0.2 and 0.1 respectively, these factors will also be included in the calculation.

[0070] Generation of linkage production line description vector: For each feature recognition weight, the gold foil processing production control system combines the equipment status monitoring data, the production environment sampling data, and the equipment operation state vector under the previous weight (if any) to generate a linkage production line description vector. For example, under the second feature recognition weight (efficiency), the gold foil processing production control system may consider the operation state vector under the first weight (stability) as one of the inputs, combined with the equipment status and environmental data of the current weight, to generate a more comprehensive efficiency description vector.

[0071] Repeat the process to generate linkage line description vectors under X feature recognition weights: The gold foil processing production control system will repeat the above process until the corresponding linkage line description vectors are generated for all X feature recognition weights. Ultimately, these linkage line description vectors together constitute the control decision data set, providing a specific numerical basis for the real-time adjustment and optimization of the gold foil processing production line.

[0072] In some other examples, step 120 involves three important feature vectors: an equipment operation state vector of the target processing production line under a first feature recognition weight, a linkage production line description vector of the target processing production line under X feature recognition weights, and an equipment operation state vector of the target processing production line under a second feature recognition weight.

[0073] 1) The equipment operation state vector of the target processing production line under the first feature recognition weight. Under the first feature recognition weight (stability), the equipment operation state vector can be expressed as: stability_operation state_vector=[0.3*75, 0.4*100, 0.3*1500]=[22.5, 40, 450]; this vector includes the influence of three dimensions of temperature, pressure and speed on equipment stability. The value of each dimension is the product of the original data and the weight of the dimension.

[0074] 2) The equipment operation state vector of the target processing production line under the second feature recognition weight. Under the second feature recognition weight (efficiency), the operation state vector considering only the equipment state can be expressed as: efficiency_operation state_vector_equipment=[0.2*75, 0.5*100, 0.3*1500]=[15, 50, 450]; considering the impact of the production environment sampling data, the environmental data can also be included in the vector. Assuming that the impact of environmental data on efficiency has been normalized and represented in an additional dimension, the efficiency operation state vector (combining equipment status and environmental data) can be expressed as: efficiency_operation state_vector=[15, 50, 450, 20] (where 20 is the set value of the impact of environmental data on efficiency)

[0075] It should be noted that since environmental data usually includes multiple factors, in practical applications, those skilled in the art assign a dimension to each environmental factor and perform weighted processing according to its weight. In the above example, the influence of all environmental factors can be combined into a single dimension.

[0076] 3) The linkage line description vector of the target processing line under X feature recognition weights. For the linkage line description vector under X feature recognition weights, each feature recognition weight will generate a corresponding equipment operation state vector similar to the above. These vectors will together form the basis of the control decision data set. In other words, for each feature recognition weight, the system will generate a feature vector that includes the impact of equipment status and environmental data.

[0077] For example, under the third feature recognition weight (product quality), the linkage production line description vector may contain the operating state vectors under the first two weights as input (or some combination or transformation of these vectors), as well as the equipment status and environmental data under the current weight. The dimension of this vector will depend on the number of features actually selected and how to combine these features. Ultimately, these multi-dimensional linear feature vectors will together constitute a control decision data set, providing a specific numerical basis for the real-time adjustment and optimization of the gold foil processing production line.

[0078] The present invention application constructs a comprehensive control decision data set by comprehensively acquiring the equipment status monitoring data and production environment sampling data of the target processing production line. This data set not only contains the equipment operation state vector of the target processing production line under different feature recognition weights, but also incorporates the influence of environmental factors, thereby achieving an accurate characterization of the production line working conditions. On this basis, through in-depth analysis of the control decision data set, the present invention application can obtain working condition adjustment discrimination information, and then guide the precise control and processing of the intelligent robot on the target processing production line. In this way, the intelligence level of the production line is significantly improved, the pertinence and effectiveness of the working condition adjustment are enhanced, and the production efficiency decline and product quality problems caused by equipment status or environmental changes are effectively avoided, ensuring the stability and efficiency of the production process.

[0079] In some preferred embodiments, the control decision data set of the target processing production line is obtained based on the equipment status monitoring data and the production environment sampling data, including: integrating the equipment operation state vector of the target processing production line under the second feature recognition weight and the production environment sampling data to obtain equipment environment cross-monitoring data; associating the equipment operation state vector of the target processing production line under the second feature recognition weight with the equipment environment cross-monitoring data to obtain the linkage production line description vector of the target processing production line under the second feature recognition weight; obtaining the control decision data set of the target processing production line based on the equipment operation state vector of the target processing production line under the first feature recognition weight and the linkage production line description vector of the target processing production line under the second feature recognition weight.

[0080] In the field of gold foil processing and production, precise control of the equipment status and production environment of the production line is the key to ensuring product quality and production efficiency. As a high-precision, high-value-added material, the processing of gold foil has extremely high requirements for the stability of the equipment and the cleanliness of the environment. In order to achieve this goal, the gold foil processing production control system needs to generate a control decision data set based on the equipment status monitoring data and the production environment sampling data to guide the intelligent adjustment of the production line.

[0081] In the gold foil processing and production control technology, a key function is to integrate the equipment status monitoring data and the production environment sampling data, and obtain the control decision data set through correlation analysis. The core of this function is to achieve precise control of the production line by deeply exploring the intrinsic connection between the equipment status and environmental factors. For example, the equipment status monitoring data reflects the operating status of each device in the production line, including the key parameters such as the switch status, operating speed, temperature, and pressure of the equipment. The changes in these parameters directly affect the processing quality and production efficiency of the gold foil. The production environment sampling data includes various factors in the production environment, such as temperature, humidity, dust concentration, etc., which also have an important impact on the processing of gold foil.

[0082] By integrating equipment status monitoring data and production environment sampling data, the gold foil processing production control system can obtain a comprehensive equipment environment cross-monitoring data. This data not only contains equipment status information, but also incorporates the impact of environmental factors on equipment performance. Then, through association analysis, the gold foil processing production control system can further reveal the interaction between equipment status and environmental factors, thereby obtaining a linkage production line description vector. This vector is a comprehensive description of the production line under the set feature recognition weight, which contains comprehensive information on equipment status and environmental factors.

[0083] Finally, based on the linkage production line description vector and the equipment operation state vector under the first feature recognition weight, the gold foil processing production control system can obtain a control decision data set. This data set is a comprehensive guide to the production line control strategy, including equipment adjustment plans and environmental control strategies under different feature recognition weights. By implementing these strategies, the gold foil processing production control system can achieve precise control of the production line and ensure the processing quality and production efficiency of the gold foil.

[0084] In some examples, the gold foil processing production line includes three key equipment: rolling mill, annealing furnace and cutting machine. At the same time, the key factors in the production environment include temperature, humidity and dust concentration. The goal of the gold foil processing production control system is to generate a control decision data set to guide the intelligent adjustment of the three equipment and the optimization of the production environment.

[0085] 1) Data Acquisition

[0086] Equipment status monitoring data: Real-time monitoring of the operating status of the rolling mill, annealing furnace and cutting machine through sensors. For example, under the first feature recognition weight (such as stability), the equipment operating state vector obtained is [R1, F1, C1], where R1, F1, and C1 represent the values ​​of the operating state parameters (such as speed, temperature, pressure, etc.) of the rolling mill, annealing furnace, and cutting machine, respectively.

[0087] Production environment sampling data: The temperature, humidity and dust concentration data in the production environment are collected in real time through environmental monitoring equipment. For example, the environmental sampling data obtained is [E_temp, E_humidity, E_dust], where E_temp, E_humidity and E_dust represent the values ​​of temperature, humidity and dust concentration respectively.

[0088] 2) Data Integration

[0089] Under the second feature recognition weight (such as efficiency), for example, the obtained equipment operation state vector is [R2, F2, C2].

[0090] The equipment operation state vector [R2, F2, C2] is integrated with the production environment sampling data [E_temp, E_humidity, E_dust]. In the present invention, the gold foil processing production control system can adopt an exemplary integration method, that is, combining the equipment state parameters with the environmental factor values ​​to obtain an equipment environment cross-monitoring data matrix.

[0091] 3) Data association and linkage description

[0092] Through association analysis, the equipment operation state vector [R2, F2, C2] is associated with the equipment environment cross-monitoring data. In the present invention, the gold foil processing production control system can use an exemplary linear association method, that is, for example, there is a linear relationship between the equipment state parameters and the environmental factors. For example, the operation state parameter R2 of the rolling mill can be linearly related to the temperature E_temp in the production environment, that is, R2_assoc=w1*R2+w2*E_temp, where w1 and w2 are weight coefficients that can be determined by experiments or experience.

[0093] Similarly, the linkage description parameters F2_assoc and C2_assoc of the annealing furnace and the cutting machine under the second feature recognition weight can be calculated. Finally, the linkage production line description vector is obtained as [R2_assoc, F2_assoc, C2_assoc].

[0094] 4) Decision making:

[0095] According to the equipment operation state vector [R1, F1, C1] under the first feature recognition weight and the linkage production line description vector [R2_assoc, F2_assoc, C2_assoc] under the second feature recognition weight, a control decision data set is generated. In the present invention application, the gold foil processing production control system can adopt a method that comprehensively considers the two feature recognition weights, that is, weighted averaging each equipment state parameter to obtain the final control strategy. For example, for a rolling mill, the final control strategy can be R_control=Q1*R1+Q2*R2_assoc, where Q1 and Q2 are weighted average coefficients, satisfying Q1+Q2=1.

[0096] Similarly, the control strategies F_control and C_control of the annealing furnace and cutting machine can be calculated. Finally, the control decision data set is [R_control, F_control, C_control].

[0097] In addition, at the algorithm level, this embodiment adopts an exemplary linear association method and a weighted average method. However, in practical applications, since the relationship between equipment status and environmental factors may be more complex, the gold foil processing production control system can use more advanced algorithms for association analysis and decision generation. For example, machine learning algorithms (such as support vector machines, neural networks, etc.) can be used to train and predict equipment status monitoring data and production environment sampling data. Through the training model, the inherent connection and interaction mechanism between equipment status and environmental factors can be learned. Then, the trained model is used to predict the new equipment status monitoring data and production environment sampling data to obtain a more accurate and effective control decision data set. In addition, data mining and statistical analysis methods can also be used to conduct in-depth analysis and mining of data. For example, cluster analysis, association rule mining and other methods can be used to reveal the potential relationship and law between equipment status and environmental factors. The application of these methods and algorithms can further improve the accuracy and efficiency of gold foil processing production control technology.

[0098] In this way, by integrating equipment status monitoring data and production environment sampling data, and performing correlation analysis and decision generation, the gold foil processing production control system can obtain the control decision data set of the gold foil processing production line. This data set is a comprehensive guide to the control strategy of the production line, including equipment adjustment plans and environmental control strategies under different feature recognition weights. By implementing these strategies, the gold foil processing production control system can achieve precise control of the production line to ensure the processing quality and production efficiency of the gold foil. At the same time, by adopting more advanced algorithms and methods for correlation analysis and decision generation, the gold foil processing production control system can further improve the accuracy and efficiency of the gold foil processing production control technology, and provide strong support for the intelligent and automated development of the gold foil processing industry.

[0099] In some further optional embodiments, the method further includes: cyclically embedding the linkage production line description vector of the target processing production line under the second feature recognition weight through Y feature embedding modules to obtain the linkage production line description vector of the target processing production line under Y feature recognition weights, where Y is a positive integer.

[0100] Furthermore, the control decision data set of the target processing production line is obtained based on the equipment operation state vector of the target processing production line under the first feature identification weight and the linkage production line description vector of the target processing production line under the second feature identification weight, including: integrating the equipment operation state vector of the target processing production line under the first feature identification weight, the linkage production line description vector of the target processing production line under the second feature identification weight, and the linkage production line description vectors of the target processing production line under Y feature identification weights to obtain the control decision data set of the target processing production line.

[0101] In actual application, realizing intelligent and refined control of production lines is the key to improving product quality, optimizing production efficiency and reducing costs. In order to achieve this goal, it is necessary not only to monitor the equipment operation status in real time, but also to deeply analyze the impact of the production environment on the processing process, and make accurate control decisions accordingly.

[0102] In the control of the gold foil processing production line, the multi-feature embedding technology aims to capture the complex dynamic behavior of the production line more comprehensively by deeply mining the linkage production line description vector under different feature recognition weights. The core of this function is that by cyclically embedding multiple feature modules, it can capture subtle changes in the production line in different dimensions, thereby providing a richer information basis for control decisions.

[0103] For example, multi-feature embedding technology involves the following key steps:

[0104] 1) Feature identification and weight assignment: First, it is necessary to identify the key features that affect the gold foil processing line and assign corresponding weights to these features. These features may include equipment operating status parameters (such as speed, temperature, pressure, etc.), production environment parameters (such as temperature, humidity, dust concentration, etc.) and process parameters (such as foil thickness, processing speed, etc.);

[0105] 2) Generation of linkage production line description vector: After identifying key features and assigning weights, a linkage production line description vector can be generated based on these features and weights. This vector is a comprehensive description of the production line under the set feature recognition weights, including comprehensive information on equipment status, environmental factors, and process parameters;

[0106] 3) Multi-feature embedding: Next, the linkage production line description vector is cyclically embedded through Y feature embedding modules. Each feature embedding module focuses on a set feature or feature combination, and generates a linkage production line description vector under Y feature recognition weights by deeply exploring the intrinsic connections and interactions between these features;

[0107] 4) Integration and decision generation: Finally, the equipment operation state vector under the first feature recognition weight, the linkage production line description vector under the second feature recognition weight, and the linkage production line description vector under Y feature recognition weights are integrated. This integration process may involve steps such as data fusion, feature selection, and weight adjustment, and finally a control decision data set is obtained. This data set contains equipment adjustment plans, environmental control strategies, and process parameter optimization suggestions under different feature recognition weights, providing comprehensive guidance for the intelligent control of the production line.

[0108] For example, the gold foil processing production line includes three key equipment: rolling mill, annealing furnace and cutting machine. At the same time, three key features are identified: equipment operation status (feature 1), production environment (feature 2) and process parameters (feature 3). The goal is to generate a control decision data set to guide the intelligent adjustment of these three equipment, the optimization of the production environment and the setting of process parameters.

[0109] 1) Feature identification and weight allocation

[0110] Feature 1 (equipment operation status): including parameters such as the speed of the rolling mill, the temperature of the annealing furnace and the pressure of the cutting machine; Feature 2 (production environment): including parameters such as the temperature, humidity and dust concentration of the production environment; Feature 3 (process parameters): including parameters such as the thickness of the foil and the processing speed. Assign a weight to each feature, set the weights to w1, w2 and w3 respectively, and satisfy w1+w2+w3=1.

[0111] 2) Linked production line description vector generation

[0112] Under the second feature recognition weight (such as production environment), the linkage production line description vector is set to [E_temp_assoc, E_humidity_assoc, E_dust_assoc], where E_temp_assoc, E_humidity_assoc, and E_dust_assoc respectively represent the degree of influence of temperature, humidity, and dust concentration in the production environment on the operating status of the equipment.

[0113] 3) Multi-feature embedding

[0114] There are Y=3 feature embedding modules, corresponding to equipment operation status, production environment and process parameters respectively.

[0115] Through the first feature embedding module, the characteristics of the equipment operation status are embedded into the linkage production line description vector, and the linkage production line description vector [R_assoc1, F_assoc1, C_assoc1] under the equipment operation status feature recognition weight is obtained.

[0116] Through the second feature embedding module (i.e., the linkage production line description vector obtained in the previous step), the characteristics of the production environment are further embedded to obtain the linkage production line description vector [R_assoc2, F_assoc2, C_assoc2] under the production environment feature recognition weight. Note that the present invention application [R_assoc2, F_assoc2, C_assoc2] is obtained by further embedding the equipment operation status characteristics on the basis of [E_temp_assoc, E_humidity_assoc, E_dust_assoc].

[0117] Through the third feature embedding module, the features of the process parameters are embedded to obtain the linkage production line description vector [R_assoc3, F_assoc3, C_assoc3] under the process parameter feature recognition weight.

[0118] 4) Integration and decision making

[0119] The equipment operation state vector [R1, F1, C1] under the first feature identification weight, the linkage production line description vector [E_temp_assoc, E_humidity_assoc, E_dust_assoc] under the second feature identification weight, and the linkage production line description vectors [R_assoc1, F_assoc1, C_assoc1], [R_assoc2, F_assoc2, C_assoc2], [R_assoc3, F_assoc3, C_assoc3] under Y=3 feature identification weights are integrated.

[0120] The integration process can adopt weighted average, data fusion or feature selection methods. For example, for the control decision of the rolling mill, the weighted average method can be adopted, that is, R_control=P1*R1+P2*R_assoc1+P3*R_assoc2+P4*R_assoc3, where P1, P2, P3 and P4 are weighted average coefficients, satisfying P1+P2+P3+P4=1.

[0121] Similarly, the control decisions F_control and C_control of the annealing furnace and cutting machine can be calculated.

[0122] Finally, the control decision data set is obtained as [R_control, F_control, C_control], which contains equipment adjustment plans, environmental control strategies and process parameter optimization suggestions under different feature recognition weights.

[0123] In the process of multi-feature embedding, a variety of algorithms can be used to explore the intrinsic connections and interactions between features. For example, machine learning algorithms (such as support vector machines, decision trees, etc.) can be used to train models and learn the association rules between different features. Through the trained model, new equipment status monitoring data, production environment sampling data, and process parameter data can be predicted to obtain the linkage production line description vector under Y feature recognition weights.

[0124] In addition, deep learning algorithms (such as neural networks) can also be used for feature embedding and integration. Neural networks have powerful feature extraction and pattern recognition capabilities and can automatically learn complex structures and rules in data. By building a suitable neural network model, features such as equipment status, environmental factors, and process parameters can be embedded and integrated to obtain a more accurate control decision data set.

[0125] It should be noted that in the process of embedding and integrating multiple features, problems such as high data dimension, feature redundancy or noise interference may be encountered. To solve these problems, methods such as feature selection, dimensionality reduction or denoising can be used to optimize the data set. For example, methods such as principal component analysis (PCA) or sparse representation can be used to reduce data dimension and remove redundant features; methods such as filters or wavelet transforms can be used to remove noise interference.

[0126] In this way, through multi-feature embedding technology, the complex dynamic behavior of the gold foil processing production line can be captured more comprehensively, and a richer information basis can be provided for control decisions. By identifying key features, generating linkage production line description vectors, and performing multi-feature embedding and integration analysis, equipment adjustment plans, environmental control strategies, and process parameter optimization suggestions under different feature recognition weights can be obtained. These suggestions can provide comprehensive guidance for the intelligent control of the production line, thereby improving product quality, optimizing production efficiency, and reducing costs. At the same time, by using advanced algorithms and methods for feature embedding and integration analysis, the accuracy and efficiency of gold foil processing production control technology can be further improved, providing strong support for the intelligent and automated development of the gold foil processing industry.

[0127] In some alternative embodiments, the equipment operation state vector of the target processing production line under the second feature recognition weight and the production environment sampling data are integrated to obtain equipment environment cross-monitoring data, including: processing the equipment operation state vector of the target processing production line under the second feature recognition weight and the production environment sampling data through a scene feature focusing strategy to obtain first feature focusing information; processing the equipment operation state vector of the target processing production line under the second feature recognition weight and the production environment sampling data through a task feature focusing strategy to obtain second feature focusing information; processing the union of the first feature focusing information and the second feature focusing information through a knowledge feature cross branch to obtain equipment environment cross-monitoring data.

[0128] In the fine manufacturing industry of gold foil processing, it is crucial to ensure the stable operation and high-quality output of the production line. In order to achieve this goal, it is necessary not only to closely monitor the operating status of the equipment, but also to pay attention to the impact of the production environment on the processing process. The generation of equipment environment cross-monitoring data in this embodiment aims to provide a more comprehensive information basis for the intelligent control of the production line by deeply exploring the intrinsic connection between the equipment operating status and the production environment. The core of this function is that through the scene feature focus strategy, task feature focus strategy and knowledge feature cross-branching, the key features of the equipment operating status and production environment under the set scenes and tasks can be captured, and cross-monitoring data can be generated accordingly to provide guidance for the optimization and adjustment of the production line.

[0129] Specifically, the generation of equipment environment cross-monitoring data involves the following key steps:

[0130] Scenario feature focusing strategy: First, it is necessary to identify key features that have a significant impact on the equipment operating status and production environment based on the specific scenario of the gold foil processing production line. These features may include operating status parameters such as the equipment's vibration frequency, temperature, and humidity, as well as environmental factors such as dust concentration and temperature fluctuations in the production environment. Through the scenario feature focusing strategy, these key features can be extracted and enhanced to obtain the first feature focusing information;

[0131] Task feature focus strategy: In addition to scene features, the specific task requirements of the gold foil processing production line also need to be considered. Different processing tasks may have different requirements for the equipment operation status and production environment. Therefore, it is necessary to further process the equipment operation status vector and production environment sampling data according to the task requirements, extract the features closely related to the task, and obtain the second feature focus information;

[0132] Knowledge feature cross branch: After obtaining the first feature focus information and the second feature focus information, the two information need to be integrated and processed through the knowledge feature cross branch. The knowledge feature cross branch can use machine learning, deep learning and other algorithms to mine and analyze the feature focus information to find out the intrinsic connection and interaction between the equipment operation status and the production environment. Through this step, the equipment environment cross monitoring data can be obtained, which contains comprehensive information about the equipment operation status and production environment under the set scenarios and tasks.

[0133] For example, the gold foil processing production line includes three key equipment: rolling mill, annealing furnace and cutting machine. The goal is to generate equipment environment cross-monitoring data to guide the intelligent adjustment of these three equipment and the optimization of the production environment.

[0134] Scene feature focus strategy: The vibration frequency and temperature of the rolling mill are set to be the key features. Through the scene feature focus strategy, these two features can be extracted and enhanced. For example, a filter can be used to bandpass filter the vibration frequency to remove noise interference; the temperature data can be smoothed to eliminate outliers. For the production environment, dust concentration and temperature fluctuation are set as key features. Similarly, these features can be extracted and enhanced using corresponding processing methods.

[0135] Task feature focus strategy: The current processing task is set to produce gold foil with a thickness of 0.01 mm. According to this task requirement, the equipment operation state vector and the production environment sampling data need to be further processed. For example, the speed of the rolling mill can be adjusted to ensure that the thickness of the foil meets the requirements; at the same time, the temperature of the annealing furnace needs to be precisely controlled to ensure the annealing quality of the foil. In this step, machine learning algorithms (such as support vector machines, decision trees, etc.) can be used to train the model to learn the equipment operation status and production environment characteristics under different tasks. Through the trained model, new data can be predicted and classified to obtain features closely related to the task.

[0136] Knowledge feature cross branch: After obtaining the first feature focus information and the second feature focus information, the two information need to be integrated and processed through the knowledge feature cross branch. This step can be achieved using deep learning algorithms (such as neural networks). A neural network model can be constructed, with the first feature focus information and the second feature focus information as input and the equipment environment cross monitoring data as output. By training the neural network model, the intrinsic connection and interaction between the equipment operation status and the production environment can be learned. For example, model structures such as multi-layer perceptron (MLP) or convolutional neural network (CNN) can be used for feature cross and integration. During the training process, optimization methods such as back propagation algorithm and gradient descent can be used to adjust the model parameters to improve the performance and accuracy of the model. Finally, through the processing of knowledge feature cross branches, the equipment environment cross monitoring data can be obtained. This data contains comprehensive information about the equipment operation status and production environment under set scenarios and tasks, which can provide comprehensive guidance for the intelligent control of the production line.

[0137] In this way, through the scene feature focus strategy, task feature focus strategy and knowledge feature cross-branch processing, equipment environment cross-monitoring data can be generated to provide a more comprehensive information basis for the intelligent control of the gold foil processing production line. This method can not only improve the stability and efficiency of the production line, but also ensure the quality and consistency of the product.

[0138] Under some other optional design ideas, the equipment operation state vector of the target processing production line under the second feature recognition weight is associated with the equipment environment cross-monitoring data to obtain the linkage production line description vector of the target processing production line under the second feature recognition weight, including: performing residual processing on the equipment environment cross-monitoring data through the target residual subnet to obtain the residual connection vector of the equipment environment cross-monitoring data; bit-by-bit summing the residual connection vector of the equipment environment cross-monitoring data with the equipment operation state vector of the target processing production line under the second feature recognition weight to obtain the linkage production line description vector of the target processing production line under the second feature recognition weight.

[0139] It can be understood that in the control of the gold foil processing production line, the generation of the linkage production line description vector aims to provide more comprehensive information support for the intelligent control of the production line by deeply exploring the intrinsic connection between the equipment operation status and the production environment. Specifically, the process involves associating the equipment environment cross-monitoring data (which has integrated the equipment operation status and production environment information) with the equipment operation status vector under the set feature recognition weight, so as to obtain a linkage description vector that can fully reflect the production line status.

[0140] The core of this function lies in two key steps: first, residual processing is performed on the equipment environment cross-monitoring data through the target residual subnet to extract key information from the data; second, the processed residual connection vector is summed bit by bit with the equipment operation status vector to generate a linkage production line description vector.

[0141] Continuing with the example of a gold foil processing line that includes three key equipment, a rolling mill, an annealing furnace, and a cutting machine, the goal is to generate a linkage line description vector under the second feature recognition weight (which is set to focus on the vibration and temperature characteristics of the equipment).

[0142] Acquisition of equipment environment cross-monitoring data: First, it is necessary to collect equipment operation status data and production environment data. For example, for a rolling mill, you may focus on its vibration frequency and temperature; for a production environment, you may focus on temperature, humidity, and dust concentration. Through the set sensors and monitoring systems, you can obtain the real-time values ​​of these data. For example, the vibration frequency of a rolling mill may be 50Hz and the temperature may be 80℃; the temperature of the production environment may be 25℃, the humidity may be 50%, and the dust concentration may be 0.1mg / m³.

[0143] Residual processing of the target residual subnet: The target residual subnet is a neural network model used to process the equipment environment cross-monitoring data to extract key information from the data. The purpose of residual processing is to remove redundancy and noise in the data and retain information that has an important impact on the state of the production line. The target residual subnet used is an exemplary fully connected neural network, whose input is the equipment environment cross-monitoring data (such as the above-mentioned rolling mill vibration frequency, temperature, and the temperature, humidity and dust concentration of the production environment), and the output is a residual connection vector. Through the trained target residual subnet, the equipment environment cross-monitoring data can be input into it to obtain a residual connection vector. For example, the residual connection vector may be [0.1, -0.05, 0.02, -0.01, 0.005] (the present invention application is only an example, and the actual value will depend on the specific data and model).

[0144] Generating the linkage production line description vector by bitwise summing: After obtaining the residual connection vector, it is necessary to sum it bitwise with the operation state vector of the equipment under the second feature recognition weight. Set the equipment operation state vector to [1.0, 0.8, 0.6, 0.4, 0.2] (this invention application is only an example, and the actual value will depend on the specific data and feature recognition weight).

[0145] By summing each bit, we can get the linkage production line description vector. The specific calculation process is: [1.0+0.1, 0.8+(-0.05), 0.6+0.02, 0.4+(-0.01), 0.2+0.005]=[1.1, 0.75, 0.62, 0.39, 0.205].

[0146] Through the above process, the linkage production line description vector of the target processing production line under the second feature recognition weight is obtained. This vector not only contains the operating status information of the equipment under the set weight, but also integrates the impact of the production environment on the equipment status. Through the generation of this vector, the actual status of the production line can be more comprehensively understood, providing stronger support for subsequent intelligent regulation. In addition, by continuously optimizing the model structure and parameters, the accuracy and reliability of the linkage production line description vector can be improved, thereby providing a more solid foundation for the intelligent and automated development of the gold foil processing production line.

[0147] In some exemplary embodiments, the method further includes: splitting the control decision vector in the control decision data set into decision sub-vectors of multiple environmental categories; optimizing the control decision vector in the control decision data set using the decision sub-vectors of the multiple environmental categories to obtain an optimized control decision data set.

[0148] In the next step, the control decision vector in the control decision data set is split into decision sub-vectors of multiple environmental categories, including: feature mapping the equipment operation state vector of the target processing production line under the first feature recognition weight and the linkage production line description vector of the target processing production line under X feature recognition weights, respectively, to obtain X+1 key decision sub-vectors and X+1 edge decision sub-vectors; focusing the X+1 key decision sub-vectors and the X+1 edge decision sub-vectors respectively through feature focusing branches, to obtain X+1 target key decision sub-vectors and X+1 target edge decision sub-vectors.

[0149] In this embodiment, the optimization of the control decision data set aims to improve the overall performance and stability of the production line by subdividing the production environment categories and formulating more accurate control strategies for different environment categories. Specifically, the process involves splitting the control decision vector into decision sub-vectors of multiple environment categories, and using these sub-vectors to optimize the original control decision vector to obtain a control decision data set that is more suitable for different production environment conditions.

[0150] The core of this function lies in two key steps: first, converting the equipment operation state vector and the linkage production line description vector into key decision sub-vectors and edge decision sub-vectors through feature mapping; second, focusing these decision sub-vectors through feature focusing branches to obtain target key decision sub-vectors and target edge decision sub-vectors.

[0151] Continuing with the example of the gold foil processing production line, which includes three key equipment: rolling mill, annealing furnace and cutting machine, and the production environment can be divided into three categories: high temperature, normal temperature and low temperature. The goal is to optimize the control decision data set to adapt to the production needs under different temperature environments.

[0152] Splitting of control decision vectors: First, it is necessary to collect the operating state vector of the equipment under the first feature recognition weight and the linkage production line description vector of the target processing production line under X feature recognition weights. Set X=3, that is, three different feature recognition weights are considered. Through feature mapping, these vectors can be converted into key decision sub-vectors and edge decision sub-vectors. Specifically, for each vector under the feature recognition weight, a key decision sub-vector and an edge decision sub-vector can be obtained. Therefore, a total of 4 (X+1) key decision sub-vectors and 4 (X+1) edge decision sub-vectors will be obtained.

[0153] Specific implementation of feature mapping: Feature mapping can be implemented through linear transformation or nonlinear transformation. Linear transformation is assumed to be used, that is, the original vector is mapped to a new space through a matrix. This matrix can be determined based on actual production data and requirements.

[0154] For example, for the operating state vector [v1, v2, v3] of the device under the first feature recognition weight, it can be mapped to the key decision sub-vector [k1, k2, k3] and the edge decision sub-vector [e1, e2, e3] through the matrix A=[a11, a12, a13; a21, a22, a23; a31, a32, a33]. The specific mapping process is: [k1, k2, k3]=A*[v1, v2, v3], [e1, e2, e3]=B*[v1, v2, v3], where B is another matrix.

[0155] Processing of feature focusing branch: After obtaining the key decision sub-vector and the marginal decision sub-vector, they need to be focused through the feature focusing branch. The purpose of focusing is to highlight the most important features in the vector for control decision and suppress other unimportant features.

[0156] An exemplary focusing method is assumed to be used, that is, weighted summation is performed on the elements in each vector, and the weight is determined according to the importance of the element. For example, for the key decision sub-vector [k1, k2, k3], its focused value can be calculated as w1k1+w2k2+w3*k3, where w1, w2, w3 are weights. Through the focusing process, 4 target key decision sub-vectors and 4 target marginal decision sub-vectors can be obtained.

[0157] Optimization of control decision data set: After obtaining the target key decision sub-vector and the target marginal decision sub-vector, they can be used to optimize the original control decision vector. Specifically, these sub-vectors can be weighted summed with the original control decision vector to obtain the optimized control decision data set.

[0158] Assuming the original control decision vector is [c1, c2, c3, ..., cn], the optimized control decision data set can be calculated as:

[0159] [c1_new, c2_new, c3_new, ..., cn_new]

[0160] =[w_k1k1_target+w_e1e1_target+...,

[0161] w_knkn_target+w_enen_target], where w_ki and w_ei are the weights of the key decision sub-vector and the edge decision sub-vector respectively.

[0162] Through the above technical solution, an optimized control decision data set is obtained. The control decision data set not only considers the operating status of the equipment under different feature recognition weights, but also integrates the impact of the production environment on the control decision. Through this optimization, the gold foil processing production line can be controlled more accurately, improving product quality and production efficiency and reducing costs.

[0163] In a further embodiment, the step of performing feature mapping on the equipment operation state vector of the target processing production line under the first feature recognition weight and the linkage production line description vector of the target processing production line under X feature recognition weights to obtain X+1 key decision sub-vectors and X+1 edge decision sub-vectors comprises: performing residual processing on the equipment operation state vector of the target processing production line under the first feature recognition weight through the first mapping list and the second mapping list respectively to obtain the key residual vector and the edge residual vector of the target processing production line under the first feature recognition weight; performing feature mapping on the key residual vector of the target processing production line under the first feature recognition weight The residual vector and the edge residual vector are respectively average pooled to obtain the first key decision sub-vector and the first edge decision sub-vector; the linkage production line description vector of the target processing production line under the u+1th feature recognition weight is residually processed through the first mapping list and the second mapping list respectively to obtain the key residual vector and the edge residual vector of the target processing production line under the u+1th feature recognition weight, where u is a positive integer not greater than X; the key residual vector and the edge residual vector of the target processing production line under the u+1th feature recognition weight are respectively average pooled to obtain the u+1th key decision sub-vector and the u+1th edge decision sub-vector.

[0164] Based on this, the feature focusing branch may include a fully connected core, a scene feature processing core, and a task feature processing core; the step of performing focusing processing on the X+1 key decision subvectors and the X+1 edge decision subvectors through the feature focusing branch to obtain X+1 target key decision subvectors and X+1 target edge decision subvectors includes: performing processing on the vth key decision subvector and the vth edge decision subvector through the fully connected core to obtain feature focusing information of the vth key decision subvector and feature focusing information of the vth edge decision subvector. The feature focusing information of the v-th key decision sub-vector and the feature focusing information of the v-th edge decision sub-vector are checked by the task feature processing, and optimization based on the task level is performed respectively to obtain the v-th key decision implicit weight and the v-th edge decision implicit weight; the v-th key decision implicit weight and the v-th edge decision implicit weight are checked by the scene feature processing, and optimization based on the scene level is performed respectively to obtain the v-th target key decision sub-vector and the v-th target edge decision sub-vector.

[0165] In the control of the gold foil processing production line, the purpose of feature mapping and focusing is to convert the equipment operation state vector and the linkage production line description vector into key decision sub-vectors and edge decision sub-vectors, and to obtain more accurate control decisions by further optimizing these sub-vectors. This process involves two key steps: first, obtaining key decision sub-vectors and edge decision sub-vectors through feature mapping; second, optimizing these sub-vectors through focusing.

[0166] In order to illustrate the above process more specifically, the following will give specific numerical examples and related algorithm analysis in combination with the actual scenario of the gold foil processing production line.

[0167] 1) Feature Mapping

[0168] Residual processing: Set the equipment operation state vector of the target processing production line under the first feature recognition weight as [v1, v2, v3], and there are two mapping lists: the first mapping list is [a1, a2, a3], and the second mapping list is [b1, b2, b3]. Perform residual processing on the equipment operation state vector through the first mapping list, and obtain the key residual vector as [a1v1, a2v2, a3v3]; perform residual processing through the second mapping list, and obtain the marginal residual vector as [b1v1, b2v2, b3v3].

[0169] Average pooling: Average pooling is performed on the key residual vector and the edge residual vector, that is, the average value of each element is calculated to obtain the first key decision sub-vector and the first edge decision sub-vector. For example, the key decision sub-vector is [(a1v1+a2v2+a3v3) / 3], and the edge decision sub-vector is [(b1v1+b2v2+b3v3) / 3] (for the sake of simplicity, this invention is assumed to have only one element, but the actual situation may be a vector).

[0170] Processing of the linkage production line description vector: For the linkage production line description vector of the target processing production line under the u+1th feature recognition weight, residual processing is also performed through the first mapping list and the second mapping list, and average pooling is performed to obtain the u+1th key decision sub-vector and the u+1th edge decision sub-vector.

[0171] 2) Focusing

[0172] Fully connected kernel processing: Set the vth key decision subvector to [kv1, kv2] and the vth edge decision subvector to [ev1, ev2]. These two subvectors are processed by the fully connected kernel to obtain their feature focus information. This process can be achieved through linear transformation or nonlinear transformation, such as calculation through matrix multiplication or neural network layer.

[0173] Task feature processing kernel optimization: The obtained feature focus information is optimized at the task level. The purpose of this step is to adjust the feature focus information according to the specific production task requirements to obtain the decision implicit weights that better meet the task requirements. For example, the feature focus information can be processed by a task feature processing kernel (which can be a simple linear transformation or a more complex neural network model) to obtain the vth key decision implicit weight and the vth edge decision implicit weight.

[0174] Optimization of scenario feature processing kernel: optimize the obtained decision implicit weights based on the scenario level. The purpose of this step is to consider different scenarios of the production environment (such as high temperature, normal temperature, low temperature, etc.) and further adjust the decision implicit weights to obtain control decisions that are more suitable for different scenarios. For example, the decision implicit weights can be processed by a scenario feature processing kernel (which can also be a simple linear transformation or a more complex neural network model) to obtain the vth target key decision sub-vector and the vth target marginal decision sub-vector.

[0175] Through the above feature mapping and focusing process, optimized key decision sub-vectors and edge decision sub-vectors are obtained. These sub-vectors not only consider the operating status of the equipment under different feature recognition weights, but also integrate the impact of the production environment on the control decision, and are optimized based on the task and scenario levels.

[0176] In some other preferred embodiments, the decision sub-vectors of the multiple environmental categories include X + 1 target key decision sub-vectors and X + 1 target edge decision sub-vectors; the step of optimizing the control decision vectors in the control decision data set by using the decision sub-vectors of the multiple environmental categories to obtain an optimized control decision data set includes: connecting the X + 1 target key decision sub-vectors to obtain a key linkage production line vector relationship network, and connecting the X + 1 target edge decision sub-vectors to obtain an edge linkage production line vector relationship network; integrating, by a residual subnet, the key linkage production line vector relationship network and the equipment operation state vector of the target processing production line under the first feature recognition weight to obtain a first feature integration result; integrating, by the residual subnet, the edge linkage production line vector relationship network and the linkage production line description vector of the target processing production line under the Xth feature recognition weight to obtain a second feature integration result.

[0177] Immediately afterwards, the step of connecting the X + 1 target key decision sub-vectors to obtain a key linkage production line vector relationship network includes: connecting the wth input information and the (X - w + 1)th target key decision sub-vector through a cyclic connection subnet to obtain the wth connection vector set; where w is a positive integer not greater than X; when w = 1, the wth input information is the (X + 1)th target key decision sub-vector; when 1 < w < X + 1, the wth input information is the (w - 1)th connection vector set.

[0178] During the gold foil processing, the control decisions of the production line are affected by various environmental factors, such as equipment status, process parameters, environmental conditions, etc. To effectively manage these complex factors, the present invention application introduces decision sub-vectors of multiple environmental categories, including X + 1 target key decision sub-vectors and X + 1 target edge decision sub-vectors. These decision sub-vectors are obtained by performing feature mapping and focusing processing on the operation status and linkage production line description of the production line under different feature recognition weights, and they contain rich production environment information and decision-making basis.

[0179] By connecting these target key decision sub-vectors and target edge decision sub-vectors, a key linkage production line vector relationship network and an edge linkage production line vector relationship network can be constructed respectively. These two relationship networks actually comprehensively depict and represent the decision-making space of the production line under different feature recognition weights, providing a solid foundation for subsequent optimization work.

[0180] To more specifically illustrate the above process, the following will give specific numerical examples and related algorithm analyses in combination with the actual scenario of the gold foil processing production line.

[0181] Connection of decision sub-vectors: Assuming X=3 feature recognition weights, we will get 4 target key decision sub-vectors (K1, K2, K3, K4) and 4 target edge decision sub-vectors (E1, E2, E3, E4).

[0182] The step of connecting these decision sub-vectors to form a relationship network is achieved by cyclically connecting the sub-network. Taking the key linkage production line vector relationship network as an example, when w=1, the first input information is K4, which is connected with the third target key decision sub-vector K3 to obtain the first connection vector set C1; when w=2, the second input information is C1, which is connected with the second target key decision sub-vector K2 to obtain the second connection vector set C2; when w=3, the third input information is C2, which is connected with the first target key decision sub-vector K1 to obtain the third connection vector set C3. Finally, C3 is the representation of the key linkage production line vector relationship network.

[0183] Integration of residual subnetworks: The residual subnetwork in the present invention application plays the role of integrating the relationship network with the specific equipment operation state vector or the linkage production line description vector. Taking the first feature integration result as an example, the key linkage production line vector relationship network C3 is integrated with the equipment operation state vector D1 of the target processing production line under the first feature recognition weight. This integration process can be achieved by residual learning, that is, learning the residual between C3 and D1, and adding the residual to D1 to obtain the first feature integration result F1. Similarly, the second feature integration result F2 can be obtained, that is, the edge linkage production line vector relationship network (obtained by connecting in a similar manner) is integrated with the linkage production line description vector D3 of the target processing production line under the third feature recognition weight.

[0184] Optimize the control decision data set: The first feature integration result F1 and the second feature integration result F2 obtained through the above steps can optimize the original control decision data set. Specifically, F1 and F2 can be input into the control decision model as new features, and the optimized control decision data set can be obtained through model learning and reasoning. This optimized data set not only considers the operating status and linkage relationship of the production line under different feature recognition weights, but also realizes the effective integration and fusion of these complex relationships through the residual subnet.

[0185] Designed in this way, it is an effective method to optimize the control decision data set using decision subvectors of multiple environmental categories. It not only improves the accuracy and reliability of control decisions, but also enhances the adaptability and robustness of the production line to different environmental conditions.

[0186] In some independent embodiments, the cyclic connection subnetwork includes an activation network based on a sliding operator; the w-th input information and the X-w+1-th target key decision subvector are connected through the cyclic connection subnetwork to obtain the w-th connection vector set, including: matching the w-th input information with the pooling result through the activation network based on the sliding operator to obtain a matching result; the pooling result is obtained by average pooling the X-w+1-th target key decision subvector; adjusting the window step variable and the window size variable according to the matching result to obtain a window step adjustment variable and a window size adjustment variable; combining the X-w+1-th target key decision subvector with the first feature derivative result to obtain a matrix operation result; the first feature derivative result is obtained by feature deriving the window size adjustment variable; the matrix operation result and the second feature derivative result are summed bit by bit to obtain the w-th connection vector set; the second feature derivative result is obtained by feature deriving the window step adjustment variable.

[0187] It can be understood that the main task of the recurrent connection subnet, as a key component in the gold foil processing production control system, is to process and integrate input information from different time steps or feature recognition weights to generate a connection vector set for subsequent decision-making. In these embodiments, the recurrent connection subnet particularly adopts an activation network based on a sliding operator. This design aims to more flexibly capture and process key features in the input information by dynamically adjusting the window step and window size, thereby improving the accuracy and response speed of production control.

[0188] In order to understand the process more intuitively, let’s take a specific scenario as an example: assume that the gold foil processing production line needs to process decision sub-vectors under X=5 feature recognition weights, and focus on the connection process between the w-th (w is a positive integer not greater than X) input information and the X-w+1th target key decision sub-vector.

[0189] Activation network matching based on sliding operator: Set the w-th input information as Iw, and the X-w+1th target key decision subvector as Kw (X=5 in the present invention, so the value range of w is 1 to 5). Perform average pooling on Kw to obtain the pooling result Pw. Average pooling is a common dimensionality reduction operation that reduces the dimension of the data by averaging the elements in the vector while retaining key information. Input Iw and Pw into the activation network based on the sliding operator. The network matches the two through a sliding operator (an operator that can slide on the input data and calculate local features) to generate a matching result Mr. This matching process can be regarded as a feature alignment or similarity measure, which aims to ensure the effective association between the input information and the pooling result.

[0190] Adjustment of window step and window size: According to the matching result Mr, dynamically adjust the window step variable (step) and window size variable (size). This process can be achieved by setting certain thresholds or rules. For example, when Mr indicates that the input information is highly matched with the pooling result, the window step can be reduced to increase the precision of feature extraction, and the window size can be appropriately adjusted to cover a wider range of features. The adjusted window step and window size are recorded as step_adj and size_adj respectively.

[0191] Feature derivation and matrix operation: Perform feature derivation on the adjusted window size size_adj to obtain the first feature derivation result FD1. Feature derivation is a method of generating new features by transforming or expanding the original features, which helps to enhance the model's ability to represent input data. Combine the X-w+1th target key decision subvector Kw with FD1 for matrix operation (such as multiplication or addition) to obtain the matrix operation result MO. This step aims to effectively fuse the adjusted window size information with the target key decision subvector.

[0192] Bitwise summation and generation of connection vector sets: Perform feature derivation on the adjusted window step size step_adj to obtain the second feature derivation result FD2. Perform bitwise summation (i.e., add corresponding elements) of the matrix operation result MO and FD2 to obtain the wth connection vector set Cv_w. This step further integrates the feature information after the window step size and window size are adjusted, providing a more comprehensive and accurate basis for subsequent production control decisions.

[0193] It can be seen that the activation network based on the sliding operator plays a vital role in the recurrent connection subnetwork. It not only flexibly captures and processes the key features in the input information by dynamically adjusting the window step and window size, but also generates a connection vector set for subsequent production control decisions through a series of complex feature derivation and matrix operation operations.

[0194] In other independent embodiments, the operating condition adjustment analysis is performed on the target processing production line based on the control decision data set to obtain the operating condition adjustment judgment information of the target processing production line, including: performing judgment processing on the control decision data set through a judgment module to obtain the operating condition adjustment judgment information of the target processing production line; wherein the operating condition adjustment judgment information includes the operating condition adjustment judgment information of each production line link in the target processing production line.

[0195] It is worth mentioning that working condition adjustment analysis is a key link in gold foil processing production control technology. It can accurately judge the current working condition of each production line link in the target processing production line and whether adjustment is needed by conducting in-depth analysis of the control decision data set. This process not only helps to discover and solve potential production problems in a timely manner, but also ensures the stable operation of the production line and the continuous improvement of product quality.

[0196] In order to more intuitively understand the implementation process of working condition adjustment analysis, a specific scenario is set: the gold foil processing production line includes three production line links, namely raw material preparation, processing and forming, and surface treatment. Each production line link has a series of control parameters and working condition indicators, such as temperature, pressure, speed, etc.

[0197] Construction of control decision data set: First, it is necessary to collect historical data of the target processing production line under normal operation, including control parameters and working condition indicators of each production line link. By preprocessing and extracting features from these historical data, a control decision data set containing multiple feature vectors can be constructed. Each feature vector corresponds to the working condition of the production line at a certain moment.

[0198] Design of the discrimination module: The discrimination module is the core component of the working condition adjustment analysis. Its main task is to discriminate the control decision data set to obtain the working condition adjustment discrimination information of each production line link. When designing the discrimination module, machine learning algorithms such as support vector machine (SVM), decision tree or random forest can be used. These algorithms can learn and construct a discrimination model based on the input feature vector, which is used to classify and discriminate new working condition data.

[0199] Generation of working condition adjustment discrimination information: Input the constructed control decision data set into the discrimination module. The discrimination module processes each feature vector one by one, and judges whether the current working condition of each production line link is normal based on the learned discrimination model. If the working condition of a production line link is judged to be abnormal, the discrimination module will output the corresponding working condition adjustment discrimination information, including the control parameters that need to be adjusted and the recommended adjustment direction.

[0200] For example, in the raw material preparation stage, 100 sets of historical data were collected, each set of data contained three characteristics: temperature, pressure and speed.

[0201] Through preprocessing and feature extraction, a 100*3 control decision dataset was obtained.

[0202] The support vector machine algorithm is used to build a discriminant model and discriminate new working condition data.

[0203] The results showed that the temperature characteristics of 5 sets of data exceeded the normal range. The discrimination module outputted the corresponding working condition adjustment discrimination information and recommended lowering the temperature to ensure the stability of the raw materials and the quality of subsequent processing.

[0204] Algorithm optimization and expansion: In practical applications, the discriminant module can also be optimized and expanded. For example, deep learning algorithms such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs) can be introduced to improve the accuracy and generalization ability of the discriminant model.

[0205] At the same time, the discriminant model can be updated and iterated online in combination with real-time monitoring data and feedback mechanisms to better adapt to the dynamic changes of the production line.

[0206] In summary, adjusting the operating conditions of the target processing production line based on the control decision data set can not only accurately determine the operating status of each production line link, but also provide timely adjustment suggestions to ensure the stable operation of the production line and continuous improvement of product quality.

[0207] In some other independent embodiments, the acquisition of equipment status monitoring data and production environment sampling data of the target processing production line includes: performing feature embedding on the target processing production line through a first feature embedding module to obtain an equipment operation state vector of the target processing production line under a first feature recognition weight; performing feature embedding on the equipment operation state vector of the target processing production line under a first feature recognition weight through a second feature embedding module to obtain an equipment operation state vector of the target processing production line under a second feature recognition weight; using the equipment operation state vector of the target processing production line under the first feature recognition weight and the equipment operation state vector of the target processing production line under the second feature recognition weight as the equipment status monitoring data of the target processing production line; performing environmental signal feature mining on the target processing production line through a sensor signal attention module to obtain environmental signal features of the target processing production line; and performing feature denoising on the environmental signal features of the target processing production line based on environmental denoising indication to obtain production environment sampling data of the target processing production line.

[0208] It can be understood that data acquisition is the basic link of gold foil processing production control technology, which is directly related to the accuracy of subsequent data analysis, working condition adjustment and decision making. This embodiment uses advanced technologies such as feature embedding and sensor signal attention to achieve comprehensive and accurate monitoring of equipment status and production environment, providing strong support for the continuous optimization of the production line.

[0209] 1) Acquisition of equipment status monitoring data

[0210] First feature embedding module: First, the target processing production line is initially extracted through the first feature embedding module. The production line is assumed to consist of n key devices, each of which has m monitorable parameters (such as temperature, pressure, speed, etc.). The first feature embedding module performs weighted summation or more complex nonlinear transformation on these parameters according to the preset first feature recognition weight to obtain the operating state vector of each device under the first feature recognition weight. For example, for a certain device, its operating state vector may be an m-dimensional vector, in which each element represents the operating state of the device at the corresponding parameter.

[0211] Second feature embedding module: Then, the device operation state vector under the first feature recognition weight is input into the second feature embedding module for deeper feature extraction. The second feature embedding module will use different feature recognition weights to further mine the implicit information in the device operation state and obtain the device operation state vector under the second feature recognition weight. This process can be regarded as a refined description of the device state, which helps to more accurately reflect the actual operation state of the device.

[0212] Formation of equipment status monitoring data: Finally, the equipment operation status vectors under the two feature recognition weights are combined to form the equipment status monitoring data of the target processing production line. This set of data not only contains the basic operation status information of the equipment, but also contains deeper feature associations and operation status change trends.

[0213] 2) Acquisition of production environment sampling data

[0214] Sensor signal attention module: In order to obtain sampled data of the production environment, the sensor signal attention module is first used to perform feature mining on the environmental signals around the production line. This module can automatically focus on and extract key signals that have a greater impact on the environment, such as temperature, humidity, noise, etc., and ignore irrelevant or noisy signals. This process is similar to the human attention mechanism and can significantly improve the efficiency and accuracy of environmental signal processing.

[0215] Environmental denoising indication: Since there are often various noise interferences in the production environment, in order to obtain accurate production environment sampling data, it is necessary to denoise the mined environmental signal features. This embodiment introduces an environmental denoising indication, which can guide the selection and parameter setting of the denoising algorithm based on the characteristics of the environmental signal and the noise distribution. For example, for high-frequency noise, a filtering algorithm can be used to remove it; for low-frequency noise or periodic interference, more complex signal processing techniques can be used to suppress it.

[0216] Formation of production environment sampling data: After feature mining and denoising, the production environment sampling data of the target processing line is a set of values ​​that can accurately reflect the actual conditions of the production environment. These data provide an important basis for subsequent environmental analysis, working condition adjustment and production optimization.

[0217] For example, the target processing production line consists of three key devices, each of which has five monitorable parameters. The first feature embedding module uses a linear weighted summation method to obtain the operating state vector (5-dimensional) of each device under the first feature recognition weight. The second feature embedding module uses a nonlinear transformation (such as a neural network) to further process the operating state vector under the first feature recognition weight to obtain the operating state vector under the second feature recognition weight (also 5-dimensional). Finally, the operating state vectors under the two feature recognition weights are spliced ​​to form a 10-dimensional equipment status monitoring data vector.

[0218] In the process of acquiring production environment sampling data, the sensor signal attention module first collects environmental signal data around the production line through the sensor network. Then, the attention mechanism is used to process these data and extract key signals that have a greater impact on the environment (such as temperature, humidity, etc.). Next, according to the environmental denoising indication, a suitable denoising algorithm (such as a filtering algorithm) is selected to denoise the extracted key signals. Finally, the denoised signal data is sorted and analyzed to obtain the production environment sampling data.

[0219] In summary, this embodiment achieves comprehensive and accurate monitoring of the equipment status and production environment of the gold foil processing production line through technical ideas such as feature embedding and sensor signal attention. The implementation of this process not only improves the accuracy and efficiency of data acquisition, but also provides strong support for subsequent data analysis, working condition adjustment and decision making.

[0220] Further, Figure 2 The structure diagram of a gold foil processing and production control system 200 provided by the present invention is shown in FIG. Figure 2 The gold foil processing production control system 200 shown includes a processor 210, which can call and run a computer program from a memory to implement the method in the present invention application.

[0221] Alternatively, if Figure 2 As shown, the gold foil processing and production control system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method in the present invention.

[0222] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .

[0223] Alternatively, if Figure 2 As shown, the gold foil processing and production control system 200 may further include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices, specifically, may send information or data to other devices, or receive information or data sent by other devices.

[0224] Optionally, the gold foil processing production control system 200 can implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as processing modules) or devices deployed with storage engines in various methods of the present invention, which will not be described here for the sake of brevity.

[0225] It should be understood that the processor of the present invention may be an integrated circuit chip having signal processing capabilities.

[0226] It is understood that the memory in the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memory. It should be noted that the memory of the systems and methods described herein is intended to include but is not limited to suitable types of memory.

[0227] Based on the above, a readable storage medium is provided, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the above method are implemented.

[0228] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises one..." does not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the method and device in the embodiment of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0229] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention application in essence or the part that contributes to the prior art can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), including a number of instructions for a terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention application.

[0230] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the inspiration of the present invention, ordinary technicians in this field can also make many forms without departing from the purpose of the present invention and the scope of protection of the present invention, all of which are within the protection of the present invention.

Claims

1. A gold foil processing and production control method based on artificial intelligence, characterized in that: The method is implemented by a gold foil processing production control system, and the method comprises: Acquire equipment status monitoring data and production environment sampling data of the target processing production line; the equipment status monitoring data includes an equipment operation state vector of the target processing production line under a first feature recognition weight and an equipment operation state vector of the target processing production line under a second feature recognition weight; A control decision data set of the target processing production line is obtained according to the equipment state monitoring data and the production environment sampling data; the control decision data set includes an equipment operation state vector of the target processing production line under a first feature recognition weight and a linkage production line description vector of the target processing production line under X feature recognition weights, and the linkage production line description vector of the target processing production line under any feature recognition weight is obtained according to the equipment operation state vector of the target processing production line under a second feature recognition weight and the production environment sampling data, where X is a positive integer; According to the control decision data set, the working condition adjustment analysis of the target processing production line is performed to obtain working condition adjustment discrimination information of the target processing production line; the working condition adjustment discrimination information is used to instruct at least one intelligent robot of the target processing production line to perform control processing; The control decision data set of the target processing production line is obtained based on the equipment status monitoring data and the production environment sampling data, including: Integrate the equipment operation state vector of the target processing production line under the second feature recognition weight and the production environment sampling data to obtain equipment environment cross-monitoring data; Associating the equipment operation state vector of the target processing production line under the second feature recognition weight with the equipment environment cross-monitoring data to obtain a linkage production line description vector of the target processing production line under the second feature recognition weight; Obtaining a control decision data set of the target processing production line according to the equipment operation state vector of the target processing production line under the first feature recognition weight and the linkage production line description vector of the target processing production line under the second feature recognition weight; The method further includes: cyclically embedding the linkage production line description vector of the target processing production line under the second feature recognition weight through Y feature embedding modules to obtain the linkage production line description vector of the target processing production line under Y feature recognition weights, where Y is a positive integer; The step of obtaining a control decision data set of the target processing production line according to the equipment operation state vector of the target processing production line under the first feature recognition weight and the linkage production line description vector of the target processing production line under the second feature recognition weight comprises: integrating the equipment operation state vector of the target processing production line under the first feature recognition weight, the linkage production line description vector of the target processing production line under the second feature recognition weight, and the linkage production line description vectors of the target processing production line under Y feature recognition weights to obtain the control decision data set of the target processing production line; The device operation state vector of the target processing production line under the second feature recognition weight and the production environment sampling data are integrated to obtain the device environment cross monitoring data, including: The equipment operation state vector of the target processing production line under the second feature recognition weight and the production environment sampling data are processed by a scene feature focusing strategy to obtain first feature focusing information; Processing the equipment operation state vector of the target processing production line under the second feature recognition weight and the production environment sampling data through a task feature focusing strategy to obtain second feature focusing information; Processing the union of the first feature focus information and the second feature focus information through a knowledge feature cross branch to obtain device environment cross monitoring data; The step of associating the equipment operation state vector of the target processing production line under the second feature recognition weight with the equipment environment cross-monitoring data to obtain the linkage production line description vector of the target processing production line under the second feature recognition weight includes: Performing residual processing on the device environment cross-monitoring data through a target residual subnet to obtain a residual connection vector of the device environment cross-monitoring data; The residual connection vector of the equipment environment cross-monitoring data is summed bit by bit with the equipment operation state vector of the target processing production line under the second feature recognition weight to obtain the linkage production line description vector of the target processing production line under the second feature recognition weight.

2. The method according to claim 1, characterized in that The method further comprises: Splitting the control decision vector in the control decision data set into decision sub-vectors of multiple environmental categories; The control decision vectors in the control decision data set are optimized using the decision sub-vectors of the multiple environment categories to obtain an optimized control decision data set.

3. The method according to claim 2, characterized in that The step of splitting the control decision vector in the control decision data set into decision sub-vectors of multiple environmental categories includes: Feature mapping is performed on the equipment operation state vector of the target processing production line under the first feature recognition weight and the linkage production line description vector of the target processing production line under X feature recognition weights to obtain X+1 key decision sub-vectors and X+1 edge decision sub-vectors; The X+1 key decision sub-vectors and the X+1 edge decision sub-vectors are respectively focused through the feature focusing branch to obtain X+1 target key decision sub-vectors and X+1 target edge decision sub-vectors.

4. The method according to claim 3, characterized in that The step of performing feature mapping on the equipment operation state vector of the target processing production line under the first feature recognition weight and the linkage production line description vector of the target processing production line under X feature recognition weights to obtain X+1 key decision sub-vectors and X+1 edge decision sub-vectors comprises: Residual processing is performed on the equipment operation state vector of the target processing production line under the first feature recognition weight through the first mapping list and the second mapping list respectively to obtain the key residual vector and the edge residual vector of the target processing production line under the first feature recognition weight; Performing average pooling on the key residual vector and the edge residual vector of the target processing production line under the first feature recognition weight, respectively, to obtain the first key decision sub-vector and the first edge decision sub-vector; Residual processing is performed on the linkage production line description vector of the target processing production line under the u+1th feature recognition weight through the first mapping list and the second mapping list respectively, to obtain the key residual vector and the edge residual vector of the target processing production line under the u+1th feature recognition weight, where u is a positive integer not greater than X; Average pooling is performed on the key residual vector and the edge residual vector of the target processing production line under the u+1th feature recognition weight to obtain the u+1th key decision sub-vector and the u+1th edge decision sub-vector; The feature focusing branch includes a fully connected core, a scene feature processing core, and a task feature processing core; the step of performing focusing processing on the X+1 key decision sub-vectors and the X+1 edge decision sub-vectors respectively through the feature focusing branch to obtain X+1 target key decision sub-vectors and X+1 target edge decision sub-vectors includes: The vth key decision sub-vector and the vth edge decision sub-vector are checked by the full connection, and are processed respectively to obtain feature focusing information of the vth key decision sub-vector and feature focusing information of the vth edge decision sub-vector, where v is a positive integer not greater than X+1; The feature focus information of the v-th key decision sub-vector and the feature focus information of the v-th edge decision sub-vector are checked by the task feature processing, and optimization based on the task level is performed respectively to obtain the v-th key decision implicit weight and the v-th edge decision implicit weight; The vth key decision implicit weight and the vth edge decision implicit weight are checked by the scene feature processing, and optimization based on the scene level is performed respectively to obtain the vth target key decision sub-vector and the vth target edge decision sub-vector.

5. The method according to claim 2, characterized in that The decision sub-vectors of the multiple environmental categories include X+1 target key decision sub-vectors and X+1 target edge decision sub-vectors; the step of optimizing the control decision vectors in the control decision data set by using the decision sub-vectors of the multiple environmental categories to obtain the optimized control decision data set includes: Connecting the X+1 target key decision sub-vectors to obtain a key linkage production line vector relationship network, and connecting the X+1 target edge decision sub-vectors to obtain an edge linkage production line vector relationship network; Integrate the key linkage production line vector relationship network and the equipment operation state vector of the target processing production line under the first feature recognition weight through the residual subnet to obtain a first feature integration result; Integrate the edge linkage production line vector relationship network and the linkage production line description vector of the target processing production line under the Xth feature recognition weight through the residual subnet to obtain a second feature integration result; The step of connecting the X+1 target key decision sub-vectors to obtain a key linkage production line vector relationship network includes: Connect the w-th input information and the (X - w + 1)-th target key decision sub-vector through a cyclic connection subnet to obtain the w-th connection vector set; where w is a positive integer not greater than X; when w = 1, the w-th input information is the (X + 1)-th target key decision sub-vector; when 1 < w < X + 1, the w-th input information is the (w - 1)-th connection vector set.

6. A gold foil processing and production control system, characterized in that: Comprising at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, such that the at least one processor executes the method according to any one of claims 1-5.

Images