Dual-purpose wrench production whole-process intelligent management method and system under industrial internet of things architecture

Abstract

The application discloses a two-purpose wrench production full-process intelligent management method and system under an industrial internet of things architecture, and relates to the technical field of industrial internet of things and intelligent manufacturing. The method comprises the following steps: taking the use characteristics of a target two-purpose wrench as a constraint, obtaining single-function production sample data and full-process historical production data; constructing and training a plurality of function performance evaluation bodies based on the data to form an evaluation body set; constructing a bidirectional master-slave optimization channel comprising a first master-slave optimization channel, a second master-slave optimization channel and an information exchange channel according to the evaluation body set; performing bidirectional parallel optimization iteration in combination with the full-process historical production data and the bidirectional master-slave optimization channel, and outputting an optimal process scheme; and finally applying the optimal scheme to a full production environment to realize intelligent management. The application effectively solves the coupling conflict problem of dual-function end production process parameters through a bidirectional optimization architecture, realizes global collaborative optimization, improves product quality and production efficiency, and reduces production cost and the waste product rate.

Description

Technical Field

[0001] This invention relates to the field of industrial Internet of Things (IoT) and intelligent manufacturing technology, specifically to an intelligent management method and system for the entire production process of dual-purpose wrenches under an industrial IoT architecture. Background Technology

[0002] Modern manufacturing widely adopts the Industrial Internet of Things (IIoT) architecture, which connects production equipment and sensors to achieve data collection and process monitoring, thereby improving the level of production visualization to a certain extent.

[0003] However, in the production process of dual-purpose wrenches under the Industrial Internet of Things (IIoT) architecture, existing technologies still face significant challenges: First, there is a significant process coupling conflict between the dual-function ends, such as the open end and the swivel end. Optimizing the performance parameters of one end often leads to a decrease in the performance of the other end. For example, increasing the heat treatment parameters for the hardness of the open end will damage the toughness or coating adhesion of the swivel end. Existing single-objective or simple weighted multi-objective optimization methods are difficult to effectively handle this deep conflict based on physicochemical mechanisms. Second, there is a data gap between macroscopic process parameters such as temperature and pressure and microscopic quality data such as metallographic structure and grain size. There is a lack of effective cross-scale fusion analysis methods, which makes it impossible to establish an accurate predictive model from process parameters to microscopic performance and then to macroscopic performance, leaving quality control in a lagging and passive state. In addition, traditional multi-objective optimization algorithms are computationally expensive and prone to getting trapped in local optima when dealing with such high-dimensional, nonlinear, and strongly constrained engineering problems, making it difficult to meet the requirements of actual production for optimization efficiency and effectiveness. Summary of the Invention

[0004] This invention addresses the technical problems in existing technologies, such as the difficulty in coordinating the coupling conflicts of dual-function processes, insufficient fusion of cross-scale production data, and low efficiency of multi-objective optimization. It provides an intelligent management method and system for the entire production process of dual-purpose wrenches under an industrial Internet of Things (IoT) architecture.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] In a first aspect, this invention provides an intelligent management method for the entire production process of dual-purpose wrenches under an industrial Internet of Things (IoT) architecture, including:

[0007] Constrained by the application characteristics of the target dual-purpose wrench, single-function production sample data and full-process historical production data are obtained;

[0008] Based on the single-function production sample data and the full-process historical production data, multiple functional performance evaluation bodies are constructed and trained to generate an evaluation body set.

[0009] A bidirectional master-slave optimization channel is constructed based on the functional performance evaluation set, wherein the bidirectional master-slave optimization channel includes a first master-slave optimization channel, a second master-slave optimization channel, and an information exchange channel;

[0010] The entire process of historical production data is combined with the bidirectional master-slave optimization channel to perform bidirectional parallel optimization iteration, and the optimal decision variable is output as the optimal process scheme;

[0011] The optimal process scheme is applied to the entire production environment of the target dual-purpose wrench for full-process production management.

[0012] Secondly, this invention provides an intelligent management system for the entire production process of dual-purpose wrenches under an industrial Internet of Things (IoT) architecture, including:

[0013] The data acquisition module is used to acquire single-function production sample data and full-process historical production data, constrained by the usage characteristics of the target dual-purpose wrench.

[0014] The evaluation body construction and training module is used to construct and train multiple functional performance evaluation bodies based on the single-function production sample data and the full-process historical production data, and generate an evaluation body set.

[0015] The optimized channel construction module is used to construct a bidirectional master-slave optimized channel according to the functional performance evaluation set, wherein the bidirectional master-slave optimized channel includes a first master-slave optimized channel, a second master-slave optimized channel and an information exchange channel;

[0016] The optimization iteration module is used to combine the historical production data of the entire process with the bidirectional master-slave optimization channel to perform bidirectional parallel optimization iteration, and output the optimal decision variable as the optimal process scheme.

[0017] The production management execution module is used to apply the optimal process scheme to the entire production environment of the target dual-purpose wrench for full-process production management.

[0018] The beneficial effects of this invention are:

[0019] Compared to existing technologies, this invention firstly solves the coupling conflict problem of production process parameters on both ends by constructing a bidirectional master-slave optimization channel and a migration mapping model, achieving true global collaborative optimization and ensuring that the performance of both ends simultaneously reaches the optimal standard. Secondly, by combining a parallel computing framework with a lightweight proxy model, optimization efficiency is significantly improved, computational costs are reduced, and a better Pareto solution can be found quickly. Thirdly, by establishing a cross-scale data fusion mechanism, accurate prediction from process parameters to microscopic performance and then to macroscopic performance is achieved, transforming quality control from passive response to proactive prevention. Finally, by directly distributing the optimal process plan to the production execution system, intelligent management of the entire production process is realized, significantly improving product consistency and reducing scrap rate and production costs. Attached Figure Description

[0020] Figure 1 A flowchart illustrating the intelligent management method for the entire production process of dual-purpose wrenches under the industrial Internet of Things architecture provided by this invention;

[0021] Figure 2 This is a schematic diagram of the intelligent management system for the entire production process of dual-purpose wrenches under the industrial Internet of Things architecture provided by the present invention.

[0022] In the attached diagram, the components represented by each number are as follows:

[0023] Data acquisition module 11, evaluation body construction and training module 12, optimization channel construction module 13, optimization iteration module 14, production management execution module 15. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0026] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0027] Example 1, as Figure 1 As shown, this embodiment of the invention provides an intelligent management method for the entire production process of dual-purpose wrenches under an industrial Internet of Things (IoT) architecture, including:

[0028] S10: Using the application characteristics of the target dual-purpose wrench as constraints, obtain single-function production sample data and full-process historical production data;

[0029] Specifically, constrained by the intended use characteristics of the target dual-purpose wrench, single-function production sample data and full-process historical production data are obtained, including:

[0030] The first and second functional ends of the target dual-purpose wrench are defined as functional constraints, and the raw material specifications of the target dual-purpose wrench are defined as basic constraints.

[0031] Based on the functional constraints and the basic constraints, the first single-function production sample data corresponding to the first functional end and the second single-function production sample data corresponding to the second functional end are obtained respectively.

[0032] Obtain the full-process historical production data of the target dual-purpose wrench, including the coupled production process of the first functional end and the second functional end;

[0033] Both the single-function production sample data and the full-process historical production data include process parameter sequences and corresponding performance test data.

[0034] A dual-purpose wrench is a specialized tool with two different functional ends, such as an open-end wrench and a box wrench, or a metric wrench and an imperial wrench. During production, both functional ends must meet specified performance standards, and these wrenches are typically manufactured from the same raw material through continuous processing. Therefore, it is essential to first clarify the constraints in the production process of the target dual-purpose wrench. Specifically, functional constraints refer to the performance requirements for the first and second functional ends of the wrench, such as the hardness of the open-end end and the toughness of the box wrench end. Basic constraints specify the raw material specifications that must be used in the production process, including material grade, chemical composition, and mechanical properties.

[0035] Furthermore, based on clearly defined constraints, single-function production sample data corresponding to the first and second functional ends of the wrench were collected separately. For the first functional end, first single-function production sample data was collected from historical production records focused on optimizing the first functional end; for the second functional end, second single-function production sample data was collected from historical production records focused on optimizing the second functional end.

[0036] Specific data acquisition is completed automatically and in real time through industrial IoT devices integrated into the production line. Sensors are used to capture physical parameters such as temperature, pressure, and torque; machine vision systems are used for visual recognition tasks such as appearance inspection and dimensional measurement; and RFID is used to track workpiece identity, bind data streams, and record flow time. The collected data comprehensively covers all key production factors affecting the performance of the wrench's functional end, including raw material parameters such as steel grade, chemical composition, and initial hardness; forging parameters such as initial forging temperature, final forging temperature, forging pressure, and deformation; machining parameters such as cutting speed, feed rate, torque, and depth of cut; heat treatment process parameters such as heating temperature, holding time, cooling medium type, and cooling rate; and surface treatment parameters such as electroplating current density, plating solution temperature, and coating thickness.

[0037] Furthermore, regarding performance testing data, the first single-function sample data mainly includes indicators such as dimensional accuracy and hardness, while the second single-function sample data mainly includes indicators such as torque capacity and coating thickness. By collecting the first and second single-function production sample data, a complete data chain from process input to performance output was established, laying a solid foundation for the subsequent construction of a high-precision performance prediction model.

[0038] Secondly, it is also necessary to acquire full-process historical production data. Full-process historical production data refers to a complete production record dataset from the input of raw materials, through all production stages such as forging, machining, heat treatment, and surface treatment, to the final product inspection. Specifically, full-process historical production data not only fully records the parameters and performance results of continuous processing at both functional ends of a single workpiece from its raw material stage through all processes to becoming a finished product, but also records in detail how the process parameters of each specific production stage simultaneously affect and record the status and performance data of the first and second functional ends.

[0039] Finally, it is important to note that both the collected single-function production sample data and the full-process historical production data must include complete sequences of process parameters and corresponding performance test data. The process parameter sequences cover all production stages from raw material processing to final forming, specifically including detailed records of key processes such as forging parameters, machining parameters, heat treatment parameters, and surface treatment parameters. Each process parameter sequence strictly corresponds to complete performance test data, which includes test results of various performance indicators from both functional ends, thus establishing a complete mapping relationship from process input to performance output.

[0040] In terms of data characteristics, the first and second single-function production sample data focus on establishing an idealized performance model for a single function, respectively focusing on the correspondence between process parameters and performance indicators of a single function. The full-process historical production data, on the other hand, truly reflects the performance coupling and conflict between the two functions during actual production, recording the comprehensive impact and trade-offs of process parameters on the two functions.

[0041] In summary, single-function data provides in-depth performance characteristics of a single function, while full-process data provides broad information on the interaction between two functions. Together, they form a complete data foundation for comprehensively analyzing the production process of dual-function tools, providing sufficient data support for subsequently establishing accurate performance prediction models and solving multi-objective optimization problems.

[0042] S20: Based on the single-function production sample data and the full-process historical production data, construct and train multiple functional performance evaluation bodies to generate an evaluation body set;

[0043] Based on the single-function production sample data and the full-process historical production data, multiple functional performance evaluation bodies are constructed and trained respectively, generating an evaluation body set, including:

[0044] Based on the first single-function production sample data, a first single-function performance evaluation body is constructed, which is used to predict the performance of the first function.

[0045] Based on the second single-function production sample data, a second single-function performance evaluation body is constructed, which is used to predict the performance of the second function.

[0046] Create copies of the first single-function performance evaluation body and the second single-function performance evaluation body respectively, and perform enhancement training on the first copy evaluation body and the second copy evaluation body based on the full-process historical production data to obtain the third single-function performance evaluation body and the fourth single-function performance evaluation body.

[0047] The first single-function performance evaluation body, the second single-function performance evaluation body, the third single-function performance evaluation body, and the fourth single-function performance evaluation body are combined to form the evaluation body set.

[0048] The first single-function performance evaluation body is used to predict hardness performance attributes, and the second single-function performance evaluation body is used to predict coating adhesion performance attributes.

[0049] The functional performance evaluation system is a predictive model built based on machine learning algorithms. Based on the input production process parameter sequence, it can accurately predict the performance indicators of specific functions of a dual-purpose wrench. Specifically, based on the aforementioned single-function production sample data and historical production data throughout the entire process, multiple functional performance evaluation systems are constructed and trained, ultimately forming an evaluation system set. This enables a precise mapping from process parameters to performance indicators and provides a model foundation for subsequent bidirectional optimization. The specific construction and training steps are as follows:

[0050] First, a performance evaluation model for the first single function is constructed based on the production sample data of the first single function. This performance evaluation model uses machine learning methods to establish a mapping relationship from production process parameters to the performance indicators of the first function, specifically for predicting the performance of the first function, particularly the hardness property.

[0051] Secondly, a performance evaluation system for the second single-function component is constructed based on the production sample data of the second single-function component. This evaluation system also employs machine learning methods to establish a mapping relationship between production process parameters and performance indicators of the second function, specifically for predicting the performance of the second function, particularly for predicting coating adhesion properties.

[0052] For example, the first single-function performance evaluation body is constructed using a neural network model, whose core function is to establish a nonlinear mapping relationship from production process parameters to the hardness performance attribute of the first functional end. Because there is a highly nonlinear and complex correlation between production process parameters and hardness performance attributes, and neural networks have significant advantages in automatic feature extraction and fitting of complex nonlinear relationships, a neural network model is chosen to construct this evaluation body.

[0053] Specifically, the evaluation model mainly consists of an input preprocessing layer, a feature extraction layer, and an output layer. The input preprocessing layer receives raw data such as raw material parameters, forging parameters, machining parameters, and heat treatment process parameters, and eliminates dimensional differences through standardization. The feature extraction layer employs a multi-layer fully connected neural network structure, with the number of neurons adaptively configured according to the feature dimension. Each layer is equipped with a ReLU activation function to introduce non-linear transformation capabilities, and Dropout layers are embedded between layers with a dropout rate set between 0.2 and 0.5 to suppress overfitting and improve the model's generalization performance. The output layer uses a single neuron equipped with a linear activation function to map the final features to continuous hardness prediction values.

[0054] During training, key hyperparameters, including the learning rate (0.001), training epochs (100), and batch size (64), were set. The learning rate was chosen based on the stability considerations of the gradient descent algorithm, the number of training epochs ensured the model had sufficient iterations to learn the data features, and the batch size balanced training efficiency with memory consumption. Supervised learning was employed. Specifically, sample process parameter sequences were collected from the first single-function production sample data as the input sample set; simultaneously, hardness detection data measured under the corresponding parameters were acquired to form a set of sample hardness values ​​as the label sample set. The input sample set and the corresponding label sample set were divided into training, validation, and test sets in a 7:2:1 ratio. Then, the sample parameters in the training set were used as input features, with the corresponding sample hardness values ​​as supervision labels. The network weight parameters were iteratively optimized using the backpropagation algorithm and the Adam optimizer. The mean squared error loss function is used to measure the deviation between the predicted hardness and the actual hardness. The training process is monitored by a validation set. When the validation set loss no longer decreases for several consecutive rounds and the accuracy reaches 90%, the training is terminated, and the first single-function performance evaluation body after convergence is obtained. This body is used to capture the complex nonlinear relationship between process parameters and hardness performance, thereby achieving high-precision hardness prediction.

[0055] Meanwhile, the second single-function performance evaluation body also employs a neural network model. Its core function is to establish a nonlinear mapping relationship from production process parameters to the adhesion performance attributes of the coating at the second functional end. Due to the complex nonlinear correlation between process data such as surface treatment parameters and heat treatment parameters and coating adhesion performance attributes, the neural network model can effectively learn the deep relationship between these features and performance indicators.

[0056] The second single-function performance evaluation model structure includes an input preprocessing layer, a feature extraction layer, and an output layer. The input preprocessing layer receives process data such as surface treatment parameters and heat treatment parameters, and performs data preprocessing through standardization. The feature extraction layer adopts a multi-layer fully connected neural network structure, with each layer equipped with a ReLU activation function and a Dropout layer embedded between layers, with the dropout rate set between 0.2 and 0.5. The output layer outputs the predicted coating adhesion value through a linear activation function.

[0057] During training, the second single-function performance evaluation body uses the same hyperparameter settings as the first evaluation body. Specifically, sample process parameter sequences are extracted from the second single-function production sample data as the input sample set, and the corresponding coating adhesion detection data are used as the label sample set. The same training process is employed, optimizing network parameters through backpropagation with mean squared error as the loss function, and monitoring the training process on the validation set. Training terminates when the prediction accuracy reaches a predetermined requirement, such as 90%, ultimately yielding a second single-function performance evaluation body specifically designed for predicting coating adhesion performance, capable of accurately establishing the mapping relationship between process parameters and coating adhesion performance.

[0058] In summary, after being trained with historical data, the first and second single-function performance evaluation bodies can achieve high-precision prediction of the performance of their respective functions, providing a reliable predictive basis for subsequent collaborative optimization.

[0059] Furthermore, complete copies of the first and second single-function performance evaluation modules were created, and these copies were augmented using historical production data from the entire production process. Since the historical production data comprehensively records the process parameters and corresponding performance results of both functionalities within the same production process, it fully reflects the complex coupling relationship and mutual influence mechanism between the two functionalities. Therefore, by augmenting the evaluation modules with this coupling information on historical production data from the entire production process, the evaluation modules, which originally focused only on predicting the performance of a single functionality, further learned and mastered the synergistic and conflict effects between the two functionalities, thus training a third and fourth single-function performance evaluation module with the ability to understand dual-function coupling.

[0060] Finally, the first, second, third, and fourth single-function performance evaluation bodies are combined to form a complete evaluation body set. This evaluation body set includes a high-precision basic model that focuses on the performance prediction of a single function, as well as an enhanced model that deeply understands the coupling effect of dual functions. Together, they provide a comprehensive and reliable basis for prediction and evaluation for subsequent bidirectional master-slave optimization.

[0061] S30: Construct a bidirectional master-slave optimization channel according to the functional performance evaluation set, wherein the bidirectional master-slave optimization channel includes a first master-slave optimization channel, a second master-slave optimization channel and an information exchange channel;

[0062] First, a bidirectional master-slave optimization channel is constructed based on the aforementioned functional performance evaluation set. Prior to this, the following steps are also included:

[0063] Model distillation was performed on the third and fourth single-function performance evaluation bodies based on the knowledge distillation method.

[0064] Based on the model distillation results, update the third and fourth single-function performance evaluation bodies.

[0065] Before constructing a bidirectional master-slave optimization channel based on the functional performance evaluation set, knowledge distillation is required to optimize model performance. Knowledge distillation is a model compression and knowledge transfer technique that significantly reduces model complexity and computational cost while maintaining prediction accuracy by having a smaller student model learn the output distribution of a larger teacher model.

[0066] Specifically, the third single-function performance evaluation body is used as the teacher model, which has already gained the ability to understand the dual-function coupling effect through training on historical production data throughout the entire process. Using a knowledge distillation method, a more streamlined student model is trained, ensuring its output prediction distribution is consistent with the teacher model, thus obtaining the distilled third single-function performance evaluation body. Similarly, the same distillation process is performed on the fourth single-function performance evaluation body to obtain a structurally optimized fourth single-function performance evaluation body.

[0067] Subsequently, the third and fourth single-function performance evaluation bodies are updated based on the knowledge distillation results. The updated evaluation bodies, while maintaining the original prediction accuracy, have a smaller model size, faster inference speed, and higher computational efficiency, making them particularly suitable for embedding into bidirectional master-slave optimization channels that require real-time optimization.

[0068] Furthermore, a bidirectional master-slave optimization channel is constructed based on the functional performance evaluation set, wherein the bidirectional master-slave optimization channel includes a first master-slave optimization channel, a second master-slave optimization channel, and an information exchange channel, including:

[0069] The first optimization channel is constructed by taking the performance index of the first functional end as the main optimization objective, the process parameters of the first function as the first decision variable, and combining the first single-function performance evaluation body.

[0070] Using the performance indicators of the second functional end as the main optimization objective and the process parameters of the second function as the second decision variable, and combining the second single-function performance evaluation body, a second optimization channel is constructed.

[0071] The fourth single-function performance evaluation body is integrated into the first optimization channel to obtain the first master-slave optimization channel, wherein the fourth single-function performance evaluation body takes the first decision variable as input;

[0072] The third single-function performance evaluation body is integrated into the second optimization channel to obtain the second master-slave optimization channel, wherein the third single-function performance evaluation body takes the second decision variable as input;

[0073] Establish the information exchange channel and connect it to the first master-slave optimization channel and the second master-slave optimization channel.

[0074] First, taking the hardness performance attribute of the first functional end as the primary optimization objective and its related process parameters as the first decision variables, a first optimization channel is constructed by combining the predictive capability of the first single-function performance evaluation body. This first optimization channel aims to maximize the hardness performance of the open end. It utilizes the first single-function performance evaluation body to establish a quantitative mapping relationship between process parameters such as forging temperature, heat treatment holding time, and cooling rate, and hardness performance indicators. Through a Bayesian optimization algorithm, the first decision variables, such as heating temperature and holding time, are iteratively adjusted to achieve precise optimization of the hardness performance of the first functional end.

[0075] Simultaneously, taking the coating adhesion performance attribute of the second functional end as the primary optimization objective and its related process parameters as the second decision variables, a second optimization channel is constructed by combining the predictive capability of the second single-function performance evaluation body. This second optimization channel aims to maximize the coating adhesion performance of the plum blossom end. It utilizes the second single-function performance evaluation body to establish a quantitative mapping relationship between process parameters such as electroplating current density, plating bath temperature, and cooling rate and coating adhesion performance indicators. Through Bayesian optimization algorithm, iterative adjustments are made to the second decision variables such as electroplating current density and cooling rate, thereby achieving precise optimization of the coating adhesion performance of the second functional end.

[0076] Furthermore, the fourth single-function performance evaluation module is integrated into the first optimization channel to form the first master-slave optimization channel. This fourth single-function performance evaluation module uses the first decision variable, namely the process parameters of the first functional end, as input. It can accurately predict the potential impact of parameter adjustments on the coating adhesion performance of the second functional end, enabling the first optimization channel to anticipate potential negative effects on the other functional end while pursuing optimal hardness performance. Similarly, the third single-function performance evaluation module is integrated into the second optimization channel to form the second master-slave optimization channel. This third single-function performance evaluation module uses the second decision variable, namely the process parameters of the second functional end, as input. It can accurately predict the potential impact of parameter adjustments on the hardness performance of the first functional end, enabling the second optimization channel to evaluate its interaction with the performance of the other functional end while optimizing coating adhesion performance.

[0077] Finally, an information exchange channel is established and connected to the first master-slave optimization channel and the second master-slave optimization channel. This information exchange channel serves as a bidirectional data transmission and coordination hub, responsible for real-time transmission of intermediate optimization results and performance prediction information generated by the two optimization channels during the iteration process. Specifically, this information exchange channel periodically collects the current process parameter scheme of the first master-slave optimization channel and its prediction results on the hardness of the first functional end, while simultaneously acquiring the influence value of the coating adhesion of the second functional end predicted by the fourth evaluation body. Simultaneously, it collects the current process parameter scheme of the second master-slave optimization channel and its prediction results on the coating adhesion of the second functional end, while simultaneously acquiring the influence value of the hardness of the first functional end predicted by the third evaluation body. Through this bidirectional information exchange mechanism, the first and second master-slave optimization channels can perceive the optimization status of the other channel and the cross-influence of their own decisions on the performance of the other end in real time, providing data support for collaborative optimization decisions.

[0078] Therefore, establishing an information exchange channel not only achieves data synchronization between the two channels, but also enables the first master-slave optimization channel and the second master-slave optimization channel to dynamically adjust their optimization direction based on the interaction information through a coordination mechanism based on prediction results, ultimately achieving collaborative optimization of the performance of the two functional terminals and efficient search for the global optimal solution.

[0079] S40: Combine the historical production data of the entire process with the bidirectional master-slave optimization channel to perform bidirectional parallel optimization iteration, and output the optimal decision variable as the optimal process scheme;

[0080] Combining the historical production data of the entire process with the bidirectional master-slave optimization channel, bidirectional parallel optimization iteration is performed, and the optimal decision variable is output as the optimal process scheme, including:

[0081] The master-slave optimization channel is initialized with the current optimal first functional process parameter as the starting point for selecting the first decision variable and the current optimal second functional process parameter as the starting point for selecting the second decision variable.

[0082] In each iteration step, the first master-slave optimization channel updates the first decision variable in the direction of the first functional end degradation and inputs it into the first single-function performance evaluation body to obtain a first predicted value for the performance of the first functional end. At the same time, the first decision variable is input into the fourth single-function performance evaluation body to obtain a second predicted value for the performance of the second functional end.

[0083] In each iteration step, the second master-slave optimization channel updates the second decision variable in the direction of the second functional end degradation and inputs it into the third single-function performance evaluation body to obtain a third predicted value for the performance of the first functional end. At the same time, the second decision variable is input into the third single-function performance evaluation body to obtain a fourth predicted value for the performance of the second functional end.

[0084] After each iteration step, dual-channel information exchange is performed by combining the preset information exchange constraints with the information exchange channel, and convergence determination is then performed.

[0085] If convergence is determined, the optimal decision variable is randomly selected and output as the optimal process scheme, using the first decision variable and the second decision variable as the process interval.

[0086] Bidirectional parallel optimization iteration is an intelligent optimization method based on dual-channel collaborative search, used to solve the multi-objective coupled optimization problem of the production process parameters for the dual-functional ends of a two-way wrench. By combining historical production data from the entire process with bidirectional master-slave optimization channels to perform bidirectional parallel optimization iteration, the conflicting relationship between the hardness performance of the first functional end and the coating adhesion performance of the second functional end can be effectively coordinated, facilitating the rapid search for the globally optimal process parameter scheme while ensuring that the performance of both ends meets the standards. The specific process is as follows:

[0087] First, based on the current optimal process parameter settings from the historical production data of the entire process, the starting search point of the bidirectional master-slave optimization channel is initialized. The first master-slave optimization channel uses the historically optimal first-function process parameter as the initial decision variable, focusing on optimizing the hardness performance of the first function; the second master-slave optimization channel uses the historically optimal second-function process parameter as the initial decision variable, focusing on optimizing the coating adhesion performance of the second function. Setting the current optimal process parameter as the starting search point can make full use of historical production experience, avoid exploring the invalid parameter space, and significantly improve optimization efficiency.

[0088] Furthermore, during the iterative optimization process, the first master-slave optimization channel and the second master-slave optimization channel run in parallel, updating the decision variables along specific search directions. After updating the first decision variable, the first master-slave optimization channel predicts the changing trend of the hardness performance of the first functional end through the first single-function performance evaluation body, and simultaneously predicts the potential impact of these parameter adjustments on the coating adhesion performance of the second functional end through the fourth single-function performance evaluation body. Similarly, after updating the second decision variable, the second master-slave optimization channel predicts the changing trend of the coating adhesion performance of the second functional end through the second single-function performance evaluation body, and simultaneously predicts the potential impact of these parameter adjustments on the hardness performance of the first functional end through the third single-function performance evaluation body.

[0089] After each iteration step, prediction results are synchronously exchanged between the first master-slave optimization channel and the second master-slave optimization channel via an information exchange channel. These results include the predicted hardness performance of the first functional end, the predicted coating adhesion performance of the second functional end, and the corresponding cross-influence predictions. Based on the exchanged information, a convergence determination is performed to assess whether the current process parameter scheme can simultaneously meet the performance requirements of both functional ends and whether the prediction results of the two channels are consistent.

[0090] Specifically, after each iteration step, a dual-channel information exchange is performed based on the preset information exchange constraints and the information exchange channel, followed by a convergence determination, including:

[0091] Configure the information exchange constraints, including exchange interval constraints, first iteration count, second iteration count, first residual constraints, and second residual constraints;

[0092] When the number of iterations is greater than or equal to the number of the first iterations, the second predicted value and the fourth predicted value are exchanged through the information exchange channel, taking into account the exchange interval constraint, and a convergence determination is performed accordingly.

[0093] First convergence determination: Both the first predicted value and the third predicted value have reached the preset performance threshold;

[0094] Second convergence determination: The first residual between the first predicted value and the fourth predicted value is less than the first residual constraint, and the second residual between the second predicted value and the third predicted value is less than the second residual constraint;

[0095] If both the first convergence determination and the second convergence determination are satisfied, then the optimization is terminated and the convergence determination is output.

[0096] If either the first convergence criterion or the second convergence criterion is not satisfied, then continue iterative optimization;

[0097] The optimization process terminates when the total number of iterations is greater than or equal to the number of the second iteration.

[0098] First, configure the information exchange constraint parameters, including the exchange interval constraint, the first iteration count, the second iteration count, the first residual constraint, and the second residual constraint. The exchange interval constraint specifies the frequency of information exchange; the first iteration count sets the minimum number of iterations allowed to begin information exchange; the second iteration count defines the maximum number of iterations; and the first and second residual constraints respectively limit the allowable deviation range of the prediction results from the two functional ends. The information exchange constraint parameters are comprehensively set based on specific production process requirements, historical optimization experience data, and system computing resources. For example, the exchange interval constraint is configured to exchange information once every 5 iterations, the first iteration count is set to 20, the second iteration count is set to 100, the first residual constraint is configured to ensure that the hardness performance prediction deviation does not exceed ±1.0 HRC, and the second residual constraint is configured to ensure that the coating adhesion prediction deviation does not exceed ±5%.

[0099] Furthermore, when the number of iterations reaches or exceeds the first iteration number, according to the exchange interval constraint, the second and fourth predicted values ​​are exchanged between the two channels through the information exchange channel. The second predicted value represents the predicted impact of the first channel parameter adjustment on the performance of the second functional terminal, and the fourth predicted value represents the predicted optimization of the second functional terminal performance by the second channel parameter adjustment. After the exchange is completed, a dual convergence determination is performed:

[0100] The first convergence criterion requires that both the first and third predicted values ​​reach a preset performance threshold. The first predicted value represents the optimization result of the first channel on the performance of the first functional end, and the third predicted value represents the predicted impact of the second channel parameter adjustment on the performance of the first functional end. This criterion ensures that the performance of both functional ends meets the basic requirements.

[0101] The second convergence criterion requires that the first residual between the first and fourth predicted values ​​be less than the first residual constraint, and the second residual between the second and third predicted values ​​be less than the second residual constraint. This criterion ensures the consistency of the dual-channel prediction results, indicating that the two optimized channels have sufficient accuracy in predicting mutual influences.

[0102] If both the first and second convergence criteria are met, the optimization process is considered converged, the iteration terminates, and the current decision variable is output as the optimal process solution. If either criterion is not met, the optimization process continues iteratively. When the total number of iterations reaches the second iteration count, the optimization process terminates regardless of whether the convergence criteria are met, ensuring efficient utilization of computational resources.

[0103] The convergence determination mechanism with dual verification ensures that the optimization results meet both performance requirements and prediction accuracy, effectively avoiding premature convergence or failure to converge, and providing a reliable termination criterion for production process optimization.

[0104] Finally, if the convergence condition is met, the range of values ​​for the first and second decision variables obtained during the iteration process is taken as the feasible process interval. Within this interval, a set of optimal decision variables is randomly selected as the final optimal process scheme output. This optimal process scheme output can simultaneously ensure that the performance of the two functional ends reaches the optimal balance, effectively solving the coupling conflict problem of production process parameters of the two functional ends.

[0105] S50: Apply the optimal process scheme to the entire production environment of the target dual-purpose wrench for full-process production management.

[0106] First, the optimal decision variables output after optimization iteration—that is, the optimal set of process parameters that achieves a balance between the hardness performance of the first functional end and the coating adhesion performance of the second functional end—are sent to the production execution system. Through the data communication interface under the industrial IoT architecture, these optimal process parameters are automatically allocated to the corresponding equipment controllers on the production line. During production execution, a sensor network deployed in each process stage collects equipment operating status parameters and actual process values ​​in real time, comparing them with the target values ​​set in the optimal process plan. When deviations occur, a compensation mechanism is automatically triggered, adjusting the equipment operating parameters through closed-loop control to ensure that the production process always remains within the optimal process window.

[0107] Meanwhile, based on the real-time performance data of the two functional ends acquired by the quality inspection equipment, the implementation effect of the optimal process scheme is continuously evaluated. When changes in material properties, equipment status drift, or changes in environmental factors cause product performance to deviate from expectations, the optimized process can be re-executed, realizing dynamic optimization and continuous improvement of the production process, and achieving a complete closed loop from optimization results to production practice.

[0108] In summary, the embodiments of this application have at least the following technical effects:

[0109] Compared to existing technologies, this application firstly solves the deep coupling conflict problem of production process parameters on both ends by innovatively constructing a bidirectional master-slave optimization channel and a migration mapping model, achieving true global collaborative optimization and ensuring that the performance indicators of both ends simultaneously meet the highest standard requirements. Secondly, it adopts a novel optimization architecture that combines a parallel computing framework with a lightweight proxy model, which significantly improves the optimization solution efficiency, significantly reduces the consumption of computing resources, and can find a better Pareto front solution in a shorter time.

[0110] Furthermore, by establishing a cross-scale data fusion mechanism, accurate predictions were achieved from process parameters to microscopic performance and then to macroscopic manifestations, transforming quality control from a passive response to proactive prevention. Finally, by directly distributing the optimal process plan to the production execution system, intelligent management of the entire production process was realized, significantly improving product consistency and reducing scrap rates and production costs.

[0111] Example 2, as Figure 2 As shown, based on the same inventive concept as the intelligent management method for the entire production process of dual-purpose wrenches under the industrial IoT architecture provided in Embodiment 1, this embodiment of the invention also provides an intelligent management system for the entire production process of dual-purpose wrenches under the industrial IoT architecture, including:

[0112] Data acquisition module 11 is used to acquire single-function production sample data and full-process historical production data, constrained by the usage characteristics of the target dual-purpose wrench.

[0113] The evaluation body construction and training module 12 is used to construct and train multiple functional performance evaluation bodies based on the single-function production sample data and the full-process historical production data, and generate an evaluation body set.

[0114] The optimization channel construction module 13 is used to construct a bidirectional master-slave optimization channel according to the functional performance evaluation set, wherein the bidirectional master-slave optimization channel includes a first master-slave optimization channel, a second master-slave optimization channel and an information exchange channel;

[0115] The optimization iteration module 14 is used to combine the historical production data of the entire process with the bidirectional master-slave optimization channel to perform bidirectional parallel optimization iteration, and output the optimal decision variable as the optimal process scheme.

[0116] The production management execution module 15 is used to apply the optimal process scheme to the entire production environment of the target dual-purpose wrench for full-process production management.

[0117] Specifically, the data acquisition module 11 is used for:

[0118] Specifically, constrained by the intended use characteristics of the target dual-purpose wrench, single-function production sample data and full-process historical production data are obtained, including:

[0119] The first and second functional ends of the target dual-purpose wrench are defined as functional constraints, and the raw material specifications of the target dual-purpose wrench are defined as basic constraints.

[0120] Based on the functional constraints and the basic constraints, the first single-function production sample data corresponding to the first functional end and the second single-function production sample data corresponding to the second functional end are obtained respectively.

[0121] Obtain the full-process historical production data of the target dual-purpose wrench, including the coupled production process of the first functional end and the second functional end;

[0122] Both the single-function production sample data and the full-process historical production data include process parameter sequences and corresponding performance test data.

[0123] The evaluation body construction and training module 12 is specifically used for:

[0124] Based on the single-function production sample data and the full-process historical production data, multiple functional performance evaluation bodies are constructed and trained respectively, generating an evaluation body set, including:

[0125] Based on the first single-function production sample data, a first single-function performance evaluation body is constructed, which is used to predict the performance of the first function.

[0126] Based on the second single-function production sample data, a second single-function performance evaluation body is constructed, which is used to predict the performance of the second function.

[0127] Create copies of the first single-function performance evaluation body and the second single-function performance evaluation body respectively, and perform enhancement training on the first copy evaluation body and the second copy evaluation body based on the full-process historical production data to obtain the third single-function performance evaluation body and the fourth single-function performance evaluation body.

[0128] The first single-function performance evaluation body, the second single-function performance evaluation body, the third single-function performance evaluation body, and the fourth single-function performance evaluation body are combined to form the evaluation body set.

[0129] The first single-function performance evaluation body is used to predict hardness performance attributes, and the second single-function performance evaluation body is used to predict coating adhesion performance attributes.

[0130] The optimized channel construction module 13 is specifically used for:

[0131] First, a bidirectional master-slave optimization channel is constructed based on the aforementioned functional performance evaluation set. Prior to this, the following steps are also included:

[0132] Model distillation was performed on the third and fourth single-function performance evaluation bodies based on the knowledge distillation method.

[0133] Based on the model distillation results, update the third and fourth single-function performance evaluation bodies.

[0134] Furthermore, a bidirectional master-slave optimization channel is constructed based on the functional performance evaluation set, wherein the bidirectional master-slave optimization channel includes a first master-slave optimization channel, a second master-slave optimization channel, and an information exchange channel, including:

[0135] The first optimization channel is constructed by taking the performance index of the first functional end as the main optimization objective, the process parameters of the first function as the first decision variable, and combining the first single-function performance evaluation body.

[0136] Using the performance indicators of the second functional end as the main optimization objective and the process parameters of the second function as the second decision variable, and combining the second single-function performance evaluation body, a second optimization channel is constructed.

[0137] The fourth single-function performance evaluation body is integrated into the first optimization channel to obtain the first master-slave optimization channel, wherein the fourth single-function performance evaluation body takes the first decision variable as input;

[0138] The third single-function performance evaluation body is integrated into the second optimization channel to obtain the second master-slave optimization channel, wherein the third single-function performance evaluation body takes the second decision variable as input;

[0139] Establish the information exchange channel and connect it to the first master-slave optimization channel and the second master-slave optimization channel.

[0140] Specifically, the optimization iteration module 14 is used for:

[0141] Combining the historical production data of the entire process with the bidirectional master-slave optimization channel, bidirectional parallel optimization iteration is performed, and the optimal decision variable is output as the optimal process scheme, including:

[0142] The master-slave optimization channel is initialized with the current optimal first functional process parameter as the starting point for selecting the first decision variable and the current optimal second functional process parameter as the starting point for selecting the second decision variable.

[0143] In each iteration step, the first master-slave optimization channel updates the first decision variable in the direction of the first functional end degradation and inputs it into the first single-function performance evaluation body to obtain a first predicted value for the performance of the first functional end. At the same time, the first decision variable is input into the fourth single-function performance evaluation body to obtain a second predicted value for the performance of the second functional end.

[0144] In each iteration step, the second master-slave optimization channel updates the second decision variable in the direction of the second functional end degradation and inputs it into the third single-function performance evaluation body to obtain a third predicted value for the performance of the first functional end. At the same time, the second decision variable is input into the third single-function performance evaluation body to obtain a fourth predicted value for the performance of the second functional end.

[0145] After each iteration step, dual-channel information exchange is performed by combining the preset information exchange constraints with the information exchange channel, and convergence determination is then performed.

[0146] If convergence is determined, the optimal decision variable is randomly selected and output as the optimal process scheme, using the first decision variable and the second decision variable as the process interval.

[0147] Specifically, after each iteration step, a dual-channel information exchange is performed based on the preset information exchange constraints and the information exchange channel, followed by a convergence determination, including:

[0148] Configure the information exchange constraints, including exchange interval constraints, first iteration count, second iteration count, first residual constraints, and second residual constraints;

[0149] When the number of iterations is greater than or equal to the number of the first iterations, the second predicted value and the fourth predicted value are exchanged through the information exchange channel, taking into account the exchange interval constraint, and a convergence determination is performed accordingly.

[0150] First convergence determination: Both the first predicted value and the third predicted value have reached the preset performance threshold;

[0151] Second convergence determination: The first residual between the first predicted value and the fourth predicted value is less than the first residual constraint, and the second residual between the second predicted value and the third predicted value is less than the second residual constraint;

[0152] If both the first convergence determination and the second convergence determination are satisfied, then the optimization is terminated and the convergence determination is output.

[0153] If either the first convergence criterion or the second convergence criterion is not satisfied, then continue iterative optimization;

[0154] The optimization process terminates when the total number of iterations is greater than or equal to the number of the second iteration.

[0155] The production management execution module 15 is specifically used for:

[0156] The optimal process scheme is applied to the entire production environment of the target dual-purpose wrench for full-process production management.

[0157] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0158] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0159] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for intelligent management of the entire production process of dual-purpose wrenches under an industrial Internet of Things (IoT) architecture, characterized in that: include: Constrained by the application characteristics of the target dual-purpose wrench, single-function production sample data and full-process historical production data are obtained; Based on the single-function production sample data and the full-process historical production data, multiple functional performance evaluation bodies are constructed and trained to generate a functional performance evaluation body set. A bidirectional master-slave optimization channel is constructed based on the functional performance evaluation set, wherein the bidirectional master-slave optimization channel includes a first master-slave optimization channel, a second master-slave optimization channel, and an information exchange channel; The entire process of historical production data is combined with the bidirectional master-slave optimization channel to perform bidirectional parallel optimization iteration, and the optimal decision variable is output as the optimal process scheme; The optimal process scheme is applied to the entire production environment of the target dual-purpose wrench for full-process production management; Specifically, based on the single-function production sample data and the full-process historical production data, multiple functional performance evaluation bodies are constructed and trained to generate a functional performance evaluation body set, including: Based on the first single-function production sample data, a first single-function performance evaluation body is constructed, which is used to predict the performance of the first function. Based on the production sample data of the second single function, a performance evaluation body for the second single function is constructed, which is used to predict the performance of the second function. Create copies of the first single-function performance evaluation body and the second single-function performance evaluation body respectively, and perform enhancement training on the first copy evaluation body and the second copy evaluation body based on the full-process historical production data to obtain the third single-function performance evaluation body and the fourth single-function performance evaluation body. The first single-function performance evaluation body, the second single-function performance evaluation body, the third single-function performance evaluation body and the fourth single-function performance evaluation body are combined to form the functional performance evaluation body set; Specifically, a bidirectional master-slave optimization channel is constructed based on the functional performance evaluation set. This bidirectional master-slave optimization channel includes a first master-slave optimization channel, a second master-slave optimization channel, and an information exchange channel, comprising: The first optimization channel is constructed by taking the performance index of the first functional end as the main optimization objective, the process parameters of the first function as the first decision variable, and combining the first single-function performance evaluation body. Using the performance indicators of the second functional end as the main optimization objective and the process parameters of the second function as the second decision variable, and combining the second single-function performance evaluation body, a second optimization channel is constructed. The fourth single-function performance evaluation body is integrated into the first optimization channel to obtain the first master-slave optimization channel, wherein the fourth single-function performance evaluation body takes the first decision variable as input; The third single-function performance evaluation body is integrated into the second optimization channel to obtain the second master-slave optimization channel, wherein the third single-function performance evaluation body takes the second decision variable as input; Establish the information exchange channel and connect it to the first master-slave optimization channel and the second master-slave optimization channel.

2. The intelligent management method for the entire production process of dual-purpose wrenches under the industrial Internet of Things architecture as described in claim 1, characterized in that, Constrained by the intended use characteristics of the target dual-purpose wrench, single-function production sample data and full-process historical production data were obtained, including: The first and second functional ends of the target dual-purpose wrench are defined as functional constraints, and the raw material specifications of the target dual-purpose wrench are defined as basic constraints. Based on the functional constraints and the basic constraints, the first single-function production sample data corresponding to the first functional end and the second single-function production sample data corresponding to the second functional end are obtained respectively. Obtain the full-process historical production data of the target dual-purpose wrench, including the coupled production process of the first functional end and the second functional end; Both the single-function production sample data and the full-process historical production data include process parameter sequences and corresponding performance test data.

3. The intelligent management method for the entire production process of dual-purpose wrenches under the industrial Internet of Things architecture as described in claim 2, characterized in that, Combining the historical production data of the entire process with the bidirectional master-slave optimization channel, bidirectional parallel optimization iteration is performed, and the optimal decision variable is output as the optimal process scheme, including: The master-slave optimization channel is initialized with the current optimal first functional process parameter as the starting point for selecting the first decision variable and the current optimal second functional process parameter as the starting point for selecting the second decision variable. In each iteration step, the first master-slave optimization channel updates the first decision variable in the direction of the first functional end degradation and inputs it into the first single-function performance evaluation body to obtain a first predicted value for the performance of the first functional end. At the same time, the first decision variable is input into the fourth single-function performance evaluation body to obtain a second predicted value for the performance of the second functional end. In each iteration step, the second master-slave optimization channel updates the second decision variable in the direction of the second functional end degradation and inputs it into the third single-function performance evaluation body to obtain a third predicted value for the performance of the first functional end. At the same time, the second decision variable is input into the third single-function performance evaluation body to obtain a fourth predicted value for the performance of the second functional end. After each iteration step, dual-channel information exchange is performed by combining the preset information exchange constraints with the information exchange channel, and convergence determination is then performed. If convergence is determined, the optimal decision variable is randomly selected and output as the optimal process scheme, using the first decision variable and the second decision variable as the process interval.

4. The intelligent management method for the entire production process of dual-purpose wrenches under the industrial Internet of Things architecture as described in claim 3, characterized in that, After each iteration step, a dual-channel information exchange is performed, combining the preset information exchange constraints with the information exchange channel, and a convergence determination is made, including: Configure the information exchange constraints, including exchange interval constraints, first iteration count, second iteration count, first residual constraints, and second residual constraints; When the number of iterations is greater than or equal to the number of the first iterations, the second predicted value and the fourth predicted value are exchanged through the information exchange channel, taking into account the exchange interval constraint, and a convergence determination is performed accordingly. First convergence determination: Both the first predicted value and the third predicted value have reached the preset performance threshold; Second convergence determination: The first residual between the first predicted value and the fourth predicted value is less than the first residual constraint, and the second residual between the second predicted value and the third predicted value is less than the second residual constraint; If both the first convergence determination and the second convergence determination are satisfied, then the optimization is terminated and the convergence determination is output. If either the first convergence criterion or the second convergence criterion is not satisfied, then continue iterative optimization; The optimization process terminates when the total number of iterations is greater than or equal to the number of the second iteration.

5. The intelligent management method for the entire production process of dual-purpose wrenches under the industrial Internet of Things architecture as described in claim 2, characterized in that, Based on the aforementioned functional performance evaluation set, a bidirectional master-slave optimization channel is constructed. Prior to this, the following steps are also included: Model distillation was performed on the third and fourth single-function performance evaluation bodies based on the knowledge distillation method. Based on the model distillation results, update the third and fourth single-function performance evaluation bodies.

6. The intelligent management method for the entire production process of dual-purpose wrenches under the industrial Internet of Things architecture as described in claim 1, characterized in that, The first single-function performance evaluation body is used to predict hardness performance properties, and the second single-function performance evaluation body is used to predict coating adhesion performance properties.

7. A smart management system for the entire production process of dual-purpose wrenches under an industrial IoT architecture, characterized in that: The method for intelligent management of the entire production process of dual-purpose wrenches under the industrial Internet of Things architecture according to any one of claims 1-6 includes: The data acquisition module is used to acquire single-function production sample data and full-process historical production data, constrained by the usage characteristics of the target dual-purpose wrench. The evaluation body construction and training module is used to construct and train multiple functional performance evaluation bodies based on the single-function production sample data and the full-process historical production data, and generate a functional performance evaluation body set. The optimized channel construction module is used to construct a bidirectional master-slave optimized channel according to the functional performance evaluation set, wherein the bidirectional master-slave optimized channel includes a first master-slave optimized channel, a second master-slave optimized channel and an information exchange channel; The optimization iteration module is used to combine the historical production data of the entire process with the bidirectional master-slave optimization channel to perform bidirectional parallel optimization iteration, and output the optimal decision variable as the optimal process scheme. The production management execution module is used to apply the optimal process scheme to the entire production environment of the target dual-purpose wrench for full-process production management.

Images