Method and device for determining extrusion parameters of a metal profile, and readable storage medium

Abstract

The application provides a metal profile extrusion parameter determination method, device and readable storage medium, wherein the method comprises the following steps: obtaining a target characteristic parameter of a target metal profile and corresponding historical process parameters in a historical extrusion process; determining a knowledge graph of the metal profile extrusion process according to the historical process parameters; determining a first extrusion parameter of the target metal profile according to the target characteristic parameter and the knowledge graph; inputting the target characteristic parameter into a trained metal profile extrusion parameter prediction model for prediction to determine a second extrusion parameter of the target metal profile; the metal profile extrusion parameter prediction model is obtained by training a preset neural network based on an improved zebra optimization algorithm and the historical process parameters; and determining a target extrusion parameter corresponding to the target metal profile according to the first extrusion parameter and the second extrusion parameter. The scheme of the application improves the determination efficiency and accuracy of the profile extrusion parameter, thereby improving the profile extrusion quality and reducing the extrusion energy consumption.

Description

Technical Field

[0001] This invention relates to the field of metal material extrusion processing technology, and in particular to a method, apparatus and readable storage medium for determining extrusion parameters of metal profiles. Background Technology

[0002] Traditional process optimization primarily relies on a combination of finite element simulation and orthogonal experiments. This approach is not only computationally expensive and time-consuming, but also struggles to meet the rapid response demands of producing new aluminum profiles. In actual production, the setting of key process parameters such as extrusion speed and temperature directly impacts energy consumption and product quality stability. Inappropriate parameter combinations can easily lead to problems such as profile deformation and surface defects. Although numerical simulation and machine learning technologies have been gradually applied in this field, existing methods still suffer from low computational efficiency, insufficient interpretability, and limited multi-objective optimization capabilities. This results in companies remaining highly reliant on empirical parameters, making it difficult to achieve the optimal balance between energy consumption and quality. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method, apparatus and readable storage medium for determining the extrusion parameters of metal profiles, so as to improve the accuracy of extrusion parameters in the metal profile extrusion process, thereby improving extrusion efficiency and quality and reducing extrusion energy consumption.

[0004] To address the aforementioned technical problems, embodiments of the present invention provide a method for determining extrusion parameters of metal profiles, comprising:

[0005] Obtain the target feature parameters of the target metal profile and the corresponding historical process parameters in the historical extrusion process. The historical process parameters include the historical models of multiple metal profiles and the historical feature parameters, historical extrusion parameters and historical extrusion results that correspond one-to-one with the historical models.

[0006] Based on the historical process parameters, a knowledge graph of the metal profile extrusion process is determined. The knowledge graph is a relationship structure diagram between the historical model, the historical characteristic parameters, the historical extrusion parameters, and the historical extrusion results.

[0007] Based on the target feature parameters and the knowledge graph, the first extrusion parameters of the target metal profile are determined;

[0008] The target feature parameters are input into a trained metal profile extrusion parameter prediction model for prediction to determine the second extrusion parameter of the target metal profile; the metal profile extrusion parameter prediction model is obtained by training a preset neural network based on an improved zebra optimization algorithm and the historical process parameters;

[0009] The target extrusion parameters corresponding to the target metal profile are determined based on the first extrusion parameter and the second extrusion parameter.

[0010] In one embodiment, determining a knowledge graph of the metal profile extrusion process based on the historical process parameters includes:

[0011] The ontology framework of the knowledge graph is determined based on the historical model, the historical feature parameters, the historical compression parameters, and the historical compression results.

[0012] Based on the historical model, the historical feature parameters, the historical extrusion parameters, and the structured historical process parameters corresponding to the historical extrusion results, the relationships between entities in the ontology framework are determined.

[0013] The ontology framework, entities, and relationships between entities are stored in a preset graph database to determine the knowledge graph corresponding to the metal profile extrusion process.

[0014] In one embodiment, determining the ontology framework of the knowledge graph based on the historical model, the historical feature parameters, the historical compression parameters, and the historical compression results includes:

[0015] Based on the historical model, the historical characteristic parameters, the historical extrusion parameters, and the historical extrusion results, determine the profile model entity, the characteristic parameter entity, the extrusion parameter entity, and the extrusion result entity;

[0016] According to the preset hierarchical structure order, the profile model entity, the feature parameter entity, the extrusion parameter entity, and the extrusion result entity are sorted from top to bottom to determine the main body frame.

[0017] In one embodiment, determining the first extrusion parameters of the target metal profile based on the target feature parameters and the knowledge graph includes:

[0018] Determine the similarity between the target feature parameters and the historical feature parameters corresponding to each historical model in the knowledge graph;

[0019] Candidate historical models in the knowledge graph are identified based on similarity, and the first compression parameter is determined based on the historical compression parameters corresponding to the candidate historical models.

[0020] In one embodiment, determining the similarity between the target feature parameters and the historical feature parameters corresponding to each historical model in the knowledge graph includes:

[0021] The similarity is determined using the following formula:

[0022]

[0023] Among them, D j This represents the similarity between the feature parameters of the j-th historical metal profile and the target metal profile in the knowledge graph, where j = 1, 2, ..., 3, m; x i Let y represent the i-th target feature parameter of the target metal profile. ij This represents the i-th historical feature parameter of the j-th historical metal profile in the knowledge graph; n represents the type of feature parameter.

[0024] In one embodiment, the metal profile extrusion parameter prediction model is obtained by training a preset neural network based on an improved zebra optimization algorithm and the historical process parameters, including:

[0025] Determine the initial parameter population of the preset neural network; the initial parameter population contains multiple sets of parameters, each set of parameters includes initial weights and initial threshold parameters, and the initial parameter population corresponds to an initial individual population containing multiple zebra individuals, with each set of parameters corresponding one-to-one with the position of each zebra individual;

[0026] Based on the historical process parameters, each set of parameters in the initial parameter population, and the preset neural network, determine the first fitness value of each zebra individual in the initial individual population;

[0027] The positions of zebra individuals in the initial individual population are iteratively updated based on the first fitness value and a preset optimization strategy to determine the target positions of zebra individuals in the initial individual population.

[0028] The parameter value corresponding to the target position is determined as the target parameter of the preset neural network, and the preset neural network is iteratively trained according to the target parameter and the historical process parameter until the preset convergence condition or the first preset number of iterations is reached, so as to obtain the metal profile extrusion parameter prediction model.

[0029] In one embodiment, the zebra individuals in the initial population are iteratively updated based on the first fitness value and a preset optimization strategy to determine the target location of the zebra individuals in the initial population, including:

[0030] Based on the first fitness value, determine the current optimal position in the initial individual population and the first fitness value corresponding to the current optimal position;

[0031] The positions of zebra individuals in the initial population are iteratively updated based on the preset Levi flight exploration strategy, the preset nonlinear convergence factor, and the optimal position until a second preset number of iterations is reached, in order to obtain the target position.

[0032] In one embodiment, determining the target extrusion parameters corresponding to the target metal profile based on the first extrusion parameters and the second extrusion parameters includes:

[0033] The first extrusion parameter and the second extrusion parameter are weighted and fused according to a preset weight value to obtain the target extrusion parameter.

[0034] Embodiments of the present invention also provide a device for determining extrusion parameters of metal profiles, comprising:

[0035] The acquisition module is used to acquire the target feature parameters of the target metal profile and the corresponding historical process parameters in the historical extrusion process. The historical process parameters include the historical models of multiple metal profiles and the historical feature parameters, historical extrusion parameters and historical extrusion results that correspond one-to-one with the historical models.

[0036] The processing module is configured to: determine a knowledge graph of the metal profile extrusion process based on the historical process parameters, wherein the knowledge graph is a relational structure diagram between the historical model, the historical feature parameters, the historical extrusion parameters, and the historical extrusion results; determine a first extrusion parameter of the target metal profile based on the target feature parameters and the knowledge graph; input the target feature parameters into a trained metal profile extrusion parameter prediction model for prediction to determine a second extrusion parameter of the target metal profile; the metal profile extrusion parameter prediction model is obtained by training a preset neural network based on an improved zebra optimization algorithm and the historical process parameters; and determine the target extrusion parameter corresponding to the target metal profile based on the first extrusion parameter and the second extrusion parameter.

[0037] Embodiments of the present invention also provide a computer-readable storage medium storing a program that, when executed by a processor, implements the method described above.

[0038] The above-described solution of the present invention has at least the following beneficial effects:

[0039] The above-mentioned solution of the present invention obtains the target feature parameters of the target metal profile and the corresponding historical process parameters in the historical extrusion process. The historical process parameters include multiple historical models of metal profiles and historical feature parameters, historical extrusion parameters, and historical extrusion results corresponding to each historical model. Based on the historical process parameters, a knowledge graph of the metal profile extrusion process is determined. The knowledge graph is a relational structure diagram between the historical models, the historical feature parameters, the historical extrusion parameters, and the historical extrusion results. Based on the target feature parameters and the knowledge graph, the first extrusion parameter of the target metal profile is determined. The target feature parameters are input into a trained metal profile extrusion parameter prediction model for prediction to determine the second extrusion parameter of the target metal profile. The metal profile extrusion parameter prediction model is obtained by training a preset neural network based on an improved zebra optimization algorithm and the historical process parameters. Based on the first extrusion parameter and the second extrusion parameter, the target extrusion parameter corresponding to the target metal profile is determined, so as to quickly and accurately provide the optimal combination of extrusion parameters for profiles with different cross-sectional characteristics, significantly reduce extrusion energy consumption while ensuring product quality, and effectively solve the problems of high computational cost, long cycle, and difficulty in meeting the production needs of new materials by traditional optimization methods. It provides efficient and reliable technical solutions for energy conservation, emission reduction, and process optimization in profile manufacturing. Attached Figure Description

[0040] Figure 1 This is a flowchart of a method for determining extrusion parameters of metal profiles provided in an embodiment of the present invention;

[0041] Figure 2 This is a knowledge graph of metal profile extrusion process provided in an optional embodiment of the present invention;

[0042] Figure 3 This is a flowchart of a method for determining extrusion parameters of 4040g aluminum profiles provided in an optional embodiment of the present invention;

[0043] Figure 4 This is a flowchart of an optional embodiment of the present invention, showing the improved zebra optimization algorithm for optimizing a BP neural network.

[0044] Figure 5 This is a schematic block diagram of the extrusion parameter determination device for metal profiles provided in an embodiment of the present invention;

[0045] Figure 6 This is a schematic block diagram of an electronic device provided in an embodiment of the present invention;

[0046] Figure 7 This is a schematic block diagram of a computing device provided in an embodiment of the present invention. Detailed Implementation

[0047] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0048] In the following description, certain specific details are set forth for the purpose of illustrating various disclosed embodiments in order to provide a thorough understanding of the various disclosed embodiments. However, those skilled in the art will recognize that embodiments may be practiced without one or more of these specific details. In other instances, well-known apparatuses, structures, and techniques associated with this application may not have been shown or described in detail to avoid unnecessarily obscuring the description of the embodiments.

[0049] Throughout this specification, references to "an embodiment" or "an embodiment" indicate that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Therefore, the appearance of "in an embodiment" or "an embodiment" in various places throughout the specification does not necessarily refer to the same embodiment. Furthermore, a particular feature, structure, or characteristic may be combined in any manner in one or more embodiments.

[0050] like Figure 1 As shown, an embodiment of the present invention provides a method for determining the extrusion parameters of a metal profile, comprising the following steps:

[0051] Step 11: Obtain the target feature parameters of the target metal profile and the corresponding historical process parameters in the historical extrusion process. The historical process parameters include the historical models of multiple metal profiles and the historical feature parameters, historical extrusion parameters and historical extrusion results that correspond one-to-one with the historical models.

[0052] Step 12: Based on historical process parameters, determine the knowledge graph of metal profile extrusion process. The knowledge graph is a structural diagram showing the relationship between historical models, historical characteristic parameters, historical extrusion parameters, and historical extrusion results.

[0053] Step 13: Determine the first extrusion parameters of the target metal profile based on the target feature parameters and the knowledge graph;

[0054] Step 14: Input the target feature parameters into the trained metal profile extrusion parameter prediction model for prediction to determine the second extrusion parameter of the target metal profile; the metal profile extrusion parameter prediction model is obtained by training a preset neural network based on the improved Zebra optimization algorithm and historical process parameters;

[0055] Step 15: Determine the target extrusion parameters corresponding to the target metal profile based on the first extrusion parameters and the second extrusion parameters.

[0056] In this embodiment, both the target feature parameters and the historical feature parameters include the cross-sectional perimeter L (unit: mm) and cross-sectional area A (unit: mm) of the metal profile. 2 The parameters include the perimeter-to-area ratio R (R = L / A) and the maximum dimension (circumscribed circle diameter of the profile, unit: mm); the first extrusion parameters, the second extrusion parameters, the target extrusion parameters, and the historical extrusion parameters all include the extrusion speed, the profile preheating temperature, the die preheating temperature, and the extrusion barrel preheating temperature; the historical extrusion results include the extrusion energy consumption and the standard deviation of the exit section temperature.

[0057] A knowledge graph of metal profile extrusion processes is constructed based on historical process parameters to effectively process, handle, and integrate these complex parameters, transforming them into a simple and clear network structure and aggregating a large amount of historical extrusion process data. This enables rapid response and reasoning based on historical extrusion process data. The knowledge graph includes historical profile models, corresponding historical feature parameters, historical extrusion parameters, and historical extrusion results. The knowledge graph is stored in triplet form, and the historical extrusion parameters and results for each profile model stored in the graph represent the optimal historical process parameters. Once the target feature parameters of the target metal profile are determined, the knowledge graph can recommend the first extrusion parameter for the target metal profile based on the similarity between the target feature parameters and the historical feature parameters in the knowledge graph.

[0058] Furthermore, a preset neural network is trained based on historical feature parameters and historical extrusion parameters from historical process parameters. During the training process, the model parameters of the preset neural network are optimized using an improved zebra optimization algorithm to obtain a trained metal profile extrusion parameter prediction model. Here, the preset neural network model can be a BP neural network model. Preferably, the key parameters of the BP neural network are: 4 nodes in the input layer, 25 nodes in the hidden layer, 4 nodes in the output layer, and the activation function is the sigmoid function. Optimizing the model parameters of the BP neural network using the improved zebra optimization algorithm enables the BP neural network to have the minimum error and the highest fitness on the training data, while improving the efficiency and accuracy of model parameter optimization. This ensures the accuracy of the prediction processing of target feature parameters based on the trained metal profile extrusion parameter prediction model, thereby obtaining a more accurate second extrusion parameter.

[0059] Furthermore, the first extrusion parameters and the second extrusion parameters are fused to obtain the target extrusion parameters for the target profile extrusion process. The solution provided by the above embodiments can quickly and accurately provide the optimal target extrusion parameters for metal profiles with different cross-sectional characteristics, significantly reducing extrusion energy consumption while ensuring product quality. This effectively solves the problems of high calculation cost, long cycle, and inability to meet the production needs of new materials in traditional optimization methods. It should be noted that the metal profiles in the above embodiments are made of materials with ductility and good plasticity, such as aluminum profiles and corresponding aluminum alloy profiles.

[0060] In an optional embodiment of the present invention, step 12 above may include:

[0061] Step 121: Determine the ontology framework of the knowledge graph based on historical models, historical feature parameters, historical compression parameters, and historical compression results.

[0062] Specifically, step 121 may include:

[0063] Step 1211: Based on historical models, historical characteristic parameters, historical extrusion parameters, and historical extrusion results, determine the profile model entity, characteristic parameter entity, extrusion parameter entity, and extrusion result entity;

[0064] Step 1212: According to the preset hierarchical structure order, sort the profile model entity, feature parameter entity, extrusion parameter entity and extrusion result entity from top to bottom to determine the body frame.

[0065] In this embodiment, historical model numbers, historical characteristic parameters (perimeter, cross-sectional area, perimeter-to-area ratio, and maximum size of the metal profile), historical extrusion parameters (extrusion speed, profile preheating temperature, die preheating temperature, and extrusion cylinder preheating temperature), and historical extrusion results (extrusion energy consumption and standard deviation of exit section temperature) are identified as corresponding entities. Preferably, entity identification can be performed by extracting the name of the corresponding entity through regular expressions or named entity algorithms. For each entity, a unique identifier can be determined to facilitate association and referencing in the knowledge graph.

[0066] Furthermore, entities can be arranged in a top-down order according to a preset hierarchical structure. For example, the profile model entity can be the top layer, the feature parameter entity the second layer, the extrusion parameter entity the third layer, and the extrusion result entity the fourth layer. Here, each entity in each layer represents a node in the knowledge graph, and the connecting edges (connection relationships) between nodes represent the association relationships between the corresponding entities. It should be noted that the profile model can also be used as the first layer, with all other entities directly connected to the profile model as second layers, to simplify the knowledge graph structure.

[0067] Furthermore, in an optional embodiment of the present invention, step 12 may further include:

[0068] Step 122: Determine the relationships between entities in the ontology framework based on historical models, historical characteristic parameters, historical extrusion parameters, and structured historical process parameters corresponding to historical extrusion results.

[0069] Step 123: Store the ontology framework, entities, and relationships between entities according to the preset graph database, and determine the knowledge graph corresponding to the metal profile extrusion process.

[0070] In this embodiment, historical models, historical feature parameters, historical compression parameters, and historical compression results are integrated into a CSV structure file, and knowledge extraction is performed using the transform_to_triples statement. Historical models and historical feature parameters, historical models and historical compression parameters, and historical compression parameters and historical compression results with connection relationships are determined, and the corresponding entities are connected after the corresponding connection relationships are determined. Furthermore, the constructed knowledge graph can be stored in the form of triples using the Neo4j graph database.

[0071] In an optional embodiment of the present invention, step 13 above may include:

[0072] Step 131: Determine the similarity between the target feature parameters and the historical feature parameters corresponding to each historical model in the knowledge graph.

[0073] Specifically, similarity can be determined using the following formula:

[0074]

[0075] Among them, D j This represents the similarity between the feature parameters of the j-th historical metal profile and the target metal profile in the knowledge graph, where j = 1, 2, ..., 3, m; x i Let y represent the i-th target feature parameter of the target metal profile. ij Let represent the i-th historical feature parameter of the j-th existing historical metal profile in the knowledge graph; n represents the type of feature parameter; here, the target feature parameter of the target metal profile and the type of historical feature parameter of the existing metal profile corresponding to each historical model in the knowledge graph are the same, n = 1, 2, 3, 4. It should be noted that there are multiple similarities in this embodiment, and the similarity corresponds one-to-one with the number of historical models in the knowledge graph.

[0076] Preferably, before performing similarity calculation, the target feature parameters of the target metal profile and the historical feature parameters in the knowledge graph need to be normalized to ensure the accuracy of determining the first extrusion parameter; more preferably, the target feature parameters and historical feature parameters can be normalized using the following formula:

[0077]

[0078] Among them, Normalized Value i Value represents the normalized i-th target feature parameter or historical feature parameter; i Min represents the original data corresponding to the i-th target feature parameter or historical feature parameter; i Max represents the minimum value of the i-th historical feature parameter in the knowledge graph. i This represents the maximum value of the i-th historical feature parameter in the knowledge graph. It should be noted that x in the above similarity calculation formula... i y i All of these represent the normalized feature parameters.

[0079] In an optional embodiment of the present invention, step 13 above may further include:

[0080] Step 132: Identify candidate historical models in the knowledge graph based on similarity, and determine the first compression parameter based on the historical compression parameters corresponding to the candidate historical models.

[0081] In this embodiment, multiple similarities are arranged in ascending order, and the historical models corresponding to the first preset number of similarities are determined as candidate historical models.

[0082] Furthermore, the first extrusion parameter is determined based on the historical extrusion parameters corresponding to the candidate historical models; preferably, the first extrusion parameter can be determined based on the following formula and the historical extrusion parameters corresponding to the candidate historical models:

[0083]

[0084] Among them, P KGt This represents the t-th extrusion parameter among the first extrusion parameters, where t = 1, 2, 3, 4 correspond to the first extrusion speed, the first profile preheating temperature, the first die preheating temperature, and the first extrusion cylinder preheating temperature, respectively; p tk This represents the t-th historical extrusion parameter of the k-th historical existing metal profile in the preset number l with the lowest similarity. k As weight, and D kThe similarity between the k-th historical metal profile in the preset number l with the lowest similarity and the target metal profile (compared to the above D) j (The calculation method is the same).

[0085] In an optional embodiment of the present invention, a metal profile extrusion parameter prediction model is obtained by training a preset neural network based on an improved zebra optimization algorithm and historical process parameters, which may include:

[0086] Step 21: Determine the initial parameter population of the preset neural network; the initial parameter population contains multiple sets of parameters, each set of parameters includes initial weights and initial threshold parameters, and the initial parameter population corresponds to the initial individual population containing multiple zebra individuals, with each set of parameters corresponding one-to-one with the position of each zebra individual.

[0087] Step 22: Determine the first fitness value of each zebra individual in the initial individual population based on historical process parameters, each set of parameters in the initial parameter population, and the preset neural network.

[0088] Step 23: Iteratively update the positions of zebra individuals in the initial individual population based on the first fitness value and the preset optimization strategy to determine the target positions of zebra individuals in the initial individual population.

[0089] Step 24: Determine the parameter value corresponding to the target position as the target parameter of the preset neural network, and iteratively train the preset neural network according to the target parameter and historical process parameters until the preset convergence condition or the first preset number of iterations is reached, so as to obtain the metal profile extrusion parameter prediction model.

[0090] In this embodiment, the initial parameter population corresponds to the initial zebra population of the improved zebra optimization algorithm. Each set of parameters in the initial parameter population corresponds to the position vector of each zebra individual in the initial zebra population. Further, a preset neural network is trained based on historical process parameters and the initial weights and initial threshold parameters (preset neural network biases) in each set of parameters. Here, the historical extrusion parameters in the historical process parameters are used as input values, and the extrusion results output by the model are used as predicted values. The actual historical extrusion results corresponding to the historical extrusion parameters are used as true values, and the error between the true value and the predicted value is calculated. This error is used as the first fitness value. It should be noted that each set of parameters in the initial parameter population corresponds to a first fitness value. Each first fitness value represents the initial fitness value of each zebra individual in the initial zebra population, and the smaller the initial fitness value, the better the position of the corresponding zebra individual.

[0091] Furthermore, in an optional embodiment of the present invention, step 23 may include:

[0092] Step 231: Determine the current optimal position and the corresponding first fitness value in the initial individual population based on the first fitness value;

[0093] Step 232: Based on the preset Levi flight exploration strategy, preset nonlinear convergence factor, and optimal position, iteratively update the position of zebra individuals in the initial individual population until the second preset number of iterations is reached to obtain the target position.

[0094] Here, based on the initial fitness value (first fitness value) of the zebra individual, the zebra individual with the smallest initial fitness value (smallest error) is selected from the initial zebra population as the current optimal individual, and the position corresponding to the optimal individual is taken as the current optimal position (optimal solution). The positions of zebra individuals in the initial population are then iteratively updated based on the current optimal position to obtain the optimal individual and its corresponding target position after iterative processing; specifically:

[0095] In the exploration phase of improving the zebra optimization algorithm: other zebra individuals in the population move according to the current best position corresponding to the current best individual to update their corresponding initial positions; wherein, the step size of other zebra individuals in the population during the movement process can adopt the preset Levy flight exploration strategy (alternating between long and short steps) to enhance the global search capability; here, updating the initial position of zebra individuals through the preset Levy flight exploration strategy can be expressed as:

[0096]

[0097] in, This indicates the first updated position of the a-th zebra in the initial zebra population after being updated by the preset Levy flight exploration strategy; Let X represent the initial position of the a-th zebra in the initial zebra population; Levy(β) represents the random step size following a Levy distribution; β represents the parameter of the Levy distribution; X best This indicates the current optimal position.

[0098] In the development phase of the improved zebra optimization algorithm, a preset nonlinear convergence factor can be used to slightly and randomly perturb the first updated position of the zebra after the preset Levy flight exploration strategy, so as to update the first updated position and obtain the second updated position, thus avoiding getting trapped in local optima. Here, updating the first updated position by the preset nonlinear convergence factor can be expressed as:

[0099]

[0100] in, This indicates the second update position after the preset nonlinear convergence factor is updated; F(b) represents the perturbation amplitude factor at the b-th iteration; Let represent the random perturbation vector generated for the a-th zebra in the b-th iteration; here, Cauchy mutation can also be introduced to increase randomness and prevent premature convergence.

[0101] Furthermore, the updated second position of the zebra individual is converted into the weights and threshold parameters of a preset neural network, and the preset neural network is trained again, while the corresponding error is calculated as the second fitness value. At this time, the first fitness value and the second fitness value are compared, and the second update position corresponding to the second fitness value being less than the first fitness value is retained. The above-mentioned position iterative update process using a preset optimization strategy is repeated for the retained second update positions until the maximum number of iterations is reached. After the iteration is completed, the final update position corresponding to the zebra individual with the smallest fitness value is selected as the target position, and the target position is decoded into the target weights and target threshold parameters in the target parameters of the preset neural network. The preset neural network is initialized using the target weights and target threshold parameters to obtain the final metal profile extrusion parameter prediction model. In an optional embodiment of the present invention, step 15 above may include:

[0102] Step 151: According to the preset weight value, the first extrusion parameter and the second extrusion parameter are weighted and fused to obtain the target extrusion parameter.

[0103] In this embodiment, the dual-channel fusion strategy of recommending the first extrusion parameter based on feature similarity knowledge reasoning and predicting the second extrusion parameter based on the model not only preserves the interpretability of knowledge in the field of metal profile extrusion, but also gives full play to the generalization ability of the data model, ensuring the accuracy and efficiency of obtaining the target extrusion parameter.

[0104] Preferably, the first extrusion parameter and the second extrusion parameter can be weighted and fused using the following formula:

[0105] P fin =ω1×P KGt +ω2×P NNt ;

[0106] Where t = 1, 2, 3, 4 represents the first extrusion parameter or the t-th extrusion parameter among the first extrusion parameters; P fin ω1 and ω2 represent the target extrusion parameters; ω1 and ω2 represent preset weights; preferably, an equal weight strategy is adopted here, that is, ω1 = ω2 = 0.5; P KGt =(P KG1 ,P KG2 ,P KG3 ,P KG4 These correspond to the first extrusion speed, the first profile preheating temperature, the first die preheating temperature, and the first extrusion cylinder preheating temperature in the first extrusion parameters, respectively; P NNt =(P KG1,P NN2 ,P NN3 ,P NN4 These correspond to the second extrusion speed, the second profile preheating temperature, the second die preheating temperature, and the second extrusion cylinder preheating temperature in the second extrusion parameters, respectively.

[0107] like Figure 3 As shown, in an optional embodiment of the present invention, taking a 4040g aluminum profile as an example, the method provided in the above embodiment will be further explained. The target characteristic parameters corresponding to the 4040g aluminum profile are L = 460.01mm and A = 657.1mm. 2 R = 0.7, D ​​= 53.3 mm; the specific process is as follows:

[0108] Step 31: Using the obtained optimal process parameters for aluminum profile extrusion energy consumption and structural characteristic parameters, construct a knowledge graph of aluminum profile extrusion process.

[0109] Specifically, the knowledge graph ontology is constructed by defining four entity types: aluminum profile model, feature parameters, process parameters, and extrusion results. Relationships are used to connect aluminum profile model and feature parameters; relationships are used to connect aluminum profile model and process parameters; and relationships are used to connect process parameters and extrusion results.

[0110] Step 32 involves integrating the obtained optimal process parameters for aluminum profile extrusion energy consumption, structural characteristic parameters, and extrusion results into a CSV file. Knowledge extraction is performed using the `transform_to_triples` statement, and the constructed knowledge graph is stored in triples using the Neo4j graph database. Note that the required software environment and dependencies, including the Neo4j local database, have been correctly installed before starting the construction and storage process.

[0111] Step 33: Optimize the BP neural network using the improved zebra optimization algorithm. The key parameters of the BP neural network are: 4 nodes in the input layer, 25 nodes in the hidden layer, 4 nodes in the output layer, sigmoid activation function, 20 zebras in the population, and 1000 training epochs. Figure 4 As shown, the specific optimization process is as follows:

[0112] Step 331, Initialization: First, a set of zebra positions and velocities needs to be initialized to initialize the initial weights and initial threshold parameters of the BP neural network. The position of each zebra represents a set of parameter values ​​in the neural network, while the velocity represents the parameter update magnitude.

[0113] Step 332, Fitness Evaluation: For each zebra's position, the fitness of its corresponding neural network needs to be calculated. Fitness is calculated by using a BP neural network to perform forward propagation and error backpropagation on the training data, and then evaluated based on the network's performance loss function.

[0114] Step 333, Update Optimal Position: Determine the current optimal position and fitness value based on the zebra's fitness evaluation. Step 334, Update Speed ​​and Position: Update the zebra's speed and position based on the behavior patterns of the improved zebra optimization algorithm. This includes updating the speed using the optimal position to move it towards the optimal position, and updating the position based on the speed.

[0115] Step 335, stopping condition met: Repeat steps 332 to 334 until the stopping condition of reaching the maximum number of iterations is met.

[0116] Step 336, Optimization result: When the optimization algorithm ends, the parameter values ​​can be extracted from the optimal position and applied to the BP neural network.

[0117] Here, the mean absolute error (MAE), mean square error (MSE), and root mean square error (RMSE) are selected to evaluate the prediction results of the metal profile extrusion parameter prediction model in this invention and the existing BP neural network model, as shown in Table 1 below. Table 1 shows that the metal profile extrusion parameter prediction model in this invention has lower loss and higher convergence accuracy compared to the existing BP neural network model on the extrusion parameter dataset in this experiment. The MAE, MSE, and RMSE are lower, and the performance indicators are comprehensively superior to the existing BP neural network model.

[0118] Table 1. Evaluation index reference table for the metal profile extrusion parameter prediction model (Model 1) in this invention and the existing BP neural network model (Model 2).

[0119]

[0120] Furthermore, the target feature parameters are processed through multiple different schemes to obtain recommended target extrusion parameters and extrusion result parameters determined after extrusion processing based on the target extrusion parameters, as shown in Table 2. Through experimental verification, the scheme of the present invention can reduce energy consumption by more than 6.5% and reduce the standard deviation of outlet section temperature by more than 16.4% compared with the traditional method, providing an efficient and reliable technical solution for energy saving, emission reduction and process optimization in aluminum profile production.

[0121] Table 2. Target extrusion parameters and extrusion results for each 4040g scheme.

[0122]

[0123]

[0124] This invention constructs a knowledge graph of metal profile extrusion processes, combines an improved zebra optimization algorithm with a backpropagation neural network model, and employs a recommendation strategy that integrates feature similarity matching and data-driven prediction to achieve intelligent optimization and recommendation of extrusion process parameters for novel aluminum profiles. This method can quickly and accurately provide optimal process parameter combinations for aluminum profiles with different cross-sectional characteristics, significantly reducing extrusion energy consumption while ensuring product quality. It effectively solves the problems of high computational cost, long cycle time, and inability to meet the production needs of new materials associated with traditional optimization methods. This provides an efficient and reliable technical solution for energy conservation, emission reduction, and process optimization in aluminum profile production.

[0125] The technical advantages of this invention are mainly reflected in three aspects: First, at the methodological level, it achieves the organic unity of knowledge representation and data modeling. Through the synergistic effect of the semantic association network of the knowledge graph and the nonlinear mapping capability of the neural network, a hybrid intelligent recommendation framework with both interpretability and predictability is constructed. Second, by improving the zebra optimization algorithm on the preset neural network model, and through the synergistic optimization of the Lévy flight strategy, nonlinear convergence factor, and Cauchy mutation operator, the efficiency and accuracy of neural network parameter optimization are significantly improved. Finally, at the application level, a complete technical chain from feature extraction to parameter recommendation is established, forming a scalable process knowledge accumulation and application closed loop, which is suitable for the optimization of metal profile extrusion processes and has broad application prospects.

[0126] like Figure 5 As shown, an embodiment of the present invention also provides a device 40 for determining the extrusion parameters of a metal profile, comprising:

[0127] The acquisition module 41 is used to acquire the target feature parameters of the target metal profile and the corresponding historical process parameters in the historical extrusion process. The historical process parameters include the historical models of multiple metal profiles and the historical feature parameters, historical extrusion parameters and historical extrusion results that correspond one-to-one with the historical models.

[0128] Processing module 42 is used to determine a knowledge graph of the metal profile extrusion process based on historical process parameters. The knowledge graph is a relational structure diagram between historical models, historical feature parameters, historical extrusion parameters, and historical extrusion results. Based on the target feature parameters and the knowledge graph, the first extrusion parameter of the target metal profile is determined. The target feature parameters are input into a trained metal profile extrusion parameter prediction model for prediction to determine the second extrusion parameter of the target metal profile. The metal profile extrusion parameter prediction model is obtained by training a preset neural network based on an improved zebra optimization algorithm and historical process parameters. Based on the first extrusion parameter and the second extrusion parameter, the target extrusion parameter corresponding to the target metal profile is determined.

[0129] It should be noted that this device is the same as the method for determining the extrusion parameters of the metal profiles described above. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0130] like Figure 6 As shown, embodiments of the present invention also provide an electronic device 50, comprising: a memory 51 for storing one or more computer programs; and one or more processors 52 for executing the one or more computer programs. When the computer programs are run by the processors, they perform the method for determining extrusion parameters of metal profiles as described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects. The electronic device 50 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown in this invention, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the invention described and / or claimed herein.

[0131] like Figure 7 As shown, electronic device 50 is a computing device or computer system, which may include CPU 501 (computing unit), which can perform various appropriate actions and processes according to a computer program stored in ROM 502 (read-only memory) or a computer program loaded from storage unit 508 into random access RAM 503 (memory). RAM 503 may also store various programs and data required for the operation of device 500. CPU 501, ROM 502, and RAM 503 are interconnected via bus 504. I / O interface 505 (input / output interface) is also connected to bus 504.

[0132] Multiple components in electronic device 500 are connected to I / O interface 505, including: input unit 506, such as keyboard, mouse, etc.; output unit 507, such as various types of monitors, speakers, etc.; storage unit 508, such as disk, optical disk, etc.; and communication unit 509, such as network card, modem, wireless transceiver, etc. Communication unit 509 allows device 500 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0133] CPU 501 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of CPU 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. CPU 501 performs the various methods and processes described above. For example, in some embodiments, the method 10 for determining the extrusion parameters of a metal profile can be implemented as a computer software program tangibly contained in a computer-readable storage medium, such as storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed on device 500 via ROM 502 and / or communication unit 509. When the computer program is loaded into RAM 503 and executed by CPU 501, one or more steps of the method 10 for determining the extrusion parameters of a metal profile described above can be performed. Alternatively, in other embodiments, CPU 501 may be configured by any other suitable means (e.g., by means of firmware) to perform the extrusion parameter determination method 10 for metal profiles.

[0134] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method for determining extrusion parameters of a metal profile as described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0135] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0136] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0137] In the embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0138] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0139] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0140] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0141] Furthermore, it should be noted that in the apparatus and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent solutions of the present invention. Moreover, the steps performing the above-described series of processes can naturally be executed in the order described, but are not necessarily required to be executed in chronological order; some steps can be executed in parallel or independently of each other. Those skilled in the art will understand that all or any step or component of the method and apparatus of the present invention can be implemented in any computing device (including processors, storage media, etc.) or network of computing devices, in hardware, firmware, software, or a combination thereof. This is something that those skilled in the art can achieve by using their basic programming skills after reading the description of the present invention.

[0142] Therefore, the object of the present invention can also be achieved by running a program or a set of programs on any computing device. The computing device can be a known general-purpose device. Therefore, the object of the present invention can also be achieved simply by providing a program product containing program code for implementing the method or apparatus. That is, such a program product also constitutes the present invention, and the storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any known storage medium or any storage medium developed in the future. It should also be noted that in the apparatus and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent to the present invention. Furthermore, the steps for performing the above series of processes can naturally be performed in the order described, but are not necessarily required to be performed in chronological order. Some steps can be performed in parallel or independently of each other.

[0143] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method of determining extrusion parameters for a metal profile, characterized in that, The method comprises the following steps: obtaining target characteristic parameters of a target metal profile and corresponding historical process parameters in a historical extrusion process, the historical process parameters comprising historical models of a plurality of metal profiles and historical characteristic parameters, historical extrusion parameters and historical extrusion results corresponding to the historical models in a one-to-one manner; determining a knowledge graph of a metal profile extrusion process according to the historical process parameters, the knowledge graph being a relationship structure diagram among the historical models, the historical characteristic parameters, the historical extrusion parameters and the historical extrusion results; determining first extrusion parameters of the target metal profile according to the target characteristic parameters and the knowledge graph; inputting the target characteristic parameters into a trained metal profile extrusion parameter prediction model for prediction to determine second extrusion parameters of the target metal profile, the metal profile extrusion parameter prediction model being obtained by training a preset neural network based on an improved zebra optimization algorithm and the historical process parameters; determining target extrusion parameters corresponding to the target metal profile according to the first extrusion parameters and the second extrusion parameters, wherein the target characteristic parameters and the historical characteristic parameters both comprise a cross-sectional perimeter L, a cross-sectional area A, a perimeter-to-area ratio R and a profile circumscribed circle diameter of the metal profile; the first extrusion parameters, the second extrusion parameters, the target extrusion parameters and the historical extrusion parameters all comprise an extrusion speed, a profile preheating temperature, a die preheating temperature and an extrusion cylinder preheating temperature; and the historical extrusion results comprise an extrusion energy consumption and an outlet cross-sectional temperature standard deviation.

2. The method of determining extrusion parameters for a metal profile according to claim 1, characterized in that, According to the historical process parameters, the knowledge graph of the metal profile extrusion process is determined, comprising: determining an ontology framework of the knowledge graph according to the historical models, the historical characteristic parameters, the historical extrusion parameters and the historical extrusion results; determining the association relationship among entities in the ontology framework according to the structured historical process parameters corresponding to the historical models, the historical characteristic parameters, the historical extrusion parameters and the historical extrusion results; storing the ontology framework, the entities and the association relationship among entities in a preset graph database to determine the knowledge graph corresponding to the metal profile extrusion process.

3. The method of determining extrusion parameters for a metal profile according to claim 2, characterized in that, According to the historical models, the historical characteristic parameters, the historical extrusion parameters and the historical extrusion results, the ontology framework of the knowledge graph is determined, comprising: determining profile model entities, characteristic parameter entities, extrusion parameter entities and extrusion result entities according to the historical models, the historical characteristic parameters, the historical extrusion parameters and the historical extrusion results; sequentially arranging the profile model entities, the characteristic parameter entities, the extrusion parameter entities and the extrusion result entities from top to bottom according to a preset hierarchical structure sequence to determine the ontology framework.

4. The method of determining extrusion parameters for a metal profile according to claim 1, characterized in that, According to the target characteristic parameters and the knowledge graph, the first extrusion parameters of the target metal profile are determined, comprising: determining the similarity between the target characteristic parameters and the historical characteristic parameters corresponding to each historical model in the knowledge graph; According to the similarity, a candidate historical model in the knowledge graph is confirmed, and a first extrusion parameter is determined according to a historical extrusion parameter corresponding to the candidate historical model.

5. The method of determining extrusion parameters for a metal profile according to claim 4, characterized in that, The similarity between the target feature parameter and a historical feature parameter corresponding to each historical model in the knowledge graph is determined, including: The similarity is determined by the following formula: ； wherein, represents the similarity between the jth historical existing metal profile in the knowledge graph and the target metal profile characteristic parameter, j = 1, 2, …, 3, m; represents the ith target characteristic parameter of the target metal profile, represents the ith historical characteristic parameter of the jth historical existing metal profile in the knowledge graph; n represents the type of characteristic parameter.

6. The method of determining extrusion parameters for a metal profile according to claim 1, characterized in that, The metal profile extrusion parameter prediction model is obtained by training the preset neural network based on the improved zebra optimization algorithm and the historical process parameters, including: An initial parameter population of the preset neural network is determined; the initial parameter population includes multiple groups of parameters, each group of parameters includes an initial weight and an initial threshold parameter, and the initial parameter population corresponds to an initial individual population including multiple zebra individuals, and each group of parameters corresponds to the position of each zebra individual one by one; According to the historical process parameters, each group of parameters in the initial parameter population, and the preset neural network, a first fitness value of each zebra individual in the initial individual population is determined; According to the first fitness value and a preset optimization strategy, the positions of the zebra individuals in the initial individual population are iteratively updated to determine the target positions of the zebra individuals in the initial individual population. The parameter values corresponding to the target positions are determined as the target parameters of the preset neural network, and the preset neural network is iteratively trained according to the target parameters and the historical process parameters until a preset convergence condition or a first preset iteration number is reached to obtain the metal profile extrusion parameter prediction model.

7. The method of determining extrusion parameters for a metal profile according to claim 6, characterized in that, According to the first fitness value and a preset optimization strategy, the positions of the zebra individuals in the initial individual population are iteratively updated to determine the target positions of the zebra individuals in the initial individual population, including: According to the first fitness value, a current optimal position in the initial individual population and a first fitness value corresponding to the current optimal position are determined; According to a preset Levy flight exploration strategy, a preset nonlinear convergence factor, and the optimal position, the positions of the zebra individuals in the initial individual population are iteratively updated until a second preset iteration number is reached to obtain the target positions.

8. The method of determining extrusion parameters of a metal profile according to claim 1, characterized in that, According to the first extrusion parameter and the second extrusion parameter, a target extrusion parameter corresponding to the target metal profile is determined, including: According to a preset weight value, the first extrusion parameter and the second extrusion parameter are weighted and fused to obtain the target extrusion parameter.

9. A device for determining extrusion parameters of a metal profile, characterized in that, Including: An acquisition module is configured to acquire a target feature parameter of a target metal profile and a corresponding historical process parameter in a historical extrusion process, the historical process parameter including historical models of multiple metal profiles and historical feature parameters, historical extrusion parameters, and historical extrusion results corresponding to the historical models one by one; The processing module is configured to determine a knowledge graph of the metal profile extrusion process according to the historical process parameters, the knowledge graph being a relationship structure diagram among the historical model, the historical characteristic parameters, the historical extrusion parameters, and the historical extrusion results; determine first extrusion parameters of the target metal profile according to the target characteristic parameters and the knowledge graph; input the target characteristic parameters into a trained metal profile extrusion parameter prediction model for prediction to determine second extrusion parameters of the target metal profile; the metal profile extrusion parameter prediction model is obtained by training a preset neural network based on an improved zebra optimization algorithm and the historical process parameters; and determine target extrusion parameters corresponding to the target metal profile according to the first extrusion parameters and the second extrusion parameters, wherein the target characteristic parameters and the historical characteristic parameters each include a cross-sectional perimeter L, a cross-sectional area A, a perimeter-to-area ratio R, and a profile circumscribed circle diameter of the metal profile; the first extrusion parameters, the second extrusion parameters, the target extrusion parameters, and the historical extrusion parameters each include an extrusion speed, a profile preheating temperature, a die preheating temperature, and an extrusion cylinder preheating temperature; and the historical extrusion results include an extrusion energy consumption and an outlet cross-sectional temperature standard deviation.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a program which, when executed by the processor, implements the method of any one of claims 1 to 8.

Images