Dimension measurement control method for steel billet forging

Abstract

The invention discloses a size measurement control method for steel billet forging, and belongs to the technical field of industrial visual intelligence. The method comprises the following steps: acquiring material information, three-dimensional point cloud and temperature information of a target steel billet; based on the three-dimensional point cloud, performing three-dimensional modeling on the target steel billet to determine size information of the target steel billet; under the condition that it is determined that the temperature information meets the forging condition, the to-be-machined position of the target steel billet is determined according to the size information; the material information and the temperature information corresponding to the to-be-machined position are input into a machining prediction model, the forging force, corresponding to the to-be-machined position, output by the machining prediction model is obtained, and the machining prediction model is obtained based on a data set comprising a plurality of training samples; and based on the to-be-machined position and the forging force, parameter configuration is conducted on the forging machine, so that the forging machine is controlled to forge the to-be-machined position of the target steel billet with the forging force. The method improves the forging precision and the forging effect of the steel billet.

Description

Technical Field

[0001] The present application belongs to the field of industrial visual intelligence technology, and in particular, relates to a method for measuring and controlling the size of a steel billet forging. Background Art

[0002] Billet forging is an important production process in the metallurgical industry, which is mainly used to transform crude steel billets into steel products with specific shapes, sizes and mechanical properties through heating and forging processes. The forging process usually includes heating the billet to a certain temperature to achieve good plasticity, and then forging it through a forging machine to make the billet meet the design requirements in shape, while improving its internal density and mechanical properties. The resulting billet can be processed into parts through processes such as turning and grinding, and is widely used in high-demand industries such as aviation, automobiles, and energy.

[0003] At present, the billet forging process usually relies on experience to set the process parameters, which include the heating temperature of the billet, the forging force, the working speed of the forging machine, the shape of the die, and how the billet is positioned in the die during the forging process. The determination of process parameters mostly relies on the experience and historical data of engineers, and is adjusted according to the standardized process of specific steel grades, shapes and sizes.

[0004] However, the force and processing position required for the forging process of steel billets of different steel grades and shapes may vary greatly. The standardized process cannot flexibly adapt to these changes, resulting in problems such as insufficient precision and poor forging effects. Summary of the invention

[0005] The present application aims to solve at least one of the technical problems existing in the prior art. To this end, the present application proposes a size measurement and control method for steel billet forging, which improves the forging accuracy and forging effect of the steel billet.

[0006] In a first aspect, the present application provides a method for measuring and controlling the size of a steel billet forging, the method comprising:

[0007] Obtain material information, 3D point cloud and temperature information of the target steel billet;

[0008] Based on the three-dimensional point cloud, three-dimensional modeling is performed on the target steel billet to determine size information of the target steel billet;

[0009] When it is determined that the temperature information satisfies the forging conditions, determining the position of the target steel billet to be processed according to the size information;

[0010] Inputting the material information and the temperature information corresponding to the position to be processed into a processing prediction model to obtain the forging force corresponding to the position to be processed output by the processing prediction model, wherein the processing prediction model is obtained based on a data set including a plurality of training samples;

[0011] Based on the position to be processed and the forging force, parameters of the forging machine are configured to control the forging machine to forge the position to be processed of the target steel billet with the forging force.

[0012] According to one embodiment of the present application, the process detection model is constructed based on a feedforward neural network, and the processing prediction model includes an input layer, a hidden layer, and an output layer connected in sequence. The material information and the temperature information corresponding to the position to be processed are input into the processing prediction model to obtain the forging force corresponding to the position to be processed output by the processing prediction model, including:

[0013] The material information and the temperature information are feature mapped and converted through the input layer to obtain a comprehensive feature vector;

[0014] Extracting and predicting the comprehensive feature vector through the hidden layer to obtain a force vector, wherein the hidden layer is constructed based on a multi-head attention mechanism;

[0015] The forging force vector is mapped through the output layer to obtain the forging force and output it.

[0016] According to one embodiment of the present application, the input layer includes a design layer, a fully connected layer, and a subsequent layer connected in sequence;

[0017] The temperature information and the material information are normalized and mapped to a vector space through a design layer to obtain a temperature vector representing the temperature data distribution of the position to be processed and a material vector corresponding to the material information, and the temperature vector and the material vector are spliced ​​to obtain a spliced ​​vector;

[0018] Calling a nonlinear activation function through a fully connected layer to perform a nonlinear transformation on the concatenated vector to obtain a transformation vector;

[0019] The transformation vector is linearly transformed and dimensionally reduced through subsequent layers to obtain the comprehensive feature vector.

[0020] According to one embodiment of the present application, the nonlinear activation function is:

[0021] Mish(x)=min(x·tanh(α·ln(1+e βx )), C)

[0022] Among them, x is the concatenated vector of the input of the fully connected layer; α and β are learnable parameters; tanh is the hyperbolic tangent function, and C is a preset constant.

[0023] According to one embodiment of the present application, before inputting the material information and the temperature information corresponding to the position to be processed into the processing prediction model, the method further includes:

[0024] Obtaining sample temperature information and sample material information of a plurality of sample steel billets, as well as sample forging forces;

[0025] According to the sample sequences of the plurality of sample steel billets, the sample temperature information, the sample material information and the sample forging force are sorted to obtain a first temperature array, a first material array and a first forging force array;

[0026] Based on the interpolation points of the sample sequence, synchronous interpolation calculations are performed on the first temperature array, the first material array and the first forging intensity array to obtain a second temperature array, a second material array and a second forging intensity array;

[0027] Performing linear dimensionality reduction on the second material array, and performing nonlinear dimensionality reduction on the linearly reduced second material array;

[0028] Verify the second temperature array, the second material array with nonlinear dimensionality reduction, and the second forging intensity array respectively, and if the verification passes, construct a data set for the second temperature array, the second material array, and the second forging intensity array;

[0029] The training samples in the data set are input into the feedforward neural network for training in sequence to obtain the processing prediction model.

[0030] According to one embodiment of the present application, the training samples in the data set are sequentially input into a feedforward neural network for training to obtain a processing prediction model, including:

[0031] Extracting a plurality of first training samples from the data set;

[0032] Multiple first training samples are sequentially input into the feedforward neural network, and each hidden layer in the feedforward neural network calculates the activation value according to the output of the previous layer:

[0033]

[0034] Among them, h i is the output of the i-th hidden layer neuron, w ij is the weight of the neurons from the jth layer to the ith hidden layer, the jth layer is the input layer or the previous layer, b i is the bias of the i-th hidden layer neuron, f(·) is the activation function, x j is the output of the jth layer, which serves as the input of the i-th hidden layer neuron;

[0035] Obtain the prediction result output by the feedforward neural network, and calculate the loss value corresponding to the prediction result;

[0036] Using the chain rule, the backpropagation algorithm calculates the gradient of the loss function with respect to the hidden layer weights and biases:

[0037]

[0038] in, is the gradient of the loss function L with respect to the output of the hidden layer;

[0039] According to the calculated gradient, update the weights and bias of the hidden layer:

[0040]

[0041]

[0042] Where μ is the learning rate;

[0043] Repeatedly extract multiple second training samples from the data set, and input the multiple second training samples into the feedforward neural network in sequence until the obtained loss value is less than the loss threshold, the feedforward neural network training is completed, and the processing prediction model is obtained.

[0044] According to an embodiment of the present application, the forging machine is configured with parameters based on the position to be processed and the forging force to control the forging machine to forge the position to be processed of the target steel billet with the forging force, including:

[0045] According to the forging force, configure the working frequency of the air hammer in the forging machine and the force of a single blow;

[0046] According to the position to be processed, configuring the moving trajectory of the air hammer in the forging machine;

[0047] Based on the position to be processed and the moving trajectory, the air hammer is globally positioned by a photoelectric sensor, and the air hammer is locally positioned by a laser sensor;

[0048] Based on the working frequency and the force of a single blow, forging the target steel billet by the positioned air hammer;

[0049] According to the comparison between the forging position of the air hammer and the position to be processed, the position of the air hammer is corrected;

[0050] The operating frequency and the force of a single blow are adjusted according to the deformation degree and temperature change of the target steel billet.

[0051] According to one embodiment of the present application, after configuring parameters of a forging machine based on the position to be processed and the forging force to control the forging machine to forge the position to be processed of the target steel billet with the forging force, the method further includes:

[0052] After each forging, the target steel billet is subjected to temperature and deformation testing;

[0053] When it is determined that the temperature of the target steel billet is lower than the lower temperature limit of the forging condition, heating the target steel billet by a heating furnace until the temperature of the target steel billet is higher than the upper temperature limit of the forging condition;

[0054] The three-dimensional point cloud and temperature information of the target steel billet are reacquired to determine the position to be processed and predict the forging force in combination with the material information, and the forging machine is controlled to forge the target steel billet until it is determined according to the deformation detection that the size difference between the target steel billet and the standard steel billet template is within a preset error range, thereby obtaining the forged target steel billet.

[0055] According to one embodiment of the present application, after obtaining the material information, the three-dimensional point cloud and the temperature information of the target steel billet, the method further includes:

[0056] When it is determined that the temperature information is lower than the lower limit of the temperature of the forging condition, heating the target steel billet by a heating furnace and detecting a temperature change of the target steel billet;

[0057] Based on the temperature change, the furnace environment and real-time power of the heating furnace, predicting the temperature change trend of the target steel billet through data modeling;

[0058] Based on the temperature change and the temperature change trend, the power of the heating furnace is adjusted until the detected temperature of the target steel billet reaches the upper temperature limit of the forging condition.

[0059] According to one embodiment of the present application, based on the temperature change and the temperature change trend, the power of the heating furnace is adjusted until the detected temperature of the target steel billet reaches the upper temperature limit of the forging condition, including:

[0060] When it is determined according to the temperature change that the detected temperature of the target steel billet is lower than the lower limit of the temperature, heating the target steel billet with the initial power by the heating furnace;

[0061] When it is determined according to the temperature change that the detected temperature of the target steel billet is not lower than the lower temperature limit, the power of the heating furnace is adjusted based on the temperature change trend and the initial power until the detected temperature of the target steel billet reaches the upper temperature limit of the forging conditions.

[0062] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application.

[0063] The present invention provides a method for measuring and controlling the size of a steel billet forging, which has the following beneficial effects compared with the prior art:

[0064] (1) By real-time three-dimensional modeling of the target steel billet, the dimensional detection accuracy is improved, the real-time monitoring and adjustment of the steel billet temperature are responded to in a timely manner, and the operating parameters of the forging machine are automatically adjusted according to the output of the processing prediction model. The different forging forces and processing positions required for steel billets of different steel grades and shapes during the forging process can be flexibly adjusted and accurately controlled, with a wider range of applications, and the precise control of the size of the target steel billet can be achieved, ensuring the stability of the quality of the target steel billet, improving production efficiency, as well as the forging accuracy and forging effect of the target steel billet, reducing the defective rate, saving materials in the processing stage of the steel billet, and reducing production costs.

[0065] (2) By performing temperature detection and deformation detection after each forging, the forging force and heating process are dynamically adjusted to form a closed-loop control system to achieve precise temperature and deformation control to ensure that the size and quality of the target steel billet meet the standards, improve the processing efficiency of the target steel ingot, and effectively control the dimensional error. The obtained forged target steel billet is close to the standard steel billet template, which is convenient for subsequent processing, reduces the waste generated by subsequent processing, and saves materials and costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0067] Figure 1 This is one of the flow charts of the size measurement and control method for steel billet forging provided in the embodiment of the present application;

[0068] Figure 2 This is the second flow chart of the dimension measurement and control method for steel billet forging provided in the embodiment of the present application;

[0069] Figure 3 This is the third flow chart of the size measurement control method for steel billet forging provided in the embodiment of the present application;

[0070] Figure 4 This is a fourth flow chart of the size measurement control method for steel billet forging provided in an embodiment of the present application;

[0071] Figure 5 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0072] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application belong to the scope of protection of this application.

[0073] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0074] In combination with the accompanying drawings, a size measurement control method for steel billet forging, a size measurement control device for steel billet forging, an electronic device and a readable storage medium provided in an embodiment of the present application are described in detail through specific embodiments and their application scenarios.

[0075] Among them, the size measurement and control method of steel billet forging can be applied to a terminal, and can be specifically executed by hardware or software in the terminal.

[0076] The terminal includes, but is not limited to, a portable communication device such as a mobile phone or tablet computer with a touch-sensitive surface (e.g., a touch screen display and / or a touch pad). It should also be understood that in some embodiments, the terminal may not be a portable communication device, but a desktop computer with a touch-sensitive surface (e.g., a touch screen display and / or a touch pad).

[0077] In the following various embodiments, a terminal including a display and a touch-sensitive surface is described. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, a mouse and a joystick.

[0078] An embodiment of the present application provides a size measurement and control method for steel billet forging. The execution subject of the size measurement and control method for steel billet forging may be an electronic device or a functional module or functional entity in the electronic device that can implement the size measurement and control method for steel billet forging. The electronic devices mentioned in the embodiment of the present application include but are not limited to mobile phones, tablet computers, computers, cameras, and wearable devices. The following describes a size measurement and control method for steel billet forging provided in the embodiment of the present application by taking an electronic device as an execution subject.

[0079] like Figure 1 As shown, the size measurement and control method of the steel billet forging includes:

[0080] Step 110, obtaining material information, three-dimensional point cloud and temperature information of the target steel billet;

[0081] In the billet processing workshop, the production line uses intelligent forging machines to forge steel billets to produce steel parts needed for industrial equipment manufacturing.

[0082] It is understandable that information such as the size, shape, material composition, and temperature of the target billet has an important influence on the processing position and force of forging, which can improve the mechanical properties and dimensional accuracy.

[0083] Among them, material information includes chemical composition such as carbon content and alloying elements of the target steel billet, as well as physical properties such as density, thermal conductivity, expansion coefficient, yield strength and hardness data.

[0084] In this step, a chemical analysis instrument such as an X-ray fluorescence analyzer, a spectrum analyzer, or laser-induced breakdown spectroscopy (LIBS) is used to determine the chemical composition of the steel billet.

[0085] When the material of the target steel billet is a standard steel grade, the corresponding material information can be queried from the database of the steel plant or related literature.

[0086] The target steel billet is scanned by 3D scanners such as laser scanners and structured light scanners to obtain accurate point cloud data. The point cloud data is filtered and registered. The generated three-dimensional point cloud can represent the complete three-dimensional surface shape of the target steel billet, including its length, width, height and detailed information of surface morphology.

[0087] Before the target steel billet enters the forging machine, the surface temperature of the target steel billet is monitored in real time by multiple infrared temperature sensors and thermal imaging cameras to obtain temperature information that can characterize the surface temperature distribution of the target steel billet.

[0088] Step 120: Based on the three-dimensional point cloud, three-dimensional modeling is performed on the target steel billet to determine size information of the target steel billet.

[0089] In this step, a 3D mesh model is generated from the 3D point cloud by calling 3D modeling software using 3D reconstruction algorithms such as Poisson reconstruction and Delaunay triangulation to obtain a 3D model of the target billet. The size information of the target billet is obtained by measuring the length, width, height, diameter and other dimensions of the 3D model, and the volume of the target billet is calculated.

[0090] Among them, the 3D modeling software can be AutoCAD, SolidWorks, etc.

[0091] Step 130: When it is determined that the temperature information satisfies the forging conditions, the position of the target steel billet to be processed is determined according to the size information.

[0092] The forging temperature range is defined according to the material type of the target steel billet, such as steel grade, alloy, etc., including the lower temperature limit of the forging conditions to the upper temperature limit of the forging conditions. The forging temperature range can be adjusted according to the type of steel and processing requirements. For example, if the forging temperature of the steel is between 800°C and 1300°C, the forging temperature range can be set at 800°C-1300°C, the lower temperature limit of the forging conditions is 800°C, and the upper temperature limit of the forging conditions is 1300°C.

[0093] In this step, the measured local temperature of the target steel billet is compared with the forging temperature range based on the temperature information. When the temperature values ​​corresponding to all local temperatures are within the forging temperature range, the temperature information meets the forging conditions. When there is a local temperature exceeding the forging temperature range, the target steel billet needs to be heated or cooled.

[0094] According to the size difference between the size information and the standard billet template, the position to be forged is determined as the position to be processed. For example, the area where the size difference exceeds the preset error range may need to be forged more, as the position to be processed, while the area where the size difference is within the preset error range needs to avoid excessive forging to ensure that the position to be processed will not be excessively deformed or thermally damaged during the forging process.

[0095] Step 140: input the material information and the temperature information corresponding to the position to be processed into a processing prediction model to obtain the forging force corresponding to the position to be processed output by the processing prediction model, wherein the processing prediction model is obtained based on a data set including a plurality of training samples;

[0096] The temperature information corresponding to the position to be processed carries the three-dimensional coordinates of the position to be processed, which is characterized by the distribution of the temperature value at the position to be processed, and the forging force also carries the three-dimensional coordinates of the position to be processed.

[0097] In this step, sample temperature information and sample material information corresponding to sample processing positions of multiple sample steel billets, as well as corresponding sample forging forces are collected. These data can be obtained through experimental measurement, simulation or actual processing data.

[0098] The sample temperature information and sample material information corresponding to the sample processing position of the sample steel billet are used as sample data, and the sample forging force is used as the label of the sample data to obtain the training sample corresponding to each sample steel billet, and construct a data set for multiple training samples.

[0099] Based on multiple training samples in the data set, a neural network is trained through a machine learning algorithm, and the trained neural network is used as a processing prediction model. The obtained processing prediction model can predict the forging force corresponding to the position to be processed based on the input temperature information and material information.

[0100] For example, there are slight defect areas at both ends of the target billet, which need to be forged intensively. Since the material of the target billet is relatively hard, the forging force is set to 1500kN, and a higher forging force needs to be applied to the two end areas to ensure uniform deformation of the target billet.

[0101] Step 150: Based on the position to be processed and the forging force, configure parameters of the forging machine to control the forging machine to forge the position to be processed of the target steel billet with the forging force.

[0102] The forging force is connected to the control system of the forging machine, and the parameters such as pressure, speed, position and force of the forging machine are configured to ensure that the air hammer properly forges the processing position of the target steel billet. The model of the air hammer is selected according to the area size of the processing position and the forging force.

[0103] In addition, during the forging process of the target billet by the air hammer in the forging machine, the temperature and deformation of the billet are monitored in real time to ensure that the forging process meets the predetermined process requirements. The parameters of the forging machine can also be dynamically adjusted according to the sensor data (such as temperature, force, deformation, etc.) to deal with any emergencies or errors.

[0104] Cooling shrinkage can also be considered, and the size of the target steel billet after forging can be inspected to ensure that the size difference between the target steel billet and the standard steel billet template is within the preset error range, and to ensure that a complete standard steel billet template can be cut from the target steel billet. The forged target steel billet enters the cooling area, and the temperature gradually decreases. The surface quality of the target steel billet is checked to ensure that there are no defects such as cracks and pores. According to the quality inspection results of the forged target steel billet, the process parameters of the forging machine are adjusted to optimize the forging process.

[0105] According to a size measurement and control method for steel billet forging provided in an embodiment of the present application, the size detection accuracy is improved by real-time three-dimensional modeling of the target steel billet, and the real-time monitoring and adjustment of the steel billet temperature are promptly responded to. According to the output of the processing prediction model, the operating parameters of the forging machine are automatically adjusted. The different forging forces and processing positions required for steel billets of different steel grades and shapes during the forging process can be flexibly adjusted and accurately controlled. The scope of application is wider, and accurate control of the size of the target steel billet is achieved, ensuring the stability of the quality of the target steel billet, improving production efficiency, as well as the forging accuracy and forging effect of the target steel billet, reducing the defective rate, saving materials in the processing stage of the steel billet, and reducing production costs.

[0106] In some embodiments, Figure 2 As shown, the processing prediction model is constructed based on a feedforward neural network, and the processing prediction model includes an input layer, a hidden layer, and an output layer connected in sequence. The material information and the temperature information corresponding to the position to be processed are input into the processing prediction model to obtain the forging force corresponding to the position to be processed output by the processing prediction model, including:

[0107] Step 210: feature mapping and converting the material information and the temperature information through the input layer to obtain a comprehensive feature vector;

[0108] Step 220: extracting and predicting the comprehensive feature vector through the hidden layer to obtain a force vector, wherein the hidden layer is constructed based on a multi-head attention mechanism;

[0109] Step 230: Map the forging force vector through the output layer to obtain the forging force and output it.

[0110] The processing prediction model is built based on FNN. The architecture of FNN is relatively simple, easy to understand and implement. It has strong processing capabilities for multi-dimensional input data such as material information and temperature, and can improve the generalization ability of the processing prediction model. Based on the type of input data and the complexity of the task, and there is no timing problem in the prediction of the forging force of the target billet, the Feedforward Neural Network (FNN) has obvious applicability in the scenarios of billet geometry and surface feature analysis, and FNN can have better performance.

[0111] The input layer receives material information and temperature information as input features, including the type and composition of the material, and the temperature distribution of the target billet.

[0112] In the input layer, some linear or nonlinear mappings (such as fully connected layers or embedding layers) are used to map and transform the input features into a comprehensive feature vector to provide suitable input for the next step of calculation.

[0113] The hidden layer contains multiple neurons and uses activation functions (such as ReLU, Sigmoid, Tanh, etc.) to increase nonlinearity, allowing the network to learn complex features and relationships.

[0114] In the hidden layer, the comprehensive feature vector will undergo further feature extraction, and the multi-head attention layer will process the comprehensive features with an attention mechanism to extract deep patterns and relationships. Through the calculation of the hidden layer, FNN will learn how to extract features related to forging force from material information and temperature information and form a force vector.

[0115] Among them, the multi-head attention layer generates the query vector Q, key vector K and value vector V from the comprehensive feature vector X:

[0116] Q=XW Q , K = XW K , V = XW V

[0117] Among them, W Q , W K , W V The weight matrix obtained after training.

[0118] The similarity is measured using Scaled Dot-Product Attention to calculate the attention weight:

[0119]

[0120] Among them, d k is the dimension of the key vector K, It is a scaling factor used to scale the dot product value to avoid gradient explosion caused by too large a value.

[0121] The interaction between temperature and material information is modeled. Each head pays attention to different subspaces of the comprehensive feature vector X in parallel, and generates an independent output representation. These outputs are spliced ​​into a spliced ​​feature vector, and the spliced ​​feature vector is mapped to a new feature space through a linear layer to generate a fused comprehensive feature representation. A multi-layer perceptron (MLP) is used to perform a linear transformation on the fused feature representation in the fully connected layer to map it to the final force prediction space to obtain a predicted force vector, which can be a continuous numerical value representing the magnitude of the force. By combining the multi-head attention mechanism and MLP to process the complex interaction between temperature and material, high-quality feature representations are generated and force predictions are made. Through multi-head attention, the processing prediction model can pay attention to the input data from multiple angles and generate rich feature representations. MLP further performs nonlinear transformations to improve prediction accuracy.

[0122] The output layer maps the internal representation (force vector) to the actual forging force. The forging force can be a scalar (a single value) or multiple values. The activation function is used to ensure that the range of the output value is consistent with the problem requirements. The final forging force corresponding to the position to be processed on the target billet is output to control or adjust the forging process.

[0123] In this embodiment, by inputting material information and temperature information into the processing prediction model for feature extraction, prediction and output, and performing nonlinear transformation on the input features through the hidden layer, the forging force under different conditions can be accurately predicted.

[0124] In some embodiments, Figure 3 As shown, the input layer includes a design layer, a fully connected layer and a subsequent layer connected in sequence;

[0125] Step 310: normalize the temperature information and material information through the design layer and map them to the vector space to obtain a temperature vector representing the temperature data distribution of the position to be processed and a material vector corresponding to the material information, and splice the temperature vector and the material vector to obtain a spliced ​​vector;

[0126] Step 320, calling a nonlinear activation function through a fully connected layer to perform a nonlinear transformation on the concatenated vector to obtain a transformation vector;

[0127] Step 330: Perform linear transformation and dimension reduction on the transformation vector through subsequent layers to obtain the comprehensive feature vector.

[0128] In actual implementation, the design layer normalizes the temperature information and material information, that is, standardizes the temperature information and material information to a certain range (for example, between 0 and 1), which helps to eliminate the differences in different input feature scales and makes subsequent processing more stable.

[0129] At the same time, the design layer also maps this information into the vector space, converts the temperature information into a vector representation, obtains the temperature vectors, and uses the neural network embedding technology to map the content and physical property data of each alloy element in the material information into a dense vector of a fixed length, and splices the embedded vectors of all elements together to obtain the material vector.

[0130] Among them, the material vector can be a low-dimensional dense vector, which effectively represents the complex material properties as a low-dimensional dense vector, thereby providing support for further analysis and prediction.

[0131] The design layer concatenates the temperature vector and the material vector to form a concatenated vector that contains the features of temperature data and material information. By combining different feature information together, a new and richer feature representation is formed.

[0132] The concatenated vectors enter the fully connected layer, and are weighted and summed through the weight matrix and nonlinearly transformed by applying a nonlinear activation function to generate a new vector, the transformed vector. The fully connected layer can effectively learn the complex relationship of the input data and help the network extract more abstract features.

[0133] The transformed vector enters the subsequent layer, which multiplies the transformed vector with the weight matrix to perform a linear transformation, thereby changing the representation of the vector. After the linear transformation, dimensionality reduction is performed through methods such as principal component analysis (PCA), which can reduce the dimension of the vector, extract the most important features, and obtain a comprehensive feature vector.

[0134] In this embodiment, the comprehensive feature vector is the feature representation that best represents the temperature and material information after multiple transformations and dimensionality reduction processes, providing a basis for subsequent efficient processing.

[0135] In some embodiments, the nonlinear activation function is:

[0136] Mish(x)=min(x·tanh(α·ln(1+e βx )), C)

[0137] Among them, x is the concatenated vector of the input of the fully connected layer; α and β are learnable parameters; tanh is the hyperbolic tangent function, and C is a preset constant.

[0138] α and β can dynamically adjust the performance of the activation function according to different inputs.

[0139] Tanh is a smooth S-shaped function. By further adjusting the output nonlinearly, it becomes smooth and continuous, with better gradient flow and training performance. By presetting the constant C, the maximum output value can be controlled, effectively avoiding excessive output values, thereby avoiding the problem of gradient explosion and reducing the fluctuation of weight updates. FNN can converge more smoothly during training. The output is more symmetrical and balanced, with the center point at 0, avoiding the defect that the ReLU activation function is only active on the positive semi-axis.

[0140] Compare the input x with tanh(α·ln(1+e βx ), the coefficient x is used to adjust the influence to ensure that even if the input is in the negative region, there is still a gradient and the information can be effectively transmitted, avoiding the defect of dead neurons in the ReLU activation function. In the positive input region, the multiplication operation retains the advantage of the positive part and does not have the defect of compressed output like the sigmoid function or the tanh function, and can better maintain the changes of large inputs.

[0141] In some embodiments, Figure 4As shown, before inputting the material information and the temperature information corresponding to the position to be processed into the processing prediction model, the method further includes:

[0142] Step 410, obtaining sample temperature information and sample material information of a plurality of sample steel billets, as well as sample forging forces;

[0143] Step 420: sort the sample temperature information, the sample material information and the sample forging force according to the sample sequences of the plurality of sample steel billets to obtain a first temperature array, a first material array and a first forging force array;

[0144] Step 430: Based on the interpolation points of the sample sequence, synchronous interpolation calculations are performed on the first temperature array, the first material array, and the first forging intensity array to obtain a second temperature array, a second material array, and a second forging intensity array;

[0145] Step 440: perform linear dimensionality reduction on the second material array, and perform nonlinear dimensionality reduction on the linearly reduced second material array;

[0146] Step 450, respectively verifying the second temperature array, the second material array with nonlinear dimensionality reduction, and the second forging intensity array, and if the verification passes, constructing a data set for the second temperature array, the second material array, and the second forging intensity array;

[0147] Step 460: Input the training samples in the data set into the feedforward neural network for training in sequence to obtain a processing prediction model.

[0148] The data collected from multiple sample billets can be obtained through experimental measurement, simulation or actual processing data.

[0149] The material information includes multiple material properties of the sample steel billet, and the material information can be converted into a multi-dimensional vector, and each component can be linearly interpolated independently.

[0150] According to the sample sequence [1,2,...,n] of the multiple sample steel billets, the sample temperature information, the sample material information and the sample forging intensity are sorted, and the first temperature array T=[T1,T2,...,T n ], the first material array M=[(M 1,1 ,M 1,2 ,...,M 1,n ),(M 2,1 ,M 2,2 ,...,M 2,n ),...,(M m,1 ,M m,2 ,...,M m,n)], the first material array corresponds to the m-dimensional attribute of the sample material information, the first forging intensity array F=[F1,F2,...,F n ];

[0151] When an interpolation point is selected in the interval [i,i+1],i∈[1,2,...,n-1] of the sample sequence, and [T i ,T i+1 ]、[M j,i ,M j,i+1 ]、[F i ,F i+1 ] is calculated using the same interpolation formula.

[0152] Interpolate the first temperature array:

[0153]

[0154] Interpolate the attributes of each dimension in the first material array. For the j-th dimension of the first material array M j =(M j,1 ,M j,2 ,...,M j,n ):

[0155]

[0156] You can get M newi =[M 1,i ,M 2,i ,...,M m,i ];

[0157] Perform interpolation calculation for the first forging intensity array:

[0158]

[0159] Through the above calculation, the first temperature array, the first material array and the first forging force array can be synchronously interpolated and calculated to obtain T newi 、M newi and F newi There is a corresponding relationship between them, which provides a basis for the expansion of training samples.

[0160] After different interpolation calculations, T newi 、M newi and F newi The arrays are respectively inserted into the first temperature array, the first material array and the first forging intensity array to obtain the second temperature array, the second material array and the second forging intensity array.

[0161] The second material array is subjected to preliminary linear dimensionality reduction through principal component analysis (PCA). Linear dimensionality reduction helps to reduce computational complexity, and then t-SNE or UMAP is used for nonlinear dimensionality reduction to retain more structural information in the second material array. The combination of linear and nonlinear dimensionality reduction can maximize the retention of multidimensional information in the data while reducing the data dimension, thus avoiding possible information loss in traditional dimensionality reduction methods.

[0162] The residuals of the second temperature array, the second material array with nonlinear dimensionality reduction, and the second forging force array are calculated and a residual graph is drawn. The error distribution is determined based on the residual graph. When the residuals are close to zero and evenly distributed, the verification is passed.

[0163] According to the data correspondence in the second temperature array, the second material array and the second forging force array, the sample temperature information and sample material information of the sample steel billet are used as sample data, and the sample forging force is used as the label of the sample data to obtain the training sample corresponding to each sample steel billet, and construct a data set for multiple training samples.

[0164] The FNN is trained through multiple training samples, and the trained FNN is used as a processing prediction model.

[0165] In this embodiment, the timing and missing problems of data are solved through sorting and interpolation, which provides more accurate input for subsequent modeling and reduces the difficulty of obtaining sample data. The linear and nonlinear dimensionality reduction are combined to reduce the dimension of the data and reduce the computational burden while retaining sufficient information to enhance the prediction ability of the processing prediction model. The feedforward neural network is used for training, which can capture complex nonlinear relationships and adapt to the variable factors in the billet processing process. The constructed processing prediction model can reflect the production process in real time, provide accurate prediction results, and contribute to the intelligent production and process optimization of the target billet.

[0166] In some embodiments, the training samples in the data set are sequentially input into a feedforward neural network for training to obtain a processing prediction model, including:

[0167] Extracting a plurality of first training samples from the data set;

[0168] Multiple first training samples are sequentially input into the feedforward neural network, and each hidden layer in the feedforward neural network calculates the activation value according to the output of the previous layer:

[0169]

[0170] Among them, h i is the output of the i-th hidden layer neuron, w ijis the weight of the neurons from the jth layer to the ith hidden layer, the jth layer is the input layer or the previous layer, b i is the bias of the i-th hidden layer neuron, f(·) is the activation function, x j is the output of the jth layer, which serves as the input of the i-th hidden layer neuron;

[0171] Obtain the prediction result output by the feedforward neural network, and calculate the loss value corresponding to the prediction result;

[0172] Using the chain rule, the backpropagation algorithm calculates the gradient of the loss function with respect to the hidden layer weights and biases:

[0173]

[0174] in, is the gradient of the loss function L with respect to the output of the hidden layer;

[0175] According to the calculated gradient, update the weights and bias of the hidden layer:

[0176]

[0177] Where μ is the learning rate;

[0178] Repeatedly extract multiple second training samples from the data set, and input the multiple second training samples into the feedforward neural network in sequence until the obtained loss value is less than the loss threshold, the feedforward neural network training is completed, and the processing prediction model is obtained.

[0179] Among them, the update of the bias is used to adjust the parameters according to the learning rate and reduce the loss.

[0180] In this embodiment, through fine-tuning, the weights and biases of the hidden layer will adjust the weights according to the gradient of the loss function, so that the processed prediction can better adapt to the prediction task and achieve good training results even in the case of small samples.

[0181] In some embodiments, the forging machine is configured with parameters based on the position to be processed and the forging force to control the forging machine to forge the position to be processed of the target steel billet with the forging force, including:

[0182] According to the forging force, configure the working frequency of the air hammer in the forging machine and the force of a single blow;

[0183] According to the position to be processed, configuring the moving trajectory of the air hammer in the forging machine;

[0184] Based on the position to be processed and the moving trajectory, the air hammer is globally positioned by a photoelectric sensor, and the air hammer is locally positioned by a laser sensor;

[0185] Based on the working frequency and the force of a single blow, forging the target steel billet by the positioned air hammer;

[0186] According to the comparison between the forging position of the air hammer and the position to be processed, the position of the air hammer is corrected;

[0187] The operating frequency and the force of a single blow are adjusted according to the deformation degree and temperature change of the target steel billet.

[0188] The position to be processed is a designated position or area on the target steel billet, the working frequency is the number of strikes at the position per unit time, and the force of a single strike is the forging force that needs to be applied at the position once.

[0189] Based on the position to be processed and the forging force, the parameters of the forging machine working frequency and the force of a single blow are configured to determine the working frequency and the force of a single blow of the air hammer, especially to optimize the positioning when using the air hammer.

[0190] Configure the moving trajectory of the air hammer in the forging machine to ensure that the air hammer accurately hits the target position. In addition, the air pressure or other control parameters of the air hammer can be adjusted according to the forging force requirements.

[0191] According to the overall shape of the target billet and the position to be processed, the moving trajectory of the air hammer is preliminarily set. In the global positioning stage, a lower-precision photoelectric sensor or laser ranging equipment is used to make larger adjustments to preliminarily determine the striking position of the air hammer and ensure that the air hammer can roughly hit the predetermined area.

[0192] After global positioning, the local positioning stage of the air hammer begins. The air hammer is monitored and precisely positioned in real time through higher-precision sensors or visual systems such as high-precision lasers, industrial cameras, infrared sensors, etc. The air hammer's drop position and force are precisely adjusted to ensure that the forging force of each position to be processed meets the expected requirements.

[0193] During the local positioning stage, the information fed back by the sensor can adjust the position of the air hammer in real time to ensure that each blow reaches the expected force and position.

[0194] During the forging process, the target steel billet will gradually deform and the temperature will gradually decrease as the air hammer is forged. Combined with the real-time deformation degree and temperature changes during the forging process, the air hammer's striking position, operating frequency and single-strike force are dynamically adjusted in real time through closed-loop control to cope with possible deformation or position changes of the steel billet during processing.

[0195] During the forging process, force sensors, position sensors, pressure sensors, etc. are used to monitor the working status of the air hammer in real time. According to the data fed back by the sensor, the force, frequency and position of the air hammer are automatically adjusted to ensure that each forging meets the requirements. This can greatly improve the accuracy of the forging process and ensure that the processing effect of each forging position meets the predetermined standards.

[0196] For example, by establishing a mathematical model or machine learning model related to the forging process, the interaction between the air hammer and the billet can be predicted, so that parameter optimization can be performed. According to the real-time feedback information obtained during the production process, the working parameters of the forging machine can be adaptively adjusted to cope with different materials, working conditions and environmental changes.

[0197] In this embodiment, the forging machine's air hammer is subjected to a combination of coarse adjustment of the global positioning and fine adjustment of the local positioning to ensure that the forging machine can accurately control the forging force and processing position of the air hammer in a short time, ensure that the air hammer accurately strikes the target billet's processing position and applies the correct forging force, thereby improving the forging quality and efficiency. Through feedback adjustment, not only the processing accuracy is improved, but also adaptive adjustment can be made under different working conditions, reducing human errors and equipment wear, and improving production efficiency and forging quality.

[0198] In some embodiments, after configuring parameters of a forging machine based on the position to be processed and the forging force to control the forging machine to forge the position to be processed of the target steel billet with the forging force, the method further includes:

[0199] After each forging, the target steel billet is subjected to temperature and deformation testing;

[0200] When it is determined that the temperature of the target steel billet is lower than the lower temperature limit of the forging condition, heating the target steel billet by a heating furnace until the temperature of the target steel billet is higher than the upper temperature limit of the forging condition;

[0201] The three-dimensional point cloud and temperature information of the target steel billet are reacquired to determine the position to be processed and predict the forging force in combination with the material information, and the forging machine is controlled to forge the target steel billet until it is determined according to the deformation detection that the size difference between the target steel billet and the standard steel billet template is within a preset error range, thereby obtaining the forged target steel billet.

[0202] According to the target steel billet processing position and the expected forging force, the parameters of the forging machine are configured, and the forging force, the working mode and the action path of the forging machine are set as the preliminary conditions of the entire forging process. The forging force includes the magnitude and direction of the force applied by the air hammer.

[0203] After each forging, the target billet is subjected to temperature and deformation testing, and the status of the target billet is monitored in real time to ensure that the temperature and deformation meet expectations.

[0204] If the forging temperature is too low, it will affect the plasticity of the steel and make it difficult to deform; if the temperature is too high, it may cause oxidation damage or lead to uneven force distribution. Through temperature detection, it is ensured that the temperature of the target steel billet meets the forging requirements. Through deformation detection, it is ensured that the deformation of the target steel billet meets the design requirements to avoid excessive or insufficient deformation.

[0205] If it is detected that the temperature of the target steel billet is lower than the lower temperature limit of the forging condition, the target steel billet is heated by the heating furnace, and the heating process needs to be controlled to prevent the temperature of the target steel billet from exceeding the upper temperature limit of the forging condition:

[0206] The temperature control system of the heating furnace should be able to accurately control the temperature of the billet to ensure that the billet has an ideal temperature when entering the next stage of the forging process. The upper and lower temperature limits should be strictly set according to the material and forging requirements of the target billet.

[0207] After each forging, the 3D point cloud and temperature information of the target billet are acquired again, the 3D model of the target billet is re-created, the processing position is determined, the temperature information and material information are re-entered into the processing prediction model, and the force required for the next forging is predicted, so that the target billet is always in the correct shape during the forging process. By reasonably coordinating the forging force with the processing position, uneven force can be avoided to affect the final shape of the target billet.

[0208] The target steel billet after forging is compared with the standard steel billet model, the shape and size errors are calculated, and then fed back to the forging machine to adjust its working parameters. By adjusting the parameters of the forging machine in real time, the error of the steel billet after each forging can be minimized. For example, if the size of a certain part is too large, the forging force of the area can be reduced; if a certain part is too small, the force can be increased. During the forging process, deformation detection is used to ensure that the size error between the target steel billet and the standard steel billet template is within the preset error range.

[0209] Among them, the trend prediction model can be trained through historical data and real-time detection data to predict possible changes in the forging process, and adjust the parameters of the forging machine in advance. Based on the feedback data, the forging force, heating process and deformation detection can be optimized in real time to improve the dimensional accuracy and processing efficiency of the final steel billet.

[0210] In this embodiment, by performing temperature detection and deformation detection after each forging, the forging force and heating process are dynamically adjusted to form a closed-loop control system, and precise temperature and deformation control is achieved to ensure that the size and quality of the target steel billet meet the standards, thereby improving the processing efficiency of the target steel ingot and effectively controlling the dimensional error. The obtained forged target steel billet is close to the standard steel billet template, which is convenient for subsequent processing, reduces waste generated by subsequent processing, and saves materials and costs.

[0211] In some embodiments, after obtaining the material information, the three-dimensional point cloud and the temperature information of the target steel billet, the method further includes:

[0212] When it is determined that the temperature information is lower than the lower limit of the temperature of the forging condition, heating the target steel billet by a heating furnace and detecting a temperature change of the target steel billet;

[0213] Based on the temperature change, the furnace environment and real-time power of the heating furnace, predicting the temperature change trend of the target steel billet through data modeling;

[0214] Based on the temperature change and the temperature change trend, the power of the heating furnace is adjusted until the detected temperature of the target steel billet reaches the upper temperature limit of the forging condition.

[0215] The heating furnace is an electric heating furnace or a gas heating furnace, which can accurately control the power output in the furnace, adjust the temperature in the furnace, and accurately control the heating process.

[0216] The temperature sensor may include an infrared thermometer installed at multiple points in the furnace to collect temperature data of the target steel billet and environmental parameters in the furnace (such as furnace temperature, atmosphere humidity, etc.) in real time.

[0217] The temperature change trend can estimate the temperature value of the target steel billet at a certain time in the future and compare it with the upper temperature limit of the forging conditions.

[0218] During the heating process, the detection temperature of the target steel billet, the furnace environment parameters and the power output data of the current heating furnace are collected in real time. The historical data and real-time monitoring data are entered into the modeling platform through the data acquisition system. The relationship between the steel billet temperature and the furnace environment and power is modeled using machine learning algorithms such as linear regression and neural networks to obtain a trend prediction model to predict the temperature change trend. The trend prediction model predicts the temperature change trend of the target steel billet in the next few minutes by inputting the real-time collected data and generates a temperature change curve.

[0219] According to the temperature change trend predicted by the trend prediction model, the current temperature rise rate is judged. If the temperature rise of the target billet is predicted to be slow and lower than the expected temperature rise, it means that the power output of the heating furnace needs to be increased; if the temperature rise is too fast, the power output needs to be reduced to avoid excessive temperature rise.

[0220] Among them, the heating furnace can be temperature controlled by a PID control system, and the power output of the heating furnace can be automatically adjusted according to the temperature change trend.

[0221] Before entering the heating furnace, it is assumed that the temperature information of the target billet is T initial , and the temperature is lower than the lower limit of the forging condition temperature T lowerlimit , that is, T initial <T lowerlimit , select a suitable heating furnace power output as the initial power, such as 80% of the power, start the heating furnace to heat the target steel billet, and gradually increase the detection temperature of the target steel billet.

[0222] During this process, the temperature changes of the target steel billet are monitored in real time, and the heating strategy is adjusted using real-time data fed back by the temperature sensor. When the temperature of the target steel billet rises slowly, the power output of the heating furnace is increased, for example, adjusted to 90% or 100% to speed up the temperature rise. If the temperature is predicted to rise too quickly, the power output is reduced, possibly adjusted to 60% or 50% of the power, to avoid the target steel billet temperature being too high and exceeding the upper temperature limit of the forging conditions.

[0223] By continuously adjusting the power, the target billet temperature is ensured to gradually approach the upper temperature limit T of the forging condition. upperlimit When the detection temperature is close to the upper temperature limit, the power of the heating furnace is fine-tuned and gradually reduced to 20%-30% of the power output to ensure that the detection temperature of the target billet is stable within the target range and enters the stable holding stage to prevent the target billet from overheating or temperature fluctuations affecting the subsequent forging process.

[0224] In addition, the trend prediction model can be updated through real-time collected data to continuously optimize the prediction accuracy and make the heating process more efficient.

[0225] If the detected temperature of the target billet is lower than the lower temperature limit or higher than the upper temperature limit, the system will automatically alarm and take emergency measures, such as immediately reducing power or stopping heating.

[0226] In this embodiment, during the process of heating the target steel billet, the temperature change is monitored in real time, combined with the furnace environment and power data, and data modeling is used to predict the temperature change trend. Based on this trend, the power of the heating furnace is adjusted to ensure that the target steel billet is stably heated within the temperature range of the forging conditions. The heating method based on data prediction and precise control can not only improve the heating efficiency, but also avoid overheating or uneven temperature, thereby ensuring the quality of the target steel billet and the smooth progress of subsequent processing. Data modeling is used to predict the temperature change trend of the target steel billet, and the power of the heating furnace is adjusted to ensure that the detected temperature of the target steel billet reaches the upper limit of the forging conditions. The forging time of the target steel billet in the forging machine can be extended and the heating frequency can be reduced.

[0227] In some embodiments, based on the temperature change and the temperature change trend, the power of the heating furnace is adjusted until the detected temperature of the target steel billet reaches the upper temperature limit of the forging condition, including:

[0228] When it is determined according to the temperature change that the detected temperature of the target steel billet is lower than the lower limit of the temperature, heating the target steel billet with the initial power by the heating furnace;

[0229] When it is determined according to the temperature change that the detected temperature of the target steel billet is not lower than the lower temperature limit, the power of the heating furnace is adjusted based on the temperature change trend and the initial power until the detected temperature of the target steel billet reaches the upper temperature limit of the forging conditions.

[0230] At the beginning of heating, the target billet temperature is monitored in real time by the sensor. The temperature information of the target billet is T initial The upper limit of the forging temperature is T upperlimit , if T initial <T lowerlimit , that is, the temperature of the target billet is lower than the lower limit of the forging temperature, and then it enters the initial heating stage.

[0231] To ensure that the billet is quickly heated to a temperature range suitable for forging, the initial power of the heating furnace is set for heating. The initial power is a higher power output value that can quickly heat the target billet and make its temperature transition to the upper limit, thereby increasing the heating rate and saving time.

[0232] During the heating process, the temperature sensor will provide real-time feedback on the temperature change of the target billet, and the furnace power will be adjusted in real time according to the feedback data. The temperature change curve will be recorded in real time to monitor the heating effect.

[0233] When the temperature of the steel billet gradually increases, the sensor detects that the detection temperature of the target steel billet is close to or no longer lower than the lower limit of the temperature under the forging condition, T initial ≥T lowerlimit, the power of the heating furnace will be adjusted according to the temperature change trend.

[0234] By using linear regression or curve fitting methods, the temperature trend of the billet is analyzed to determine the rate of temperature increase. If the temperature increase rate is too fast, the power output is reduced to avoid the billet temperature from increasing too fast.

[0235] If the temperature rises too slowly, the power of the heating furnace should be appropriately increased to speed up the temperature rise so that the detected temperature of the target billet is stabilized within the forging temperature range.

[0236] The furnace power output is adjusted by a PID controller according to the temperature change curve, heating rate and temperature control target.

[0237] When the target billet temperature reaches the upper temperature limit T upperlimit When the temperature is too low, the heating furnace will automatically stop or adjust to maintain a stable temperature.

[0238] During the adjustment and holding stage, the power output in the furnace is continuously adjusted to ensure that the detection temperature of the target steel billet does not exceed the upper temperature limit, while maintaining the temperature range required for the forging conditions.

[0239] After the temperature stabilizes, the heating furnace enters the insulation stage to keep the billet temperature stable for a period of time to ensure that the entire billet is heated evenly and achieves the expected material properties.

[0240] In addition, a temperature over-limit alarm mechanism can be set to ensure that the temperature does not exceed the set range, preventing the target steel billet from degrading due to overheating.

[0241] In this embodiment, it is possible to ensure that the target steel billet can achieve precise temperature control during the heating process. By dynamically adjusting the power according to the real-time temperature change, not only the uniformity of steel billet heating is ensured, but also energy waste can be avoided, production efficiency and product quality can be improved, and various key parameters of the steel billet can be monitored in real time, and intelligent adjustments can be made during the forging process, which not only improves production efficiency and reduces manual intervention, but also significantly improves the dimensional accuracy and mechanical properties of the finished product, meets the high-precision and high-quality production requirements, improves the level of automation and intelligence, and controls the entire production process more finely, further reduces energy consumption, increases output value, and promotes the development of intelligent manufacturing.

[0242] The embodiment of the present application provides a size measurement control method for steel billet forging, and the execution body can be a size measurement control device for steel billet forging. The embodiment of the present application takes a size measurement control device for steel billet forging executing a size measurement control method for steel billet forging as an example to illustrate a size measurement control device for steel billet forging provided in the embodiment of the present application.

[0243] The embodiment of the present application also provides a size measurement and control device for steel billet forging.

[0244] The size measurement and control device for forging a steel billet comprises:

[0245] An acquisition module is used to obtain material information, three-dimensional point cloud and temperature information of the target steel billet;

[0246] A first processing module is used to perform three-dimensional modeling on the target steel billet based on the three-dimensional point cloud to determine size information of the target steel billet;

[0247] A second processing module is used to determine a position of the target steel billet to be processed according to the size information when it is determined that the temperature information satisfies the forging conditions;

[0248] a third processing module, for inputting the material information and the temperature information corresponding to the position to be processed into a processing prediction model, and obtaining the forging force corresponding to the position to be processed output by the processing prediction model, wherein the processing prediction model is obtained based on a data set including a plurality of training samples;

[0249] The fourth processing module is used to configure parameters of the forging machine based on the position to be processed and the forging force, so as to control the forging machine to forge the position to be processed of the target steel billet with the forging force.

[0250] A size measurement and control device for steel billet forging in the embodiment of the present application can be an electronic device, or a component in an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal, or other devices other than a terminal. Exemplarily, the electronic device can be a mobile phone, a tablet computer, a laptop computer, a PDA, a mobile Internet device (MID), an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), etc., which is not specifically limited in the embodiment of the present application.

[0251] A size measurement control device for steel billet forging in the embodiment of the present application may be a device having an operating system. The operating system may be a Windows operating system, a Linux operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0252] A size measurement and control device for steel billet forging provided in an embodiment of the present application can implement various processes implemented by a size measurement and control method for steel billet forging in the above-mentioned embodiment, and can achieve the same technical effect. To avoid repetition, it will not be described here.

[0253] In some embodiments, Figure 5 As shown, an embodiment of the present application also provides an electronic device 500, including a processor 501, a memory 502, and a computer program stored in the memory 502 and executable on the processor 501. When the program is executed by the processor 501, each process of the above-mentioned embodiment of the dimension measurement control method for forging a steel billet is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0254] It should be noted that the electronic devices in the embodiments of the present application include the mobile electronic devices and non-mobile electronic devices mentioned above.

[0255] An embodiment of the present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the above-mentioned embodiment of the dimension measurement control method for forging a steel billet are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0256] The processor is the processor in the electronic device in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0257] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned dimension measurement control method for steel billet forging.

[0258] The processor is the processor in the electronic device in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk.

[0259] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, which are coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-mentioned embodiment of the size measurement and control method for forging a steel billet, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0260] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0261] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0262] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment method can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute a size measurement control method for steel billet forging in each embodiment of the present application.

[0263] In the description of this application, "first feature" or "second feature" may include one or more of the features.

[0264] In the description of the present application, “plurality” means two or more.

[0265] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

[0266] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0267] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the claims and their equivalents.

Claims

1. A method for measuring and controlling the size of a steel billet forging, characterized in that: include: Obtain material information, 3D point cloud and temperature information of the target steel billet; Based on the three-dimensional point cloud, three-dimensional modeling is performed on the target steel billet to determine size information of the target steel billet; When it is determined that the temperature information satisfies the forging conditions, determining the position of the target steel billet to be processed according to the size information; Inputting the material information and the temperature information corresponding to the position to be processed into a processing prediction model to obtain the forging force corresponding to the position to be processed output by the processing prediction model, wherein the processing prediction model is obtained based on a data set including a plurality of training samples; Based on the position to be processed and the forging force, parameters of the forging machine are configured to control the forging machine to forge the position to be processed of the target steel billet with the forging force.

2. The method for measuring and controlling the size of steel billet forging according to claim 1, characterized in that: The processing detection model is constructed based on a feedforward neural network. The processing prediction model includes an input layer, a hidden layer and an output layer connected in sequence. The material information and the temperature information corresponding to the position to be processed are input into the processing prediction model to obtain the forging force corresponding to the position to be processed output by the processing prediction model, including: The material information and the temperature information are feature mapped and converted through the input layer to obtain a comprehensive feature vector; Extracting and predicting the comprehensive feature vector through the hidden layer to obtain a force vector, wherein the hidden layer is constructed based on a multi-head attention mechanism; The forging force vector is mapped through the output layer to obtain the forging force and output it.

3. The method for measuring and controlling the size of steel billet forging according to claim 2, characterized in that: The input layer includes a design layer, a fully connected layer and a subsequent layer connected in sequence; The temperature information and the material information are normalized and mapped to a vector space through a design layer to obtain a temperature vector representing the temperature data distribution of the position to be processed and a material vector corresponding to the material information, and the temperature vector and the material vector are spliced ​​to obtain a spliced ​​vector; Calling a nonlinear activation function through a fully connected layer to perform a nonlinear transformation on the concatenated vector to obtain a transformation vector; The transformation vector is linearly transformed and dimensionally reduced through subsequent layers to obtain the comprehensive feature vector.

4. The method for measuring and controlling the size of steel billet forging according to claim 3, characterized in that: The nonlinear activation function is: Mish(x)=min(x·tanh(α·ln(1+e βx )),C) Among them, x is the concatenated vector of the input of the fully connected layer; α and β are learnable parameters; tanh is the hyperbolic tangent function, and C is a preset constant.

5. The method for measuring and controlling the size of steel billet forging according to claim 1, characterized in that: Before inputting the material information and the temperature information corresponding to the position to be processed into the processing prediction model, the method further includes: Obtaining sample temperature information and sample material information of a plurality of sample steel billets, as well as sample forging forces; According to the sample sequences of the plurality of sample steel billets, the sample temperature information, the sample material information and the sample forging force are sorted to obtain a first temperature array, a first material array and a first forging force array; Based on the interpolation points of the sample sequence, synchronous interpolation calculations are performed on the first temperature array, the first material array and the first forging intensity array to obtain a second temperature array, a second material array and a second forging intensity array; Performing linear dimensionality reduction on the second material array, and performing nonlinear dimensionality reduction on the linearly reduced second material array; Verify the second temperature array, the second material array with nonlinear dimensionality reduction, and the second forging intensity array respectively, and if the verification passes, construct a data set for the second temperature array, the second material array, and the second forging intensity array; The training samples in the data set are input into the feedforward neural network for training in sequence to obtain the processing prediction model.

6. The method for measuring and controlling the size of steel billet forging according to claim 5, characterized in that: The training samples in the data set are sequentially input into the feedforward neural network for training to obtain a processing prediction model, including: Extracting a plurality of first training samples from the data set; Multiple first training samples are sequentially input into the feedforward neural network, and each hidden layer in the feedforward neural network calculates the activation value according to the output of the previous layer: Among them, h i is the output of the i-th hidden layer neuron, w ij is the weight of the neurons from the jth layer to the ith hidden layer, the jth layer is the input layer or the previous layer, b i is the bias of the i-th hidden layer neuron, f(·) is the activation function, x j is the output of the jth layer, which serves as the input of the i-th hidden layer neuron; Obtain the prediction result output by the feedforward neural network, and calculate the loss value corresponding to the prediction result; Using the chain rule, the backpropagation algorithm calculates the gradient of the loss function with respect to the hidden layer weights and biases: in, is the gradient of the loss function L with respect to the output of the hidden layer; According to the calculated gradient, update the weights and bias of the hidden layer: Where μ is the learning rate; Repeatedly extract multiple second training samples from the data set, and input the multiple second training samples into the feedforward neural network in sequence until the obtained loss value is less than the loss threshold, the feedforward neural network training is completed, and the processing prediction model is obtained.

7. The method for measuring and controlling the size of steel billet forging according to claim 1, characterized in that: The forging machine is configured with parameters based on the position to be processed and the forging force, so as to control the forging machine to forge the position to be processed of the target steel billet with the forging force, including: According to the forging force, configure the working frequency of the air hammer in the forging machine and the force of a single blow; According to the position to be processed, configuring the moving trajectory of the air hammer in the forging machine; Based on the position to be processed and the moving trajectory, the air hammer is globally positioned by a photoelectric sensor, and the air hammer is locally positioned by a laser sensor; Based on the working frequency and the force of a single blow, forging the target steel billet by the positioned air hammer; According to the comparison between the forging position of the air hammer and the position to be processed, the position of the air hammer is corrected; The operating frequency and the force of a single blow are adjusted according to the deformation degree and temperature change of the target steel billet.

8. The method for measuring and controlling the size of steel billet forging according to claim 1, characterized in that: After configuring parameters of a forging machine based on the position to be processed and the forging force to control the forging machine to forge the position to be processed of the target steel billet with the forging force, the method further includes: After each forging, the target steel billet is subjected to temperature and deformation testing; When it is determined that the temperature of the target steel billet is lower than the lower temperature limit of the forging condition, heating the target steel billet by a heating furnace until the temperature of the target steel billet is higher than the upper temperature limit of the forging condition; The three-dimensional point cloud and temperature information of the target steel billet are reacquired to determine the position to be processed and predict the forging force in combination with the material information, and the forging machine is controlled to forge the target steel billet until it is determined according to the deformation detection that the size difference between the target steel billet and the standard steel billet template is within a preset error range, thereby obtaining the forged target steel billet.

9. The method for measuring and controlling the size of steel billet forging according to claim 1, characterized in that: After obtaining the material information, the three-dimensional point cloud and the temperature information of the target steel billet, the method further includes: When it is determined that the temperature information is lower than the lower limit of the temperature of the forging condition, heating the target steel billet by a heating furnace and detecting the temperature change of the target steel billet; Based on the temperature change, the furnace environment and real-time power of the heating furnace, predicting the temperature change trend of the target steel billet through data modeling; Based on the temperature change and the temperature change trend, the power of the heating furnace is adjusted until the detected temperature of the target steel billet reaches the upper temperature limit of the forging condition.

10. The method for measuring and controlling the size of steel billet forging according to claim 9, characterized in that: Based on the temperature change and the temperature change trend, the power of the heating furnace is adjusted until the detected temperature of the target steel billet reaches the upper temperature limit of the forging condition, including: When it is determined according to the temperature change that the detected temperature of the target steel billet is lower than the lower limit of the temperature, heating the target steel billet with the initial power by the heating furnace; When it is determined according to the temperature change that the detected temperature of the target steel billet is not lower than the lower temperature limit, the power of the heating furnace is adjusted based on the temperature change trend and the initial power until the detected temperature of the target steel billet reaches the upper temperature limit of the forging conditions.

Images