Method and system for controlling flattening of valve shell forge piece height limiting block

Abstract

The invention relates to the technical field of metal plastic forming control, and particularly discloses a method and system for controlling flattening of a height limiting block of a valve shell forging. The method comprises the steps that the initial size, temperature and material attribute parameters of a forged piece are obtained, and initial working condition feature vectors are generated; a control model is constructed based on historical forging samples, and a target pressure control curve and initial height limiting block setting parameters are output; collecting real-time voltage control process data and generating a deviation analysis result; calculating the correction displacement of the height limiting block and generating a control instruction; executing a fine tuning action, collecting corrected data, and constructing a feedback matrix; and matching analysis is carried out in combination with final state structure detection parameters, and model optimization and database updating are completed. According to the method, dynamic correction and closed-loop optimization of the height limit control model are realized, and the forming consistency and intelligent control level of the forge piece are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of metal plastic forming control, and in particular to a method and system for controlling the flattening of a height limiting block of a valve housing forging. Background Art

[0002] In key manufacturing sectors such as aviation and energy, valve housing forgings, due to their complex structures and high precision requirements, place extremely high demands on forming control stability during the forging process. Traditional forging control methods, which rely on fixed height limiter setting parameters and empirical operation, struggle to respond in real time to forming deviations caused by material properties, initial state, and process fluctuations. This can easily lead to defects such as overpressure, underpressure, or uneven wall thickness during the forging flattening process, compromising the final structural quality and subsequent assembly accuracy.

[0003] Existing control methods are mainly adjusted through static limits and process experience. They lack the ability to integrate and analyze the actual initial working conditions of forgings (such as size, temperature, material, etc.) and the dynamic parameters of the pressure control process (such as stroke, load, deformation behavior, etc.). They also fail to effectively establish a feedback learning mechanism between process data and forming results, making it difficult to meet the needs of high consistency and high precision forming control.

[0004] Therefore, how to dynamically construct and optimize the height limit block control model based on the initial working condition characteristic vector of the forging, the pressure control process behavior parameters and the final structure deviation data, and realize the intelligent regulation of the flattening process of the valve shell forging and the guarantee of forming consistency has become a key technical problem that needs to be solved urgently in the current field of intelligent forging control. Summary of the Invention

[0005] The present invention provides a method and system for controlling the flattening of a valve shell forging by a height limit block, so as to solve the problem of how to dynamically construct and optimize a height limit block control model based on the initial working condition characteristic vector of the forging, the pressure control process behavior parameters and the final structure deviation data, so as to realize intelligent regulation of the flattening process of the valve shell forging and ensure the forming consistency.

[0006] In order to solve the above technical problems, the present invention provides a method for controlling the flattening of a valve housing forging height limit block, comprising: Obtaining the initial outer diameter, length, wall thickness, surface temperature, internal temperature, material yield strength, elastic modulus, and thermal expansion coefficient of the forging, constructing a dimensional parameter set and a physical parameter set, and normalizing the sets to generate an initial working condition feature vector; Based on the initial working condition feature vector and a forging sample database containing historical working condition samples and pressure control records, a height limit block control model is constructed to output a target pressure control curve and initial height limit block setting parameters; Collect real-time flattening force, pressure control stroke, real-time temperature and equivalent plastic strain to build process parameter sequence; Dynamically comparing the process parameter sequence with the target pressure control curve to generate a pressure control deviation sequence, and constructing a deviation analysis result set through trend extraction and tolerance judgment; Calculate the corrected displacement parameters of the height limit block based on the deviation analysis result set, real-time temperature and material parameters, and generate a control instruction including a control type, an adjustment range and an execution mode; Executing the control instruction to complete the position correction of the height limit block, collecting the adjusted voltage control behavior parameters, and constructing a correction feedback voltage control matrix; Comparing the feedback pressure control matrix with the final structure detection parameters of the forging to construct a structure matching deviation set; Error backtracking and sample encoding are performed based on the initial eigenvector, feedback matrix and structural deviation set, the forging sample database is updated, and the control model and target pressure control curve are optimized.

[0007] Furthermore, the step of generating the initial operating condition characteristic vector includes: The size parameter set and physical parameter set are normalized and feature fused respectively, and the principal component extraction method is used to construct the feature vector dimensional expression structure.

[0008] Furthermore, the height limit block control model is a multivariable nonlinear regression model or a regression prediction model based on a neural network, and is pre-trained using historical samples with annotated final state deviations.

[0009] Furthermore, the collected real-time pressure control stroke and the target pressure control curve are aligned in a dynamic window manner to construct a pressure control deviation sequence, and smoothing filtering and cumulative offset processing are performed.

[0010] Furthermore, the calculation process of the height limit block correction displacement parameter adopts a weighted factor fusion method to fuse the average pressure control deviation, material yield strength and temperature gradient to construct a dynamic control weight.

[0011] Furthermore, the control instruction includes a control type field, a correction direction field, a correction amplitude field and an execution mode field, and is transmitted to the height limit control unit through a standard industrial communication interface.

[0012] Furthermore, the rows of the feedback pressure control matrix are sampling timing, and the columns are corrected flattening force, stroke, temperature and deformation rate, and the data are filtered, merged and structured.

[0013] Furthermore, the final structural detection parameters of the forging include the final outer diameter, length and wall thickness after forming, and are collected using a three-dimensional optical scanning device.

[0014] Furthermore, the error backtracking and sample updating process includes structural deviation clustering, control behavior association and label supplementation operations, which are used to expand the multi-operating condition coverage of historical samples.

[0015] Furthermore, a valve housing forging height limit block controlled flattening system comprises: The initial working condition acquisition module is used to obtain and encode the initial parameters of the forging and generate a working condition feature vector; The control model building module is used to generate a pressure control prediction model and a target pressure control curve; The real-time voltage control monitoring module is used to collect and calculate the voltage control deviation sequence; The height limit correction control module is used to generate a height limit correction control instruction according to the deviation analysis result; The execution feedback collection module is used to complete the height limit block adjustment and feedback data collection; The model optimization module is used to perform structure matching analysis and model updating.

[0016] The following are its main beneficial effects: (1) The working condition perception and model adaptive update mechanism of the forging height limit block control process are realized, improving the accuracy of the pressure control process and the forming consistency. The present invention obtains the initial outer diameter, length, wall thickness, temperature and material parameters of the forging to generate a standardized initial working condition feature vector, and inputs it into the height limit block control model trained based on historical samples, thereby dynamically generating the target pressure control curve and setting parameters for the current forging characteristics, avoiding the inaccuracy problem of traditional empirical settings.

[0017] (2) A dynamic correction calculation method that integrates pressure control deviation trends, material parameters, and thermal state effects has been established to enhance the real-time response capability of the height limit block control. By constructing a pressure control deviation sequence and error analysis result set, combined with real-time temperature and material properties, this method constructs a multi-factor correction parameter calculation formula, outputs the height limit block correction displacement, and generates control instructions, achieving fine-tuning control. Compared with traditional fixed correction values, this method can effectively reduce misjudgments, over-adjustments, or adjustment lags.

[0018] (3) A closed-loop learning mechanism of final structure and process feedback was constructed to achieve continuous evolution of the forging model and improvement of its adaptability to multiple working conditions.

[0019] The present invention performs structural matching analysis on the final structure detection parameters and the pressure control behavior matrix after height limit execution, traces the source of errors, and updates the forging sample database and height limit control model to achieve iterative optimization of the model and improve the prediction accuracy and control stability of subsequent forgings. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1A schematic flow chart of a method for controlling flattening of a valve housing forging by a height limiting block provided in an embodiment of the present application; Figure 2 A structural block diagram of a system for controlling flattening of a valve housing forging height limiting block provided in an embodiment of the present application; Figure 3 A diagram showing a folding defect of a valve housing forging provided in an embodiment of the present application; Figure 4 A diagram showing the steps of the die pressing process for the valve housing forging provided in an embodiment of the present application; Figure 5 A schematic diagram of the flattening of a valve housing forging provided in an embodiment of the present application; Figure 6 Schematic diagram of the valve housing forging provided in the embodiment of the present application being flattened by a height limiting block. DETAILED DESCRIPTION

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as commonly understood by those skilled in the art to which this application belongs. The terms used in the specification of the application are for the purpose of describing specific embodiments only and are not intended to limit this application. The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of this application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0022] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0023] Example 1: Reference Figure 1 , is a flow chart of a method for controlling flattening of a valve housing forging by a height limiting block provided by an embodiment of the present invention. The flow chart may include at least steps S100 to S600: S100, obtaining initial size parameters, temperature parameters, and material properties of the forging, and performing feature encoding on the parameters to generate an initial working condition feature vector; S200, initializing a height limit block control model based on the initial working condition characteristic vector and historical forging data, and generating a target pressure control curve and initial height limit block setting parameters; S300, collecting real-time flattening force, stroke, temperature, and forming deformation parameters during the forging process, and calculating a pressure control deviation sequence based on the target pressure control curve; S400, performing error judgment and dynamic adjustment calculation on the voltage control deviation sequence, generating a height limit block correction displacement parameter, and sending an execution instruction to the height limit control unit; S500, based on the displacement parameter of the height limit block, performing a fine adjustment action of the height limit block, collecting a sequence of adjusted flattening behavior parameters, and constructing a process feedback pressure control matrix; S600, performing structural matching analysis on the process feedback pressure control matrix and the final state detection parameters of the forging, and updating the working condition sample data set to optimize the height limit control model; Step S100 at least includes steps S110-S130: S110: Obtain an initial size parameter set of the forging to be processed.

[0024] In this step, the initial geometric state of the valve housing forging to be machined is obtained before forging. Specifically, the system retrieves basic geometric information of the forging from the upstream pre-processing station or loading process. This information includes but is not limited to: outer diameter parameters, length parameters, wall thickness parameters, inner diameter parameters, and initial volume estimation.

[0025] Specifically, the outer diameter parameter is the maximum diameter of the outer circle of the forging, which is defined as the initial size parameter ; The length parameter is defined as the total axial length of the forging from end to end, denoted as ;Wall thickness parameters is the radial thickness difference between the outer diameter and the inner diameter. Understandably, in order to ensure the pressure control accuracy, the size parameter set also includes auxiliary parameters such as forging ovality, end eccentricity and center hole roundness, which are uniformly included in the size parameter set and recorded as: size set = { 、 、 、 、 、 }.

[0026] During the dimension acquisition process, a non-contact laser measurement system or 3D contour scanning device can be used to acquire a spatial geometric point cloud. This point cloud data is then fitted to a standard CAD model to extract a standardized parameter set. Subsequent steps can then be based on this parameter set for standardization. In actual implementation, the output of the dimension parameter acquisition module directly serves as one of the input conditions for physical parameter modeling in S120, and together with it, serves as the original input vector set for feature vector construction in S130.

[0027] S120. Collect forging material properties and temperature parameters to construct a physical parameter set.

[0028] After completing the dimensional parameter collection, the system activates the physical parameter acquisition module to acquire the metallurgical characteristics, mechanical properties, and processing temperature of the materials corresponding to the current batch of forgings. These parameters are not only used as input factors in the subsequent height limit block control modeling, but are also highly correlated with the behavioral deviation model during the actual pressure control process in S300.

[0029] Specifically, the material property parameter set Including but not limited to: Yield strength of the material , elastic modulus , thermal expansion coefficient , plasticity index , work hardening coefficient , thermal conductivity and specific heat capacity Some of these parameters can be retrieved from the material database, while others require confirmation by the upstream pre-furnace sample inspection module. For example, for valve housing forgings made of high-strength titanium alloy or aviation aluminum alloy, the system will retrieve the corresponding material number and corresponding thermodynamic properties from the material database based on the specific alloy model.

[0030] At the same time, the system collects the surface and internal temperatures of the forgings through infrared temperature measuring devices or embedded thermocouples, which are defined as surface temperature parameters and internal temperature parameters The surface temperature and internal temperature constitute the processing temperature subset, which provides the necessary basis for controlling the flattening response time and rheological properties.

[0031] The formation of physical parameter set can be expressed as: physical parameter set = { 、 、 、 、 、 、 、 、 }.

[0032] The above physical parameter set will be input together with the size parameter set obtained in S110 in S130 to construct the initial working condition feature vector. 、 、 It will also directly participate in the estimation of actual pressure control deviation in S300, forming a cross-module indicator sharing mechanism.

[0033] S130. Generate the initial working condition feature vector of the forging.

[0034] After constructing the dimensional and physical parameter sets, the system enters the feature extraction and normalization phase. This step corresponds to the core sub-step of initial state modeling in the present invention, namely, forming a unified initial forging condition feature vector by normalizing, weighted combining, and high-dimensional mapping heterogeneous parameters.

[0035] Specifically, the system first performs vector splicing on various indicators in the size parameter set and the physical parameter set to construct a comprehensive parameter matrix. Then, based on the preset feature weight system, different indicators are weighted summed and normalized according to their weights in subsequent control modeling, and finally a feature vector is generated. The eigenvector It is a multidimensional vector, including the shape characteristics, material property characteristics and hot processing characteristics of the forging, and is the core input quantity in the control process of the present invention.

[0036] Furthermore, the system reduces the dimensionality of these high-dimensional parameters through principal component analysis (PCA), Bayesian feature mapping, or convolutional feature fusion methods to ensure convergence efficiency during model calculation. During the construction process, feature selection strategies from prior empirical models can be introduced to filter out redundant and low-contribution features.

[0037] The final generated initial working condition feature vector This feature vector is transmitted to S200 for initial modeling of the height limit block control model and generation of the pressure control curve. In subsequent processes, this feature vector also serves as an input benchmark for constructing the target pressure control curve in S220, comparing it with the real-time pressure control behavior to determine whether the deviation is reasonable in S300, and determining whether the final structural error matches the modeling residual in S600, forming the main control input indicator throughout the entire process.

[0038] Description of the connection between the front and back: The various parameter indicators collected in S110 and S120 are integrated and processed in S130 and uniformly mapped to ; described Used as input data by S210 for matching modeling with historical sample data; Among them 、 and The indicators will also be called again during the voltage control deviation correction in S300; In S600, the structural mapping difference between the final structural state of the forging and the initial features also depends on the principal component of the feature vector; Therefore, S100 is not only the front-end acquisition module of the system, but also the key support source for model construction, error judgment and back-end optimization processes.

[0039] Description of the technical effects of this paragraph: This section constructs the initial feature vector of multi-source fusion of size, material and temperature to provide accurate, standardized and repeatable input benchmark for the subsequent height limit block control model, significantly enhances the accuracy and generalization ability of flattening process modeling, and realizes the data closed-loop support for the entire process from the initial state to the final structure.

[0040] Step S200 at least includes steps S210-S230: S210: Construct a pressure-controlled modeling input data set.

[0041] The initial working condition feature vector of the forging in completing S130 After the generation of , the system starts the modeling process of the height limit block control model. Specifically, the system calls a set of fully labeled sample data from the historical forging sample database to construct a pressure control modeling input data set that matches the current working conditions. The historical forging sample database is defined as a set ,in, Indicates the The initial working condition characteristic vector of the historical forgings, structure and consistent; It represents the pressure control curve of the forging process corresponding to the characteristic vector of the working condition, including the three-dimensional relationship data of stroke-pressure-deformation.

[0042] Specifically, the system first converts the feature vector of the forging to be processed into All sample vectors in the historical database Perform similarity calculation. In this process, the difference between the current vector and the sample vector is compared using weighted Euclidean distance, Mahalanobis distance or cosine similarity, and the previous vector is selected according to the similarity. The closest sample data to form a modeling candidate subset .

[0043] In the selected candidate sample subset, each group It will be used to fit the initial state space of the height limit control model. Understandably, in order to improve the adaptability of the model to complex working conditions, the system needs to pay special attention to the vector output by S130 during the sample matching process. Key parameters such as surface temperature , internal temperature , yield strength and outer diameter parameters , which is used as a strong weight indicator for similarity matching.

[0044] At the end of this step, the system forms a data set for modeling , providing input conditions for S220 model fitting.

[0045] S220: Generate a pressure control prediction model.

[0046] After building the model input dataset After that, the system enters the pressure control prediction model fitting stage. The goal of the model fitting is to generate a set of function expression models for predicting the pressure control response behavior of the height limit block, denoted as ,in Represents the pressure-controlled height variable during the forging process.

[0047] Specifically, the system will input each set of forging curves in the sample Decomposed into a discrete sequence of data points, including dimensions such as stroke variable, flattening force response, deformation rate, etc., and combined with its corresponding working condition vector In this embodiment, the modeling method can use a variety of regression strategies, such as polynomial regression, support vector regression, deep neural network regression or random forest regression.

[0048] The system is based on the current vector The difference between the working conditions in the candidate sample set and the parameter fitting weights of different dimensions are set to control the model's performance on specific feature dimensions (such as 、 ) to improve the response sensitivity of the model and enhance the local adaptability of the model.

[0049] After fitting is completed, the system generates a pressure control prediction model , the model can be used at any pressure-controlled height The corresponding theoretical crushing force, deformation, and material response are output. This prediction model is not only used to generate the target pressure control curve in S230, but also compared with the actual collected pressure control behavior data in S300 for deviation, serving as a reference for error evaluation.

[0050] At the same time, to ensure that the model forms a closed loop with the model optimization process in S600, the system identifies and records the fitting parameters and sample reference relationships used in modeling, so that the model structure and parameter weights can be reversely corrected based on the actual final state results.

[0051] S230: Output the target pressure control curve and the initial value parameters of the height limit.

[0052] After completing the pressure control prediction model After the generation of the current working condition vector With model The function expression relationship is used to generate the target pressure control curve , and derive the initial height limit block setting parameters from the key point position in the curve .

[0053] Specifically, during the process of generating the pressure control curve, the system divides the pressure control interval into several discrete height points. , calculate the corresponding theoretical flattening force value respectively, and form a function map. The system determines the end point of the pressure control process based on the forming limit criteria (such as maximum flattening rate, maximum allowable stress, critical compression), thereby obtaining the target pressure control end point position This position is used as the lower limit of the required forming height of the forging, which is the reference position parameter of the initial height limit block. .

[0054] In the specific implementation, the system can also integrate state variables such as material yield point, processing temperature threshold, elastic-plastic transition critical point, etc. The curve is segmented and marked to identify areas such as "linear compression section", "plastic flow section", and "stable forming section" to assist in stage judgment during the S300 pressure control monitoring process.

[0055] Finally, the system sets the target pressure control curve The data is passed to S300 and used as a reference for comparison with the real-time pressure control behavior. At the same time, the initial setting parameters of the height limit block are set. The position is sent to the height limit control unit as the initial setting position. If it is determined in the subsequent S400 that the pressure control deviation exceeds the tolerance range, a correction operation will be performed based on this.

[0056] Description of the connection between the front and back: This section is highly consistent with the parameter generation process in the previous S100, and the initial working condition feature vector used is The output from S130 includes a multi-source feature set of dimensions, temperatures, and materials, ensuring the consistency and integrity of the modeling input data.

[0057] S230 generated pressure control curve It is the direct input for S300 to perform deviation comparison; Height limit block setting parameters Referenced by S400 to determine the corrected displacement value; Model The structure and weight parameters of will be updated in S600 to form a model optimization feedback path.

[0058] Description of the technical effects of this paragraph: This section constructs a height limit block pressure control prediction model based on feature vectors and sample data, effectively realizing the intelligent mapping from working conditions to control behaviors, improving the dynamic adaptability and initial control accuracy of the height limit block flattening control, and forming a data flow closed loop running through S100 to S600.

[0059] Step S300 at least includes steps S310-S330: S310: Construct a forging process parameter sequence.

[0060] Output the target pressure control curve in S230 and the initial height limit block position parameters After that, the system enters the real-time monitoring stage of the forging process. Specifically, at the beginning of the flattening process, the system starts the embedded acquisition control module and relies on multiple sets of distributed sensing devices to continuously collect key parameters of the flattening process.

[0061] The key parameters include the following categories: Real-time crushing force: , indicating that at time Below, the axial force of the height limit block acting on the forging; Real-time pressure control stroke: , indicating that the height limit block is from the initial position Move to time The instantaneous pressure-controlled displacement at ; Real-time temperature parameters: , including the average temperature of the forging surface and interior; Real-time forming deformation: , represents the actual equivalent plastic strain of the forging under the current working conditions; Current voltage control stage label: , used to identify whether it is in the linear compression zone, plastic flow zone or stable forming zone.

[0062] The subscripts in the above variables Indicates the processing number of this forging, and the working condition feature vector generated in S100 Maintain the binding relationship. Each parameter is sampled at a high frequency (such as 1kHz) to form multi-channel time series data and constructed into the following matrix form: in, Indicates the end time point of this flattening process. That is, the original collected data sequence of this voltage control process, which serves as the input variable of S320.

[0063] Technical effect description: This data structure not only provides basic information for real-time control, but also constitutes the actual sample data source for subsequent modeling and correction, supporting the model update closed-loop logic (see S600).

[0064] S320: Calculate the real-time travel deviation sequence.

[0065] Furthermore, in obtaining the process parameter sequence After that, the system calls the target pressure control curve generated in S230 and compared with the real-time pressure control stroke data collected Perform point-by-point comparison to calculate the voltage control deviation sequence .

[0066] The target voltage control curve is defined as: in, is the target travel point, is the corresponding theoretically predicted pressure value, is the number of discrete nodes.

[0067] To ensure calculation consistency, the system will and Perform interpolation alignment to obtain the expected target travel value at each moment , and then perform differential processing: in, :Indicates at time Pressure control height deviation under ; : Measured stroke; : target trip predicted by the model; A positive value indicates insufficient pressure control, and a negative value indicates excessive pressure control.

[0068] To improve accuracy, the system also introduces the pressure control force difference as an auxiliary calibration item: Here, is the measured pressure value, The target curve The target force value obtained by interpolation. and If the signs are opposite, the system has a force-stroke response hysteresis and the height limit parameters need to be dynamically corrected.

[0069] Technical effect description: The voltage-controlled deviation sequence It is an important indicator for judging the consistency between the current pressure control behavior and the prediction model. It will be used in S400 to dynamically modify the control strategy and improve the response sensitivity and stability of the height limit control.

[0070] S330: Generate a deviation analysis result set.

[0071] The system is sensitive to the above deviation sequence Perform smoothing filtering to eliminate high-frequency noise caused by equipment vibration or data sampling errors. Specific methods can be sliding mean filtering, bidirectional exponential smoothing, or wavelet denoising. After smoothing, the processed deviation sequence is generated: Then, based on the processed sequence, the system constructs a fixed window sliding cumulative deviation function, calculates the average deviation, peak deviation, and offset trend within each period, and generates a voltage control deviation analysis result set: in: : is the average deviation value in this voltage control cycle; : is the deviation variance, used to evaluate stability; :It is the offset trend symbol, which can be , representing continuous negative deviation, stable deviation and positive deviation trends respectively.

[0072] Technical effect description: The voltage control deviation analysis result set It will serve as direct input data for the calculation of the corrected displacement parameters of the height limit block in S400, providing a directional reference and strength adjustment basis for subsequent height limit control.

[0073] Description of the connection between the front and back: The operating condition characteristic vector from S100 Temperature parameters in 、 and yield strength , which is cited in the estimation of strain response in pressure-controlled processes; Target pressure control curve from S230 and the initial height limit block position , used in this section for real-time voltage control deviation judgment; Generated deviation analysis result set , used as the basis for calculating the pressure control correction amount in S400; The collected original sequence It will also serve as a comparison baseline before and after fine-tuning behavior in S500 and be used by S600 for model optimization sample updates.

[0074] Description of the technical effects of this paragraph: Through the implementation of this section, the system can continuously, in real time and accurately identify the difference between the behavior of the height limit block and the target pressure control model during the forging process, and form a structured deviation evaluation quantity to provide a feedback loop for the height limit control model, ensuring the controllability, stability and high responsiveness of the height limit control process under nonlinear conditions.

[0075] Step S400 at least includes steps S410-S430: S410, constructing a deviation judgment result sequence First, the system obtains the voltage control deviation analysis result set output by S330: in, : average voltage control deviation; : bias variance; : Trend symbol, the value is {-1,0,+1}, indicating negative offset, stability and positive offset.

[0076] Next, the system calculates the error according to the set error tolerance threshold. The average voltage control deviation is judged if the following conditions are met: It is determined that the current voltage control state has deviated from the target voltage control curve , need to enter the correction stage.

[0077] System builds deviation judgment result sequence: in, It is set by the user or system experience parameters, usually in the range of 0.2-0.5mm.

[0078] Technical effect: This step converts continuous deviation behavior into discrete correction trigger criteria, which helps to reduce invalid control actions and improve the robustness of the control system.

[0079] S420: Calculate the corrected displacement parameters of the height limit block.

[0080] In obtaining the deviation judgment result sequence Afterwards, if there is any , the system enters the dynamic correction calculation process. In order to adapt to different material properties, temperature conditions and forming stages, the system constructs a height limit correction control function based on a multi-factor weighted method: in: : No. The height limit correction displacement of the forging task; : voltage control average deviation, derived from S330; : Current temperature parameter, derived from S310; : Material yield strength, derived from S120; : Deviation trend, derived from S330; : Weighting coefficient.

[0081] The parameters are explained as follows: : Control deviation response sensitivity, generally set to 0.6–0.8, used to quickly eliminate overshoot; : Thermal state compensation factor, generally set to 0.1–0.3, used to deal with the effect of temperature on material rheology; : Material hardness weight, generally set to 0.05–0.2, used to adapt material differences to pressure-controlled behavior; : Trend feedback adjustment factor, generally set to ±0.1–±0.2, determines the direction of correction.

[0082] Description of technical effects: This correction function introduces multi-dimensional factors such as temperature, material, and trend to achieve intelligent adaptive adjustment based on traditional deviation feedback, enhance the control system's ability to respond to complex forging environments, and solve the problems of misjudgment and excessive correction caused by empirical rules in traditional methods.

[0083] Furthermore, to ensure that the corrected displacement is within a reasonable range, the system executes the following constraint function: in, : Maximum correction step, usually set to 1mm; : The minimum executable step size, usually set to 0.1mm.

[0084] This limit constraint prevents violent control fluctuations and enhances system safety and stability.

[0085] S430: Convert the control instruction and send it.

[0086] Complete the correction of displacement parameters of the height limit block After calculating , the system converts the value into a displacement control instruction that can be recognized by the underlying actuator. Specifically, the system constructs the instruction structure: The above control instruction fields are described as follows: Cmd_Type: Control type, set to "Limit Height Adjustment"; Target_Offset: Target adjustment value, unit is mm; Mode: Execution mode, servo precision control is used by default.

[0087] The command is transmitted to the height limit control unit via the controller interface, and the height limit actuator controlled by the subsequent S500 completes the correction action. , it means that the current deviation is within the tolerable range and the system does not issue any adjustment instructions.

[0088] The system also archives the pressure control deviation sequence, correction parameters, instruction records and temperature-material status in the current control cycle to form an operation log, which is used as a backward learning sample input in S600 for control model optimization.

[0089] Description of the connection between the front and back: The pressure control deviation analysis result set used in this section Output from S330; Material parameters , temperature parameters From S120 and S310; Correction results It will be directly called by S500 to correct the height limit block position; All input data can be traced back to the initial feature vector in S100 ; The current control instructions are also used as feedback samples in S600 to participate in model updating, forming a closed-loop learning system.

[0090] Description of the technical effects of this paragraph: This section constructs a multi-parameter pressure control correction function and integrates temperature and material properties to significantly improve the height limit control system's ability to identify pressure control errors, its response speed, and its adaptability. Compared to traditional fixed-threshold correction strategies, this method maintains stable control performance over a wider range of operating conditions, offering greater anti-disturbance capabilities and greater pressure control precision.

[0091] Step S500 at least includes steps S510-S530: S510: Execute the height limit block position correction action.

[0092] Specifically, the system receives the height limit block correction control instruction output from the S430 sub-step, which includes the target correction displacement parameters of this correction cycle. The initial position parameters of the height limit block are , output by S230, the corrected displacement parameter is , calculated by sub-step S420.

[0093] The system controls the height limit actuator to adjust the physical position of the height limit block according to the control instructions, so that it can complete precise movement in the pressure-controlled axial direction and finally reach the target correction position. ,Right now: The target correction position is the target displacement value of the current pressure-controlled stroke, which represents the new limit position obtained by subtracting the dynamic correction amount from the original set height.

[0094] The height limit actuator can include a servo cylinder, an electro-hydraulic proportional cylinder, or a ball screw drive module, all with sub-millimeter displacement control accuracy. To ensure consistent displacement control, the system uses a position encoder for feedback monitoring during the height limit block adjustment process, recording and verifying the consistency between the actual and target travel of the height limit block.

[0095] After the fine-tuning action is completed, the system will correct the target position It is registered in the process control data set and serves as the time anchor point in subsequent data collection and feedback behaviors to ensure that the collected data are the real response behaviors in the corrected state.

[0096] S520: Collect pressure control behavior data after height limit correction.

[0097] After the height limit block completes fine-tuning and positioning, the system starts the multi-channel pressure-controlled behavior data acquisition mechanism to sample and record the actual response behavior of the forging during the flattening process after the height limit adjustment.

[0098] Specifically, the collected data includes but is not limited to: Corrected flattening force : Indicates the real-time axial load value of the press on the forging after the height limit block completes the fine-tuning action; Corrected pressure control stroke : Indicates the actual compression depth of the forging in this pressure control cycle; Corrected deformation : Indicates the equivalent plastic strain produced by the forging material in the modified state; Corrected temperature distribution parameters : Optional indicator, characterizing the effect of forging temperature rise on forming behavior.

[0099] The above-mentioned collection behavior is completed through the collaboration of high-frequency sensor arrays and edge control units, and each parameter is synchronously archived with a unified timestamp to ensure time consistency and data closure.

[0100] The acquisition sequence constitutes a set of time-correlated data vector sequences, which are recorded as: The feedback sequence is the direct response data after the height limit control behavior is executed. Its content is consistent with the original pressure control behavior parameters collected in step S310, which facilitates the subsequent formation of a before-after comparison relationship for error backtracking, model correction and system evaluation.

[0101] To improve data consistency, the system preprocesses each type of parameter sequence, including noise reduction, time alignment, and outlier removal, to ensure the structural integrity of the final data matrix.

[0102] S530: Construct a process feedback voltage control matrix.

[0103] After completing the multi-channel pressure control feedback data collection, the system enters the pressure control feedback matrix generation phase. Specifically, the collected flattening force sequence, pressure control stroke sequence, deformation variable sequence, etc. are processed into a matrix structure according to a unified time step to construct a structured feedback pressure control behavior matrix. .

[0104] The feedback voltage control matrix is ​​based on the full record data of the voltage control response after the height limit block corrects the behavior, and its format is as follows: The direction is the time step ; Column direction corresponds to 、 、 、 ; The matrix elements represent the numerical expressions of the corresponding parameters at that moment.

[0105] It is understandable that the structure of the matrix is ​​consistent with the process parameter matrix constructed in S310, so it is comparable and can be used for the final state structure deviation backtracking and closed-loop optimization of the pressure control model in S600.

[0106] At the same time, the system packages the feedback matrix together with the height limit correction instruction, execution results, and execution start and end time and stores them in the control log to form a complete control link record, thereby improving the auditability and traceability of the system.

[0107] Description of the connection between the front and back: The correction action in S510 is based on the error judgment and correction parameter generation in S400. The correction position parameter is the output result of sub-step S420. , and the initial pressure control reference position determined in S230 Closely linked to ensure that dynamic adjustment behavior is completed within the established control framework.

[0108] The process feedback voltage control matrix constructed in S530 It will directly serve as the input basis for S600 to conduct final state detection deviation analysis and model update training, forming the final data structure of the system closed-loop control process.

[0109] Description of the technical effects of this paragraph: The implementation of S500 enables real-time collection, structured storage, and closed-loop feedback of the process response effects of height limit block fine-tuning control actions, ensuring a clear response record for each adjustment, effectively improving the height limit control system's perceptibility, modifiability, and self-learning capabilities. Combined with S600, this data can also be used as historical sample input for iterative updates to the pressure control model, enhancing the system's adaptability under changing operating conditions.

[0110] Step S600 at least includes steps S610-S630: S610: Construct a final state structure matching deviation set.

[0111] Specifically, after the height limiter is fine-tuned, the system performs final inspection of the forging using an integrated, high-precision structural inspection device. This device, which can include a 3D laser scanner, a coordinate measuring machine, or an automated visual measurement system, collects the final dimensional parameters of the forging's key structural components according to design requirements.

[0112] In this embodiment, the set of forging final state detection parameters collected by the system is: Final outer diameter parameter: indicates the final outer diameter of the forging after flattening, defined as ; Final length parameter: represents the final size in the axial direction, defined as ; Final wall thickness parameter: represents the final wall thickness value, defined as .

[0113] The above parameters are unified into the final state structural parameter set , whose format is: The parameter acquisition accuracy is controlled within ±0.05mm to ensure comparability with the original design target parameters. This parameter set is the basic indicator for judging whether the pressure control results meet the target forming requirements.

[0114] The system further calls the target parameter set output by the height limit control model in the modeling stage from S200 , which includes target parameters such as the desired outer diameter, length, and wall thickness corresponding to the end point of the pressure control curve.

[0115] Compare the two sets of parameter sets one by one, calculate their differences, and generate the final state structure matching deviation set In this set, each deviation value represents the degree of deviation of the current workpiece in the corresponding dimension. The smaller the value, the closer the height limit control process is to the expectation.

[0116] S620: Input the control model to perform error backtracking analysis.

[0117] Furthermore, the system feeds back the process voltage control matrix constructed in step S530 Deviations from the above structure The process feedback matrix records the actual behavior of the forging at each time point during the entire press control process after the height limit block is fine-tuned, including key dynamic behaviors such as changes in flattening force, press control stroke, and material deformation.

[0118] The system compares the following three sets of data relationships through the error backtracking module in the control model: Initial operating condition eigenvector , calculated and generated by S130; Feedback matrix of actual pressure control process , generated by S500; Final structure matching deviation set , generated by S610.

[0119] The system uses this to establish a mapping path between pressure-controlled behavior and final structural response. Through data fitting, deviation clustering, key behavior extraction, etc., it identifies the key factors causing final state errors from the pressure-controlled behavior and evaluates the parameter sensitivity of the corresponding parts in the model.

[0120] During the implementation process, the system can establish individual labels for the current forging samples, marking whether their control behavior is effective and whether the formed structure is qualified, so as to be used for the label learning optimization mechanism of subsequent models.

[0121] S630: Update historical samples and control models.

[0122] After completing the error backtracking analysis, the system will return all the data elements of the current forging process, including the initial working condition feature vector , Process Behavior Matrix , final state deviation set , control adjustment history and height limit action execution trajectory and other contents are fused and encoded to generate new forging sample entries and append them to the historical sample database In

[15] , we extend the model learning data foundation.

[0123] At the same time, the system adjusts the pressure control prediction model based on the pressure control behavior deviation trend shown by the current data. and target voltage control curve Perform optimization adjustments. The optimization may adopt one of the following strategies: If there is a systematic drift trend in the final state deviation, the model update is mainly based on parameter retraining; If the deviation is personalized and sporadic, the model update is mainly based on the incremental learning mechanism; If the deviation direction is consistent for several consecutive times and exceeds the tolerance threshold, the model will be reconstructed at the structural level.

[0124] The updated model parameters are redeployed to S200 as the input basis for the subsequent forging modeling stage. This process forms a closed-loop learning mechanism of "forging → monitoring → judgment → correction → reflow → remodeling".

[0125] Description of the connection between the front and back: Target parameter set used by S610 The predicted results of the pressure control model in S230 and the compared detection parameters It is the actual final size obtained after S500, which is the key bridge for building a feedback loop.

[0126] Based on the initial feature vector generated by S100, S620 connects the entire process data flow from modeling (S200), pressure control (S300-S500) to the final forming state (S600), ensuring that the error generation mechanism is explainable at the system level and avoiding the black box of the model optimization process.

[0127] The completion of S630 execution marks the formal closure of the system's self-learning function. Its input comes from the output results of S100 to S500, and is finally updated on the S200 model construction module, enabling the system to have the ability to continuously evolve across cycles and batches.

[0128] Description of the technical effects of this paragraph: Through the implementation of this section, the system has realized a closed-loop data flow for the entire process from working condition identification, control execution, behavior feedback to model correction. It not only greatly improves the forming consistency of the height limit block control behavior, but also enhances the system's generalization ability, robustness and continuous evolution ability, providing a technical foundation for future multi-variety and multi-batch intelligent forging.

[0129] The key innovations of the present invention include: (1) A predictive control model construction mechanism of “initial working condition—historical sample modeling—target curve output” is proposed to solve the problem that the forming control strategy cannot be generalized between different forgings.

[0130] (2) Construct a calculation formula for the height limit block correction parameters that includes pressure control deviation, temperature, and material property factors to achieve dynamic fine-tuning and accurate issuance of control instructions.

[0131] (3) An error backtracking mechanism involving structural matching deviation and process feedback matrix is ​​introduced to form a closed-loop optimization link and realize the adaptive learning ability of the control model.

[0132] The following are its main beneficial effects: (1) The working condition perception and model adaptive update mechanism of the forging height limit block control process are realized, improving the accuracy of the pressure control process and the forming consistency. The present invention obtains the initial outer diameter, length, wall thickness, temperature and material parameters of the forging to generate a standardized initial working condition feature vector, and inputs it into the height limit block control model trained based on historical samples, thereby dynamically generating the target pressure control curve and setting parameters for the current forging characteristics, avoiding the inaccuracy problem of traditional empirical settings.

[0133] (2) A dynamic correction calculation method that integrates pressure control deviation trends, material parameters, and thermal state effects has been established to enhance the real-time response capability of the height limit block control. By constructing a pressure control deviation sequence and error analysis result set, combined with real-time temperature and material properties, this method constructs a multi-factor correction parameter calculation formula, outputs the height limit block correction displacement, and generates control instructions, achieving fine-tuning control. Compared with traditional fixed correction values, this method can effectively reduce misjudgments, over-adjustments, or adjustment lags.

[0134] (3) A closed-loop learning mechanism of final structure and process feedback was constructed to achieve continuous evolution of the forging model and improvement of its adaptability to multiple working conditions.

[0135] The present invention performs structural matching analysis on the final structure detection parameters and the pressure control behavior matrix after height limit execution, traces the source of errors, and updates the forging sample database and height limit control model to achieve iterative optimization of the model and improve the prediction accuracy and control stability of subsequent forgings.

[0136] Example 2: Figure 2 The structural block diagram of a system for controlling the flattening of a valve housing forging height limiting block according to an embodiment of the present invention is shown. Figure 2 As shown, the structure may include: The initial working condition acquisition module 10 is used to collect multi-dimensional data on the valve housing forging before forging, acquiring information including initial dimensional parameters (including outer diameter, length, and wall thickness), material property parameters (including yield strength, elastic modulus, and thermal expansion coefficient), and processing hot temperature information (including surface and internal temperatures). This module combines a high-precision laser measurement device, a thermocouple sensor, and a material database interface to complete structured modeling of the initial working condition characteristics and generate the forging's initial working condition feature vector, providing input for subsequent modeling modules.

[0137] The control model construction module 20 is used to construct a model input based on the initial working condition feature vector and a database of historical forging samples. It generates a height limiter control prediction model through multi-sample comparison and intelligent regression fitting. This module outputs the target pressure control curve and the corresponding initial setting parameters for the height limiter. It then performs adaptive model optimization based on multi-dimensional data weights, providing a precise control basis for the pressure control process.

[0138] The real-time pressure control monitoring module 30 collects the forging's real-time flattening force, pressure control stroke, temperature distribution, and deformation behavior during the forging process. It dynamically compares this data with the target pressure control curve, constructs a pressure control deviation sequence, and extracts stage-by-stage deviation trends and stability changes. This module relies on a multi-channel data acquisition card and high-frequency response sensors to ensure the continuity and accuracy of the acquisition process.

[0139] The height limit correction control module 40 determines whether height limit block correction is necessary based on the pressure control deviation analysis results. It then calculates dynamic correction displacement parameters for the height limit block by combining real-time temperature, material yield parameters, and pressure control offset trends. This module converts these correction parameters into standardized control instructions and sends them to the height limit block control actuator, driving it to perform real-time fine-tuning to adapt to the dynamically changing pressure control state.

[0140] The control execution and feedback acquisition module 50 is used to collect real-time behavior data from the next round of pressure control after the height limit block corrects. This includes parameters such as the corrected flattening force, stroke, and deformation rate, thereby constructing a complete pressure control behavior feedback sequence. This module further performs time-series merging on the feedback sequence to generate a structured process feedback pressure control matrix, providing core data support for final state structural comparison analysis and model retraining.

[0141] The final state detection and model optimization module 60 is used to obtain the final dimensional parameters of the forging (including final outer diameter, length, and wall thickness) after flattening. These parameters are then analyzed for structural matching with the target forming parameters output by the pressure-controlled prediction model to construct a final state deviation set. Based on the pressure-controlled behavior feedback matrix and final state deviation data, the system performs error backtracking analysis and updates the historical forging sample database to further optimize the height control prediction model and pressure-controlled curves, achieving closed-loop intelligent evolution of the system.

[0142] Figure 3 A diagram showing a folding defect of a valve housing forging provided in an embodiment of the present application; Figure 4 A diagram showing the steps of the die pressing process for the valve housing forging provided in an embodiment of the present application; Figure 5 A schematic diagram of the flattening of a valve housing forging provided in an embodiment of the present application; Figure 6 Schematic diagram of the valve housing forging provided in the embodiment of the present application being flattened by a height limiting block.

[0143] Beneficial effects of this embodiment: The present invention provides a valve housing forging height limit block controlled flattening system. By building an intelligent control link for the entire process, from initial state modeling to final state structural verification, this system achieves highly refined and intelligent control of the forging height limit flattening process. This system has significant benefits in the following aspects: 1. Improved control accuracy: By building a pressure control prediction model based on actual operating conditions and integrating a dynamic deviation correction mechanism, highly responsive adjustment of the height limit block control behavior is achieved, significantly reducing forming errors.

[0144] 2. Enhanced system adaptability: This system builds a model learning closed-loop mechanism through structural matching analysis and model optimization modules. It can automatically update the control strategy based on the feedback of each forging task, and effectively respond to the pressure control requirements of multi-material and multi-working condition forgings.

[0145] 3. Improved pressure control safety: Dynamically determine the correction amount of the height limit block and automatically constrain the adjustment range, effectively preventing overpressure or structural damage to forgings caused by overadjustment, and improving system safety and robustness.

[0146] 4. Production consistency optimization: Structured collection of pressure control behavior data and final shape data enables the system to continuously improve control accuracy, optimize the forming consistency of subsequent batches, and reduce the frequency of manual debugging.

[0147] 5. Strong scalability: Each functional module of the system has independence and interface compatibility, which makes it easy to deploy in different forging equipment and control platforms, and has good engineering promotion value.

[0148] Obviously, the embodiments described above are only some of the embodiments of the present application, rather than all of the embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions described in the aforementioned specific embodiments, or to make equivalent replacements for some of the technical features therein. Any equivalent structure made using the contents of the present application specification and the accompanying drawings, directly or indirectly used in other related technical fields, is also within the scope of patent protection of the present application.

Claims

1. A method for controlling the flattening of a valve housing forging height limit block, characterized in that: The following steps are involved: Obtaining the initial outer diameter, length, wall thickness, surface temperature, internal temperature, material yield strength, elastic modulus, and thermal expansion coefficient of the forging, constructing a dimensional parameter set and a physical parameter set, and normalizing the sets to generate an initial working condition feature vector; Based on the initial working condition feature vector and a forging sample database containing historical working condition samples and pressure control records, a height limit block control model is constructed to output a target pressure control curve and initial height limit block setting parameters; Collect real-time flattening force, pressure control stroke, real-time temperature and equivalent plastic strain to build process parameter sequence; Dynamically comparing the process parameter sequence with the target pressure control curve to generate a pressure control deviation sequence, and constructing a deviation analysis result set through trend extraction and tolerance judgment; Calculate the corrected displacement parameters of the height limit block based on the deviation analysis result set, real-time temperature and material parameters, and generate a control instruction including a control type, an adjustment range and an execution mode; Executing the control instruction to complete the position correction of the height limit block, collecting the adjusted voltage control behavior parameters, and constructing a correction feedback voltage control matrix; Comparing the feedback pressure control matrix with the final structure detection parameters of the forging to construct a structure matching deviation set; Error backtracking and sample encoding are performed based on the initial eigenvector, feedback matrix and structural deviation set, the forging sample database is updated, and the control model and target pressure control curve are optimized.

2. The method according to claim 1, characterized in that The step of generating the initial operating condition characteristic vector comprises: The size parameter set and physical parameter set are normalized and feature fused respectively, and the principal component extraction method is used to construct the feature vector dimensional expression structure.

3. The method according to claim 1, characterized in that The height limit block control model is a multivariable nonlinear regression model or a regression prediction model based on a neural network, and is pre-trained using historical samples with annotated final state deviations.

4. The method according to claim 1, wherein The collected real-time pressure control stroke and target pressure control curve are aligned in a dynamic window manner to construct a pressure control deviation sequence, and smoothing filtering and cumulative offset processing are performed.

5. The method according to claim 1, wherein The calculation process of the height limit block correction displacement parameter adopts a weighted factor fusion method to fuse the average pressure control deviation, material yield strength and temperature gradient to construct a dynamic control weight.

6. The method according to claim 1, characterized in that The control instruction includes a control type field, a correction direction field, a correction amplitude field and an execution mode field, and is transmitted to the height limit control unit through a standard industrial communication interface.

7. The method according to claim 1, characterized in that The rows of the feedback pressure control matrix are sampling timing, and the columns are corrected flattening force, stroke, temperature and deformation rate. The data are filtered, merged and structured.

8. The method according to claim 1, characterized in that The final structural detection parameters of the forging include the final outer diameter, length and wall thickness after forming, and are collected using a three-dimensional optical scanning device.

9. The method according to claim 1, characterized in that The error backtracking and sample updating process includes structural deviation clustering, control behavior association and label supplementation operations, which are used to expand the multi-condition coverage of historical samples.

10. A valve housing forging height limit block controlled flattening system, characterized in that: include: Initial working condition acquisition module, which obtains and encodes the initial parameters of the forging and generates the working condition feature vector; Control model construction module to generate pressure control prediction model and target pressure control curve; Real-time voltage control monitoring module, collects and calculates voltage control deviation sequence; The height limit correction control module generates a height limit correction control instruction based on the deviation analysis results; Execute the feedback collection module to complete the height limit block adjustment and feedback data collection; The model optimization module is used to perform structural matching analysis and model updating.

Images