A multi-directional die forging mass data analysis method and system

Abstract

The present application relates to the technical field of multi-directional die forging quality analysis, and provides a multi-directional die forging quality big data analysis method and system, which comprises collecting pressure, temperature and displacement time sequence data of each loading direction of multi-directional die forging and product defect size, and constructing process and quality corresponding data source. Then, the parameter difference is calculated to generate an interaction parameter set to capture the interaction of multi-directional loading. The forgings are divided into continuous spatial grids, the action area and overlapping proportion of each loading direction are determined based on the geometric relationship of the die, and the accurate mapping of the process parameters and the spatial position is established; the equivalent process parameter sequence is generated by taking the action proportion as the weight to quantify the contribution difference of different loading directions to the same position. Finally, the grid quality score is calculated by comprehensively considering the process deviation, loading interaction dispersion and defect size and taking the average, so as to realize the accurate full-space coverage evaluation of the multi-directional die forging quality and provide a clear basis for process optimization.

Description

Technical Field

[0001] This invention relates to the field of multi-directional forging quality analysis technology, and more specifically, to a multi-directional forging quality big data analysis method and system. Background Technology

[0002] The content in this section only provides background information related to this invention and may not constitute prior art.

[0003] Multi-directional forging is a precision forging technology that uses multiple loading directions to apply forces simultaneously or sequentially, causing metal billets to undergo plastic deformation in complex mold cavities. It can form forgings with complex internal cavities, multi-directional bosses, and irregular structures in one step, significantly improving material utilization and the mechanical properties of forgings. It is widely used in high-end manufacturing fields such as aerospace, automobile manufacturing, engineering machinery, and energy equipment, and is one of the core processes supporting the manufacturing of major national equipment.

[0004] In recent years, with the rapid development of industrial big data technology, more and more enterprises and research institutions have begun to apply data analysis technology to the quality control of multi-directional forging, establishing quality prediction and evaluation models by collecting various process parameter data during the production process. However, most existing big data analysis methods for multi-directional forging quality directly adopt the analysis framework of traditional unidirectional forging, failing to fully adapt to the core technical characteristics of multi-directional forging, and have a series of problems that urgently need to be solved: Existing methods usually only analyze the process parameters of each loading direction independently, ignoring the complex interactions and coupling effects between different directions during multi-directional loading, resulting in an inability to comprehensively and accurately reflect the actual forming conditions of the forging; Existing methods mostly evaluate quality from the overall level of the forging, lacking a refined analysis of the differences in forming conditions in different spatial regions of the forging, making it difficult to accurately explain the generation mechanism of local defects; In addition, the quality evaluation system of existing methods is relatively singular in dimension, failing to comprehensively consider the influence of multiple key factors such as the degree of deviation of process parameters, the stability of the loading process, and the actual defect status, resulting in insufficient accuracy and reliability of quality evaluation results, and failing to provide an effective scientific basis for the optimization and adjustment of process parameters.

[0005] Therefore, there is an urgent need for a multi-directional forging quality analysis method that can fully reflect the characteristics of multi-directional forging process, achieve refined spatial quality analysis, and integrate multi-dimensional influencing factors. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-directional forging quality big data analysis method and system to improve the aforementioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows: Firstly, this application provides a multi-directional forging quality big data analysis method, including: Collect time-series data of each loading direction during the multi-directional forging process, as well as defect size data of each inspection location of the finished forging; the time-series data includes pressure time-series data, temperature time-series data, and displacement time-series data; The timing data of each loading direction are aligned according to each sampling point on the time axis to generate a multi-directional process parameter combination for each sampling moment; Based on the combination of multi-directional process parameters, the process parameter difference between each loading direction is calculated to generate the loading direction interaction parameter set at each sampling time. The forging product is divided into multiple continuous spatial grid units; based on the die loading geometry of multi-directional forging, the effective area of ​​each loading direction on the forging product is determined; the effective area of ​​each spatial grid unit is overlapped with the effective area of ​​each loading direction to obtain the effective proportion of the loading direction corresponding to each spatial grid unit; Using the proportion of the loading direction effect of each spatial grid cell as the weight, the parameters in the multi-directional process parameter combination are weighted to generate the equivalent process parameter time series of the corresponding spatial grid cell; The root mean square error between the time series of equivalent process parameters and the time series of preset standard process parameters is calculated as the deviation. The standard deviation of the concentrated pressure difference of the interaction parameters in the loading direction during the forming period is calculated as the interaction dispersion. The deviation, interaction dispersion and corresponding defect size of the spatial grid cell are weighted and summed to obtain the quality score of the spatial grid cell. The average of the quality scores of all spatial grid cells is taken as the multi-directional forging quality evaluation result.

[0007] Furthermore, the step of aligning the timing data of each loading direction according to each sampling point on the time axis specifically includes: Using the time series data with the highest sampling rate as a benchmark, other time series data are resampled using cubic spline interpolation to ensure that all directions have data at the same timestamp. The interpolation nodes are determined based on the time points of the benchmark sequence.

[0008] Furthermore, the steps for calculating the process parameter differences between each loading direction specifically include: At each sampling moment, the pressure value, temperature value, and displacement value corresponding to any two different loading directions are subtracted to obtain the pressure difference, temperature difference, and displacement difference, respectively. The above differences between all two different directions at each sampling moment constitute the loading direction interaction parameter set.

[0009] Furthermore, the step of dividing the finished forging into multiple continuous spatial grid units specifically includes: A three-dimensional solid model of the forging product is obtained, and a hexahedral mesh is used to mesh the three-dimensional solid model.

[0010] Furthermore, the steps for determining the effective area of ​​each loading direction on the finished forging include: For each loading direction, taking the contact surface between the mold and the forging as the reference plane, a virtual prism is generated along the loading direction that encompasses the entire projection of the forging in that direction. The Boolean intersection operation is performed between the prism and the three-dimensional model of the forging, and the intersection volume is the area of ​​action in that direction.

[0011] Furthermore, the steps for performing a Boolean intersection operation between the prism and the three-dimensional model of the forging specifically include: The virtual prism and the three-dimensional model of the forging are spatially superimposed, retaining the space occupied by both, and removing the parts of the virtual prism that do not overlap with the forging and the parts of the forging that do not overlap with the virtual prism. The remaining part is the area of ​​influence in this direction.

[0012] Furthermore, following the results of the multi-directional forging quality evaluation, the following are also included: Spatial mesh cells with quality scores below a preset threshold are marked as quality anomaly areas. The location and extent of all quality anomaly areas are displayed on the three-dimensional solid model of the forging. At the same time, the time series of equivalent process parameters and the set of interaction parameters of loading direction corresponding to each quality anomaly area are recorded.

[0013] Secondly, this application also provides a multi-directional forging quality big data analysis system, including: The data acquisition module is used to collect time-series data of each loading direction during the multi-directional forging process, as well as defect size data of each inspection position of the finished forging; the time-series data includes pressure time-series data, temperature time-series data, and displacement time-series data; The timing alignment module is used to align the timing data of each loading direction according to each sampling point on the time axis, and generate a multi-directional process parameter combination for each sampling moment; The interactive parameter module is used to calculate the process parameter difference between each loading direction based on the combination of multi-directional process parameters, and generate the loading direction interactive parameter set for each sampling time. The proportion calculation module divides the finished forging into multiple continuous spatial grid units; based on the die loading geometry of multi-directional forging, it determines the area of ​​action of each loading direction on the finished forging; it calculates the overlap between each spatial grid unit and the area of ​​action of each loading direction to obtain the proportion of action of each loading direction corresponding to each spatial grid unit. The equivalent parameter module is used to weight each parameter in the multi-directional process parameter combination with the proportion of the loading direction of each spatial grid cell as the weight, and generate the equivalent process parameter time series of the corresponding spatial grid cell. The quality evaluation module is used to calculate the root mean square error between the time series of equivalent process parameters and the time series of preset standard process parameters, as the deviation; calculate the standard deviation of the concentrated pressure difference of the interaction parameters in the loading direction during the forming period, as the interaction dispersion; weight and sum the deviation, interaction dispersion and corresponding defect size of the spatial grid cell to obtain the quality score of the spatial grid cell; and take the average of the quality scores of all spatial grid cells as the multi-directional forging quality evaluation result.

[0014] Thirdly, this application also provides an electronic device, including: Memory, used to store computer programs; A processor is used to implement the method steps as described in the first aspect when executing a computer program.

[0015] Fourthly, this application also provides a readable storage medium on which a computer program is stored, which, when executed by a processor, implements the method steps of the first aspect.

[0016] The beneficial effects of this invention are as follows: This invention provides a complete process-quality data source for quality analysis by collecting time-series data on pressure, temperature, and displacement in each loading direction of multi-directional forging, as well as finished product defect size data. It generates multi-directional process parameter combinations by aligning the time-series data of different loading directions with sampling points, solving the problem of asynchronous multi-directional loading timing and enabling correlation analysis of multiple parameters at the same time. By calculating the differences in process parameters between each loading direction, it generates an interaction parameter set, capturing the interaction of multi-directional loading and uncovering interactive influencing factors neglected by traditional single-parameter analysis. Finally, it divides the forging into continuous spatial grid units, determines the effective area of ​​each loading direction based on the die loading geometry, and performs overlapping calculations to obtain the proportion of loading direction effect in each grid unit. A precise mapping between process parameters and the spatial position of forgings was established, enabling spatial refinement of quality analysis. Equivalent process parameter time series were generated by weighting multi-directional process parameters with their respective contribution proportions, quantifying the differences in contribution of different loading directions to the same spatial position, and obtaining a process sequence that accurately reflects the actual forming conditions at that position. Finally, the root mean square error between the equivalent sequence and the standard sequence was calculated as the deviation, and the standard deviation of the pressure difference during the forming period was used as the interaction dispersion. Combined with the defect size, a weighted summation was obtained to get a grid quality score, which was then averaged to obtain an overall evaluation. This comprehensive quantification from three dimensions—process deviation, loading interaction, and actual defects—achieved a precise, full-space coverage evaluation of multi-directional die forging quality, providing a clear basis for process optimization. Attached Figure Description

[0017] Figure 1 A flowchart of a multi-directional forging quality big data analysis method provided by the present invention; Figure 2A schematic diagram of a multi-directional forging quality big data analysis system provided by the present invention; Figure 3 This is a schematic diagram of an electronic device provided by the present invention.

[0018] In the diagram: 201, Data Acquisition Module; 202, Timing Alignment Module; 203, Interaction Parameter Module; 204, Proportion Calculation Module; 205, Equivalent Parameter Module; 206, Quality Evaluation Module; 301, Processor; 302, Memory. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] like Figure 1 As shown in the embodiment of the present invention, a multi-directional forging quality big data analysis method includes: S101 collects time-series data of each loading direction during the multi-directional forging process, as well as defect size data of each inspection position of the finished forging; the time-series data includes pressure time-series data, temperature time-series data, and displacement time-series data.

[0021] Specifically, multi-directional forging processes typically include a vertical main loading direction and multiple horizontal lateral loading directions, each driven by an independent hydraulic cylinder or mechanical pressure head. In practice, strain gauge pressure sensors and magnetostrictive displacement sensors are integrated into the actuators in each loading direction, and armored thermocouples are embedded near the interface between the die and the forging. All sensors are triggered by a unified clock source through a multi-channel synchronous data acquisition system, synchronously recording pressure, temperature, and displacement signals according to a preset sampling frequency, thereby obtaining pressure timing data, temperature timing data, and displacement timing data.

[0022] The time-series data specifically includes pressure time-series data, temperature time-series data, and displacement time-series data. These three types of time-series data are collected because pressure time-series data directly reflects the dynamic balance between material deformation resistance and loading force; temperature time-series data affects metal flow stress and microstructure evolution; and displacement time-series data quantitatively describes the die filling process and forging dimensions. Together, these three constitute a complete state vector describing the multi-directional loading process. Synchronous acquisition can fully preserve the coordination relationship and time lag information of loading in each direction, providing a real and continuous process input for subsequent steps to calculate multi-directional process parameter combinations and loading direction interaction parameter sets.

[0023] Meanwhile, after the forging is completed and cooled, a phased array ultrasonic detector or industrial computed tomography (CT) device is used to scan the critical load-bearing section of the forging point by point to obtain defect size data such as the equivalent diameter or depth of defects at each detection location, which serves as the target label for quality evaluation. For example, for the multi-directional die forging of a three-way valve body, there are three loading directions: the vertical downward main punch direction and the left and right horizontal directions. A strain gauge pressure sensor with a range-adaptive design is installed on the hydraulic cylinder in each direction, a magnetostrictive displacement sensor is fixed at the punch, and a sheathed thermocouple is set on the surface of the corresponding mold cavity. All channels are simultaneously acquired at a sampling rate of 100 Hz to obtain the pressure, temperature, and displacement time series for the vertical, left horizontal, and right horizontal directions. After the forging is processed, ultrasonic phased array detectors are used to detect pore or crack-like defects at 10 detection locations, such as the inner wall of the valve body cross hole, and the maximum defect size is recorded. This data acquisition method accurately correlates the multidimensional dynamic information of the forming process with the final quality defects, enabling subsequent steps to establish a mapping between process parameters and quality scores based on the real physical process, thereby improving the effectiveness and practicality of multi-directional die forging quality analysis.

[0024] S102 aligns the timing data of each loading direction according to each sampling point on the time axis to generate a multi-directional process parameter combination for each sampling moment.

[0025] Specifically, the purpose of this alignment operation is to eliminate timestamp misalignment caused by differences in the response characteristics of various sensors, data acquisition channel transmission delays, or different sampling rates used in different loading directions, thereby ensuring that a complete process state description in all directions can be obtained at the same time. If unaligned data is used directly, the time offset will be misjudged as process fluctuations when calculating the pressure difference, temperature difference, and displacement difference between loading directions, resulting in distortion of interactive parameters.

[0026] During the alignment process, the time series data with the highest sampling rate is used as the benchmark. Other time series data are resampled using cubic spline interpolation to ensure data is available in all directions at the same timestamp. The interpolation nodes are determined based on the time points of the benchmark sequence. The highest sampling rate sequence is chosen as the benchmark because its temporal granularity is the finest, preserving the transient details of the forming process to the greatest extent and avoiding information loss. Cubic spline interpolation resampling utilizes piecewise cubic polynomials to construct smooth curves with continuous first and second derivatives between known data points. When calculating the interpolation results at the benchmark time points, it accurately reproduces the original measurement trend without introducing the abrupt changes common in linear interpolation, thus protecting the smoothness and differentiability of physical quantities such as pressure, temperature, and displacement in the time dimension. The interpolation nodes are directly taken from each timestamp of the benchmark sequence; that is, using the timetable of the benchmark sequence as a target, the corresponding values ​​are calculated point-by-point from the spline curves of other time series, ensuring strict alignment of data in all directions on the time axis.

[0027] S103, based on the combination of multi-directional process parameters, calculate the process parameter difference between each loading direction and generate the loading direction interaction parameter set for each sampling time. Specifically, in multi-directional forging, there is a complex mechanical coupling and heat transfer coordination relationship between the vertical main loading direction and each horizontal lateral direction. The pressure, temperature, or displacement curves in a single direction cannot reflect this cross-directional interaction. This step quantifies the differences in process parameters in different directions at each time step, explicitly extracting the antagonistic, hysteretic, and unbalanced states between directions during the forming process. This provides direct dynamic characteristic evidence for subsequent accurate determination of loading coordination quality and defect causes.

[0028] The specific steps for calculating the process parameter differences between various loading directions include: at each sampling moment, subtracting the pressure, temperature, and displacement values ​​corresponding to any two different loading directions to obtain the pressure difference, temperature difference, and displacement difference, respectively. Taking a three-way valve body multi-directional forging with three loading directions (vertical main punch, left horizontal, and right horizontal) as an example, at a certain sampling moment after alignment processing, the vertical pressure value, left horizontal pressure value, and right horizontal pressure value are read from the multi-directional process parameter combination, and then the processor performs a simple subtraction operation to obtain the pressure difference between the vertical and left horizontal directions, the vertical and right horizontal directions, and the left and right horizontal directions. The temperature difference and displacement difference are generated using the same pairwise subtraction method. The principle behind this differential calculation is as follows: the pressure difference directly reflects the degree of imbalance in the flow resistance of metal in different areas of the cavity. If the vertical pressure remains normal at a certain moment while the horizontal pressure on one side drops sharply, the difference will increase instantaneously, indicating that filling on that side is obstructed or an abnormal pressure release has occurred. The temperature difference reveals the spatial unevenness of the heat exchange rate at the contact interfaces of each mold. A persistently large difference can easily cause local overcooling or overheating, affecting the uniformity of the microstructure. The displacement difference measures the synchronicity of the punch advancement in each direction in real time. An unexpected increase in the difference indicates a disorder in the filling sequence, which can induce folding or incomplete filling defects. This calculation only uses synchronously acquired time-series data, requires no additional sensing devices, has a small computational load, and is easy to implement in engineering controllers or edge computing nodes, making it highly feasible for deployment.

[0029] Subsequently, the differences between all the aforementioned two different directions at each instant constitute a loading direction interaction parameter set. That is, at the sampling instant, the pressure difference, temperature difference, and displacement difference values ​​of all obtained direction pairs are merged and packaged to form a feature vector characterizing the multi-directional interaction state at that instant, and stored in association with a time label, forming an interaction parameter dataset of the same length as the original time series. In implementation, the direction pair identifiers and their corresponding three types of differences can be organized into a structured data table according to time series, which can be directly used in subsequent steps. This interaction parameter set completely reproduces the cooperative or antagonistic relationship between loading directions at each instant, allowing the standard deviation of the pressure difference to be naturally extracted to quantify the stability of directional coordination during the forming process when calculating the interaction dispersion. This liberates quality scoring from the limitation of relying solely on a single parameter deviation, effectively improving the explanatory power and predictive accuracy of quality fluctuations in multi-directional loading forgings.

[0030] S104, the finished forging is divided into multiple continuous spatial grid units; based on the die loading geometry of multi-directional forging, the action area of ​​each loading direction on the finished forging is determined; the action area of ​​each spatial grid unit is overlapped with the action area of ​​each loading direction to obtain the action ratio of the loading direction corresponding to each spatial grid unit.

[0031] Specifically, to achieve a refined mapping from macroscopic loading conditions to the microscopic regions within the forging, the first step is to divide the finished forging into multiple continuous spatial mesh units. In practice, a three-dimensional solid model of the finished forging is obtained. This model can be directly exported from a computer-aided design system or obtained by reverse reconstruction of the physical object using industrial computed tomography (CT) equipment. A hexahedral meshing method is used to mesh the three-dimensional solid model. The principle is that hexahedral elements possess a regular spatial topology and a consistent normal direction. Compared to tetrahedral elements, when performing directional projection and volume percentage calculations, they can naturally form a regular geometric tangent relationship with the loading direction, thus avoiding ambiguity in direction determination caused by element shape distortion. After meshing, the forging is discretized into a series of seamlessly connected three-dimensional spatial mesh units. Each unit serves as the minimum spatial carrier for subsequent process parameter assignment and quality evaluation, providing a deterministic spatial reference for establishing a precise mapping relationship between process parameters and defect locations.

[0032] Based on this, the effective area of ​​each loading direction on the finished forging is determined. The purpose is to concretize the abstract loading direction into a quantifiable three-dimensional influence range within the forging. For each loading direction, using the contact surface between the mold and the forging as the reference plane, a virtual prism is generated along the loading direction that encompasses the entire projection of the forging in that direction. The technical essence of this operation is to construct an infinitely extending prism that completely encloses the outer contour of the forging, using the contact surface as the initial boundary of the force flow input and the loading direction as the tension vector. Subsequently, a Boolean intersection operation is performed between this prism and the three-dimensional model of the forging, and the intersection volume is the effective area of ​​that direction. Specifically, the Boolean intersection operation between the prism and the three-dimensional model of the forging is achieved through spatial superposition calculation: the virtual prism and the three-dimensional model of the forging are spatially superimposed, retaining the space occupied by both, and removing the parts of the virtual prism that do not overlap with the forging, as well as the parts of the forging that do not overlap with the virtual prism. The remaining part is the effective area of ​​that direction. This Boolean intersection process precisely simulates the transmission path of the loading force in that direction within the forging at the geometric level. That is, only the area that is both inside the metal entity and within the envelope of the force flow in the loading direction will be dominated by the loading in that direction. Through this geometric trimming, the boundary of the area of ​​action is completely determined by the geometric features of the mold contact surface and the loading direction, avoiding the arbitrariness brought about by human experience division, so that the boundaries of the influence range of each direction inside the forging are clear and without overlap or omission.

[0033] After obtaining the effective area of ​​each spatial grid cell and each loading direction, an operation is performed to overlap the effective area of ​​each spatial grid cell with the effective area of ​​each loading direction to obtain the percentage of the load direction effect corresponding to each spatial grid cell. In specific implementation, each spatial grid cell is traversed, and the volume of the part of the cell intersecting with the effective area of ​​the vertical main loading direction and the effective areas of each horizontal lateral loading direction is calculated. Then, the volume is divided by the total volume of the spatial grid cell itself to obtain the percentage quantitative result of the degree of load effect of each direction on the cell. Taking a three-way valve body forging as an example, a certain grid cell near the inner wall of the main pipe may completely overlap with the effective area of ​​the vertical main punch direction, but only partially intersect with the effective areas of the left and right horizontal directions. In this case, the effective percentage of the cell is recorded as 100% in the vertical direction, 20% in the left horizontal direction, and 80% in the right horizontal direction. This calculation process achieves a seamless connection from directional macroscopic loading to gridded microscopic classification, enabling the process state at each spatial location to be decomposed into the superposition result of multiple loading directions. This provides a precise quantitative basis for subsequent steps to integrate multi-directional process parameters with the proportion of action as weight, effectively supporting the in-depth analysis of the internal mass distribution law of complex forgings subjected to multi-directional forces.

[0034] The formula for calculating the proportion of the effect of the loading direction is as follows: For any spatial grid cell and arbitrary loading direction The grid cell is affected by the loading direction. The percentage of the effect is: (1) In the formula, For spatial grid units Loading direction The proportion of the effect; For spatial grid units With loading direction The intersection volume of the regions of action; For spatial grid units Its total volume; Number the spatial grid cells; Number the loading direction.

[0035] S105, using the proportion of the loading direction effect of each spatial grid cell as the weight, weights each parameter in the multi-directional process parameter combination to generate the equivalent process parameter time series of the corresponding spatial grid cell.

[0036] Specifically, the weight is based on the proportion of the loading direction of each spatial grid cell. The principle is to transform the uneven influence of multi-directional loading on different positions inside the forging into a quantifiable contribution coefficient. These proportions have been accurately obtained in step S104 through Boolean intersection operation between the virtual prism and the forging model, as well as the calculation of the overlap of grid cell volume. They directly reflect the actual degree of participation of the force flow in each loading direction in the deformation and thermal process of the cell material.

[0037] When weighting the parameters in a multi-directional process parameter combination, each sampling time is traversed. Pressure, temperature, and displacement values ​​in the vertical main loading direction and each horizontal lateral loading direction are read from the aligned multi-directional process parameter combination. These values ​​are then multiplied by the proportion of each direction's effect corresponding to that spatial grid cell. Parameters of the same type are then accumulated by direction to obtain the equivalent pressure, equivalent temperature, and equivalent displacement of that cell at that time. This weighting process is not a simple averaging but a weighted fusion based on the physical range of action. This ensures that the equivalent parameters accurately reflect the comprehensive process state of the spatial point under the coupling effect of multi-directional loads, avoiding local information distortion caused by using only a single-directional parameter.

[0038] After generating the equivalent process parameter time series for the corresponding spatial grid cell, the process curves originally recorded dispersedly according to the loading direction are transformed into the unique full-process equivalent process history of that cell, providing a continuous input that precisely matches the physical stress characteristics of that point for subsequent deviation calculations. Taking a three-way valve body forging as an example, a spatial grid cell on the inner wall of the main pipeline is determined in step S104 to have a vertical force ratio of 100%, a left horizontal force ratio of 20%, and a right horizontal force ratio of 80%. At a certain sampling moment, the vertical pressure, left horizontal pressure, and right horizontal pressure are extracted from the multi-directional process parameter combination and then weighted and accumulated with weights of 1.00, 0.20, and 0.80 to generate the equivalent pressure of that cell at that moment. The same weighting operation is performed on temperature and displacement. By traversing all sampling moments, the equivalent pressure time series, equivalent temperature time series, and equivalent displacement time series of that cell are obtained. This fusion method, which uses the proportion of effect as a weight, quantitatively injects the synergistic effect of multi-directional loading into the process description of each spatial grid cell. This enables the deviation calculated based on the time series of equivalent process parameters to sensitively capture local quality anomalies caused by directional synergistic imbalance, thereby improving the precision and interpretability of multi-directional forging quality evaluation.

[0039] The formula for equivalent pressure is as follows: At any sampling time Arbitrary spatial grid cell The equivalent pressure is: (2) In the formula, Sampling time Spatiotemporal grid cells The equivalent pressure; For spatial grid units Loading direction The proportion of the effect; Sampling time Loading direction The pressure value after alignment; For the first Each sampling time; Number the spatial grid cells; Number the loading direction.

[0040] The formula for equivalent temperature is as follows: At any sampling time Arbitrary spatial grid cell The equivalent temperature is : (3) In the formula, Sampling time Loading direction The aligned temperature value.

[0041] The formula for the equivalent displacement is as follows: At any sampling time Arbitrary spatial grid cell The equivalent displacement is : (4) In the formula, Sampling time Loading direction The displacement value after alignment.

[0042] S106, calculate the root mean square error between the time series of equivalent process parameters and the time series of preset standard process parameters, as the deviation; calculate the standard deviation of the concentrated pressure difference of the interaction parameters in the loading direction during the forming period, as the interaction dispersion; weight and sum the deviation, interaction dispersion and corresponding defect size of the spatial grid cell to obtain the quality score of the spatial grid cell; take the average of the quality scores of all spatial grid cells as the multi-directional forging quality evaluation result.

[0043] Specifically, the root mean square error (RMSE) between the equivalent process parameter time series and the preset standard process parameter time series is calculated as the deviation. The specific implementation is as follows: For a given spatial grid cell, the processor calculates the square of the difference between its equivalent pressure sequence and the standard pressure sequence at each sampling time. The mean of this squared sequence over the entire forming period is then calculated, and the square root is taken to obtain the pressure deviation component. The same operation is performed on the equivalent temperature and equivalent displacement sequences. The RMSE results of the three components are then arithmetically averaged or weighted averaged to form the comprehensive deviation of the cell. The principle is that the RMSE is extremely sensitive to the amplitude and duration of the deviation of process parameters from the standard waveform in the time dimension. It can capture both instantaneous and severe deviations and cumulatively reflect persistent small drifts. Therefore, the larger the deviation value, the more significant the deviation between the actual forming process at that grid location and the ideal process specification. The standard process parameter time series can be derived from the statistical mean of the equivalent parameters of the corresponding grid in historical batches of qualified forgings, or from the ideal loading curve extracted at that grid by finite element process simulation, ensuring the rationality of the reference benchmark process.

[0044] The formulas for each deviation component (root mean square error) are as follows: Arbitrary spatial grid cell The pressure deviation component is: (5) In the formula, For spatial grid units Pressure deviation component; This represents the total number of sampling points; Sampling time Spatiotemporal grid cells The equivalent pressure; For spatial grid units Standard pressure time series at sampling time The value at; For the first Each sampling time; Number the spatial grid cells.

[0045] Arbitrary spatial grid cell Temperature deviation component : (6) In the formula, Sampling time Spatiotemporal grid cells The equivalent temperature; For spatial grid units Standard temperature time series at sampling time The value at that location.

[0046] Arbitrary spatial grid cell The displacement deviation component is : (7) In the formula, Sampling time Spatiotemporal grid cells The equivalent displacement; For spatial grid units The standard displacement time series at the sampling time The value at that location.

[0047] The overall deviation is: (8) Simultaneously, the standard deviation of the pressure difference values ​​in the interaction parameter set of the loading directions during the forming period is calculated as the interaction dispersion. In specific implementation, the time series of all pressure difference values ​​for each direction pair corresponding to the grid cell are extracted from the interaction parameter set. The dispersion of these differences around their respective mean values ​​throughout the entire forming period is calculated, and the average of the standard deviations of each direction pair is taken as the interaction dispersion of the cell. This index is specifically used to quantify the stability of the force-flow balance state during multi-directional loading: the smaller the standard deviation, the more stable the pressure resistance relationship between the directions, and the more orderly the metal filling flow; if the standard deviation is too large, it reflects that the grid position experienced repeated and violent fluctuations in pressure between directions during the forming period, which can easily induce defects such as turbulence, folding, or local incomplete filling.

[0048] The formula for the pressure difference sequence is as follows: Any two different loading directions and ( The average value of the pressure difference sequence during the entire molding period is : (9) In the formula, This represents the total number of sampling points; Sampling time Loading direction With loading direction The pressure difference between them.

[0049] Any two different loading directions and ( The standard deviation of the pressure difference sequence during the entire molding period is: : (10) Arbitrary spatial grid cell Interaction Dispersion The average of the standard deviations of the pressure difference series in all directions: (11) In the formula, This represents the total number of loading directions. For loading direction With loading direction The standard deviation of the pressure difference series; These are the loading direction numbers; Number the spatial grid cells.

[0050] After obtaining the deviation and interaction dispersion, the deviation, interaction dispersion, and corresponding defect size of the spatial grid cell are weighted and summed to obtain the quality score of the spatial grid cell. The defect size here refers to the maximum equivalent diameter or depth of the defect measured at the detection location of the grid cell in step S101. During weighting, the weights are determined based on principal component analysis of process experimental design or historical data to reflect the relative importance of deviation, interaction dispersion, and defect size on the final quality, ensuring that the score comprehensively expresses the combined effects of process fluctuations, directional misalignment, and existing defects. Finally, the average quality score of all spatial grid cells is taken as the multi-directional forging quality evaluation result, thus achieving an objective quantification of the uniformity of the internal quality distribution of the entire forging with a single numerical value.

[0051] Among them, arbitrary spatial grid cells Quality rating for: (12) In the formula, This is the weighting coefficient for the overall deviation. These are the weighting coefficients for the interaction dispersion; The weighting factor is the defect size. For spatial grid units The corresponding defect size data.

[0052] The overall quality evaluation result of the multi-directional die forging is the average of the quality scores of all spatial mesh elements. : (13) In the formula, This represents the total number of spatial grid cells.

[0053] After obtaining the evaluation results, spatial mesh cells with quality scores below a preset threshold are marked as quality anomaly areas. The location and extent of all quality anomaly areas are displayed on the 3D solid model of the forging, and the equivalent process parameter time series and loading direction interaction parameter set corresponding to each quality anomaly area are recorded. This preset threshold can be set according to the material's allowable defect tolerance or process capability index requirements. The 3D display uses color mapping, highlighting the anomaly areas with warning colors on the surface or section plane of the solid model, allowing process engineers to intuitively locate risky areas. Recording its equivalent process parameters and interaction parameters provides a complete dynamic evidence chain for tracing the cause of defects. It allows direct comparison of the equivalent pressure, temperature, displacement differences, and directional pressure difference fluctuation characteristics between the anomaly area and the normal area at each stage of forming, thereby quickly identifying whether the quality hazard is caused by equipment control response lag, local overheating of the mold, or improper multi-directional loading timing settings.

[0054] like Figure 2 As shown, based on the same inventive concept, this embodiment provides a multi-directional forging quality big data analysis system, including: The data acquisition module 201 is used to acquire time-series data of each loading direction during the multi-directional forging process, as well as defect size data of each detection position of the finished forging; the time-series data includes pressure time-series data, temperature time-series data, and displacement time-series data; The timing alignment module 202 is used to align the timing data of each loading direction according to each sampling point on the time axis, and generate a multi-directional process parameter combination for each sampling moment; The interaction parameter module 203 is used to calculate the process parameter difference between each loading direction based on the combination of multi-directional process parameters, and generate the loading direction interaction parameter set for each sampling time. The proportion calculation module 204 divides the finished forging into multiple continuous spatial grid units; determines the area of ​​action of each loading direction on the finished forging based on the die loading geometry of multi-directional forging; and calculates the overlap between each spatial grid unit and the area of ​​action of each loading direction to obtain the proportion of action of each loading direction corresponding to each spatial grid unit. The equivalent parameter module 205 is used to weight each parameter in the multi-directional process parameter combination with the proportion of the loading direction of each spatial grid cell as the weight, and generate the equivalent process parameter time series of the corresponding spatial grid cell. The quality evaluation module 206 is used to calculate the root mean square error between the time series of equivalent process parameters and the time series of preset standard process parameters, as the deviation; calculate the standard deviation of the concentrated pressure difference of the interaction parameters in the loading direction during the forming period, as the interaction dispersion; weight and sum the deviation, interaction dispersion and corresponding defect size of the spatial grid cell to obtain the quality score of the spatial grid cell; and take the average of the quality scores of all spatial grid cells as the multi-directional forging quality evaluation result.

[0055] like Figure 3 As shown, based on the same inventive concept, this embodiment provides an electronic device, including: Memory 302 is used to store computer programs; Processor 301 is used to implement the method steps as described in the first aspect when executing a computer program.

[0056] Based on the same inventive concept, this embodiment provides a readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it implements the method steps as described in the first aspect.

[0057] It should be noted that the specific methods by which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0058] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-directional die forging quality big data analysis method, characterized in that, include: Collect time-series data of each loading direction during the multi-directional forging process, as well as defect size data of each inspection location of the finished forging; The time-series data includes pressure time-series data, temperature time-series data, and displacement time-series data; The timing data of each loading direction are aligned according to each sampling point on the time axis to generate a multi-directional process parameter combination for each sampling moment; Based on the multi-directional process parameter combination, the process parameter difference between each loading direction is calculated to generate the loading direction interaction parameter set for each sampling time. The finished forging is divided into multiple continuous spatial grid units; based on the die loading geometry of multi-directional forging, the effective area of ​​each loading direction on the finished forging is determined; the effective area of ​​each spatial grid unit and the effective area of ​​each loading direction are overlapped to obtain the effective proportion of the loading direction corresponding to each spatial grid unit; Using the proportion of the loading direction effect of each spatial grid cell as the weight, the parameters in the multi-directional process parameter combination are weighted to generate the equivalent process parameter time series of the corresponding spatial grid cell; The root mean square error between the equivalent process parameter time series and the preset standard process parameter time series is calculated as the deviation. Calculate the standard deviation of the pressure difference value of the loading direction interaction parameter concentration during the forming period as the interaction dispersion; weight the deviation, interaction dispersion and corresponding defect size of the spatial grid cell to obtain the quality score of the spatial grid cell; take the average of the quality scores of all spatial grid cells as the multi-directional forging quality evaluation result.

2. The multi-directional forging quality big data analysis method according to claim 1, characterized in that, The step of aligning the timing data of each loading direction according to each sampling point on the time axis specifically includes: Using the time series data with the highest sampling rate as a benchmark, other time series data are resampled using cubic spline interpolation to ensure that all directions have data at the same timestamp. The interpolation nodes are determined based on the time points of the benchmark sequence.

3. The multi-directional forging quality big data analysis method according to claim 1, characterized in that, The step of calculating the process parameter differences between each loading direction specifically includes: At each sampling moment, the pressure value, temperature value, and displacement value corresponding to any two different loading directions are subtracted to obtain the pressure difference, temperature difference, and displacement difference, respectively. The above differences between all two different directions at each sampling moment constitute the loading direction interaction parameter set.

4. The multi-directional forging quality big data analysis method according to claim 1, characterized in that, The step of dividing the finished forging into multiple continuous spatial grid units specifically includes: A three-dimensional solid model of the forging product is obtained, and the three-dimensional solid model is meshed using a hexahedral meshing method.

5. The multi-directional forging quality big data analysis method according to claim 1, characterized in that, The step of determining the effective area of ​​each loading direction on the finished forging specifically includes: For each loading direction, taking the contact surface between the mold and the forging as the reference plane, a virtual prism is generated along the loading direction that encompasses the entire projection of the forging in that direction. The Boolean intersection operation is performed between the prism and the three-dimensional model of the forging, and the resulting intersection volume is the area of ​​action in that direction.

6. The multi-directional forging quality big data analysis method according to claim 5, characterized in that, The step of performing a Boolean intersection operation between the prism and the three-dimensional model of the forging specifically includes: The virtual prism and the three-dimensional model of the forging are spatially superimposed, retaining the space occupied by both, and removing the parts of the virtual prism that do not overlap with the forging and the parts of the forging that do not overlap with the virtual prism. The remaining part is the area of ​​influence in this direction.

7. The multi-directional forging quality big data analysis method according to claim 1, characterized in that, Following the evaluation result of multi-directional forging quality, the following are also included: Spatial mesh cells with quality scores below a preset threshold are marked as quality anomaly areas. The location and extent of all quality anomaly areas are displayed on the three-dimensional solid model of the forging. At the same time, the time series of equivalent process parameters and the set of interaction parameters of loading direction corresponding to each quality anomaly area are recorded.

8. A multi-directional forging quality big data analysis system, based on the multi-directional forging quality big data analysis method described in claim 1, characterized in that, include: The data acquisition module is used to collect time-series data of each loading direction during the multi-directional forging process, as well as defect size data of each inspection position of the finished forging. The time-series data includes pressure time-series data, temperature time-series data, and displacement time-series data; The timing alignment module is used to align the timing data of each loading direction according to each sampling point on the time axis, and generate a multi-directional process parameter combination for each sampling moment; The interaction parameter module is used to calculate the process parameter difference between each loading direction based on the multi-directional process parameter combination, and generate the loading direction interaction parameter set for each sampling time. The proportion calculation module divides the finished forging into multiple continuous spatial grid units; determines the effective area of ​​each loading direction on the finished forging based on the die loading geometry of multi-directional forging; and calculates the overlap between each spatial grid unit and the effective area of ​​each loading direction to obtain the proportion of the loading direction effect corresponding to each spatial grid unit. The equivalent parameter module is used to weight each parameter in the multi-directional process parameter combination with the proportion of the loading direction effect of each spatial grid cell as the weight, and generate the equivalent process parameter time series of the corresponding spatial grid cell. The quality evaluation module is used to calculate the root mean square error between the equivalent process parameter time series and the preset standard process parameter time series, as the deviation. Calculate the standard deviation of the pressure difference value of the loading direction interaction parameter concentration during the forming period as the interaction dispersion; weight the deviation, interaction dispersion and corresponding defect size of the spatial grid cell to obtain the quality score of the spatial grid cell; take the average of the quality scores of all spatial grid cells as the multi-directional forging quality evaluation result.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the multi-directional forging quality big data analysis method as described in any one of claims 1 to 7.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, which, when executed by a processor, implements a multi-directional forging quality big data analysis method as described in any one of claims 1 to 7.

Images