A method and system for defining the stages of life decay in magnesium alloy automotive die-casting molds.
By constructing a three-dimensional health status index and a process scenario database, and combining a three-level judgment logic and process parameter disturbances, the problem of real-time monitoring and failure mode identification of magnesium alloy automotive die-casting mold life decay was solved, realizing accurate prediction and safe management of mold life.
Patent Information
- Application Number
- CN202611188420.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-06
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-08-06
AI Technical Summary
Existing technologies cannot detect the health degradation of magnesium alloy automotive die-casting molds in real time during service, making it difficult to distinguish different failure modes. This leads to premature mold disuse or failure to provide timely warnings, resulting in a high false alarm rate, low reliability of warnings, and an inability to accurately predict remaining lifespan.
By acquiring the peak attenuation rate of clamping force, the non-uniformity of the temperature field on the cavity surface, and the proportion of vibration and shock energy online, a three-dimensional health status index is constructed. Combined with the process scenario database, a trigger upper limit is dynamically generated. A three-level progressive judgment logic and a process parameter disturbance cycle are adopted to achieve adaptive definition of the mold attenuation stage and reliable confirmation of failure mode.
It achieves online adaptive life decay definition for magnesium alloy automotive die-casting molds, reducing the false alarm rate by 40%, improving the reliability of early warning, and ensuring prediction error of less than 5%, thus ensuring production continuity and mold service safety.
Smart Images

Figure CN122692801B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method and system for defining the stages of life decay of magnesium alloy automotive die-casting molds, belonging to the field of die-casting mold life monitoring and predictive maintenance technology. Background Technology
[0002] Magnesium alloy automotive die-casting molds operate under harsh conditions of high temperature, high pressure, and continuous impact from high-speed molten metal. Inevitably, their cavity surfaces gradually develop various failure modes, including erosion wear, thermal fatigue cracking, and plastic deformation. Accurately defining mold life directly affects the forming quality of automotive structural components, production line downtime and restart costs, and overall production cycle time. However, current mold life management largely relies on fixed mold cycle limits or offline detection after mold opening. This approach not only fails to detect the health degradation process of the mold in consecutive cycles in real time but also struggles to distinguish between different failure modes. This often leads to molds being prematurely decommissioned due to a single anomaly, wasting remaining value, or to batch quality accidents and catastrophic mold damage caused by the failure to promptly warn of rapid crack propagation.
[0003] Currently, some technologies attempt to assess mold condition by monitoring single physical quantities such as mold temperature, vibration, or clamping force online. However, these methods generally suffer from three shortcomings: First, they rely solely on fixed alarm thresholds, which cannot adapt to the normal drift of characteristic quantities caused by fluctuations in process parameters such as injection speed, casting pressure, and pouring temperature, resulting in high false alarm and false negative rates. Second, they lack the fusion and collaborative judgment of multi-physical field characteristics, making it difficult to separate specific attenuation modes from composite failure signals. Third, after an early warning, they lack a closed-loop verification and dynamic correction mechanism for the failure type, leading to low reliability of remaining life prediction and failing to provide a reliable basis for maintenance decisions. Summary of the Invention
[0004] To overcome the problems of existing technologies, such as the inability to effectively distinguish between the decay stage and failure mode during mold service, fixed warning thresholds and lack of process adaptability, and low reliability of remaining life prediction, this invention provides a method and system for defining the stage of life decay of magnesium alloy automotive die casting molds. It realizes online adaptive definition of mold decay stage, reliable confirmation of failure mode, and accurate prediction of remaining usable mold cycles.
[0005] The technical solution of this application specifically includes: According to one aspect of this application, a method for phased determination of the life decay of magnesium alloy automotive die-casting molds is provided, comprising: The peak decay rate of the clamping force, the non-uniformity of the temperature field on the cavity surface, and the proportion of impact energy of vibration extracted by multivariate variational mode decomposition are obtained online for each module and combined into a three-dimensional health status index. Using qualified samples from the initial stage of normal service, a process scenario database is constructed with injection speed, casting pressure and pouring temperature as key values, storing the historical distribution of the three-dimensional health status index; the current process scenario is matched online, the historical minimum value is extracted and a dynamic trigger limit is generated according to a preset multiple; when any two indicators exceed the limit at the same time, an initial decay tendency is marked. After marking, a three-level progressive judgment is initiated: First, if the attenuation rate exceeds a certain multiple of the dynamic trigger limit, a cavity deformation warning is output; Second, if the impact energy ratio changes from a sparse single peak to a dense multi-peak mode, a crack propagation warning is output; Third, if the temperature field non-uniformity increases rapidly and is positively coupled with the impact energy ratio increase, an erosion wear warning is output; If none of these conditions are met, the marking is removed and normal monitoring is resumed. Verify the sequence of warning types; if a single warning reaches the threshold for consecutive times, it is confirmed to be valid; when multiple warnings occur alternately, start the process parameter disturbance cycle, fix the low-speed injection, confirm the warning type based on the attenuation rate swing amplitude and flash increment, and correct the dynamic trigger upper limit by swing amplitude. Based on the confirmed warning type, predict the remaining available modules, and based on the predicted remaining available modules and safety process requirements, output control commands in real time.
[0006] Preferably, the steps for obtaining the three health status indicators include: In each die-casting cycle, the peak clamping force of the current die is extracted during the pressurization and holding phases. The peak clamping force of the reference module calibrated at the beginning of service Substitute them into the following formula to calculate the attenuation rate. : ; Capture key areas of the cavity surface at the moment of mold opening. The temperature data from each temperature measurement point were used to calculate the average temperature of each point within the sampling window. and the arithmetic mean of all temperature measurement points Substitute into the following formula to calculate the temperature field nonuniformity. : ; Vibration signals from the injection and filling stages were extracted and decomposed into multiple variational mode shapes. Calculate the energy of each mode using one eigenmode function. and center frequency In the preset impact-sensitive frequency band The percentage of impact energy is extracted using the following formula. : ;in, For indicator functions, To impact the lower limit frequency of the sensitive frequency band, In order to push the upper frequency limit of the sensitive frequency band, For indexing; attenuation rate Temperature field nonuniformity and impact energy ratio The data are combined into a three-dimensional vector, with the mold sequence number, injection speed, casting pressure, and pouring temperature appended, and then stored in a time-series database.
[0007] Preferably, the steps for constructing the process scenario database and matching the current process scenario include: Taking qualified dies with no significant degradation in the early stages of mold service as the object, the three process parameters of injection speed, casting pressure, and pouring temperature are used as the combined key values. Each process parameter is discretized according to a preset step size. The health status ternary index of dies with the same or similar key values is grouped into the same process scenario, and the decay rate sequence under each scenario is stored. Temperature field non-uniformity sequence and impact energy percentage sequence Basic statistics; The injection speed, casting pressure, and pouring temperature of the current mold are obtained online. After normalization using the standard deviation of the corresponding process parameters in the database, the Euclidean distance between the current process parameters and the combined key values of each scenario is calculated, and the process scenario with the smallest distance is selected as the matching scenario.
[0008] Preferably, the dynamic trigger upper limit generation step includes: Extract the historical minimum value of attenuation rate from the matched process scenarios. The historical minimum value of temperature field nonuniformity and the historical minimum value of impact energy percentage Multiply by the attenuation rate, temperature field non-uniformity, and the preset ratio of impact energy, respectively. , , The dynamic trigger limit of the decay rate is obtained. Temperature field nonuniformity dynamic triggering upper limit Dynamic trigger limit for impact energy percentage ; when or or When any one of the following conditions is met, the current mode is marked as having an initial attenuation tendency, where, For attenuation rate, For temperature field non-uniformity, The percentage of impact energy.
[0009] Preferably, the first step in the three-level progressive determination is: determining the attenuation rate of the current module. Does it meet the requirements? ,in To preset a specific multiple, This is the dynamic trigger limit for the attenuation rate under the current matching process scenario; if it is satisfied, it is determined that the permanent deformation in the mold closing direction has caused dimensional deviation, and the cavity plastic deformation is the current dominant failure mechanism. A cavity deformation warning is output and the judgment of this layer ends.
[0010] Preferably, the second level of the three-level hierarchical determination is: for the most recent consecutive... For each mode, the energy distribution of each mode within the impact frequency band is extracted, and the number and spacing of independent peaks are identified. If only one significant peak is detected in a single mode, it is marked as a sparse single-peak mode. If three or more significant peaks are detected and the frequency interval between the center of adjacent peaks is less than the dense threshold, it is marked as a dense multi-peak mode. When the first half of the modes in the window is dominated by the sparse single-peak mode and the second half is dominated by the dense multi-peak mode, it is determined that the impact energy distribution has completed the mode transition, and a crack propagation warning is output.
[0011] Preferably, the third level of the three-level progressive determination is: For the most recent consecutive The difference sequence for calculating the non-uniformity of the temperature field is obtained by performing multiple operations. The difference sequence of impact energy proportion ; By judgment The slope of the sequence is greater than zero and the second difference remains positive, confirming that the rate of increase in the non-uniformity of the temperature field is accelerating. Calculate the coupling strength using the following formula : ;in, and The difference sequences of temperature field inhomogeneity are respectively The difference sequence of impact energy proportion The mean, For the delay steps, The length of the continuous module window. The index of the module within the window. For the first The increase in the proportion of impact energy of each modal relative to the previous modal; when Difference sequences exceeding a preset coupling threshold and exhibiting non-uniformity in the temperature field. The difference sequence of impact energy proportion When the Pearson correlation coefficient exceeds the preset threshold, it is determined that the increase in temperature field nonuniformity and the increase in the proportion of impact energy form a positive feedback coupling, and an erosion wear warning is output.
[0012] Preferably, the condition for removing the marker and returning to normal monitoring is as follows: when the first, second, and third level judgments in the three-level step-by-step judgment are not satisfied, the preliminary attenuation tendency marker of the current mold is cleared, the preliminary attenuation tendency removal event is recorded, the mold returns to the normal monitoring state, and the process scenario matching and dynamic over-limit judgment are repeated.
[0013] Preferably, the condition for confirming the validity of a single warning consecutive module reaching a threshold is: comparing the current module with the previous module. The combination of warning types for each module is called a warning sequence. When the warning sequence contains only the same warning type from erosion wear warning, crack propagation warning, or cavity deformation warning, and the number of consecutive occurrences is not less than a preset duration threshold, the warning sequence is considered to be in sequence. When the warning type is confirmed to be valid, the remaining prediction modules will proceed.
[0014] Preferably, the step of initiating a process parameter disturbance cycle and confirming the warning type based on the attenuation rate swing amplitude and flash increment includes: When at least two different warning types appear in the warning sequence and the number of consecutive occurrences of each type is less than the alternation judgment value, based on the currently matched process scenario, the injection speed is fixed at a low speed and the holding time remains unchanged, and continuous production continues. Each module; Collect the peak value of the mold closing force during the disturbance and calculate the attenuation rate; measure the maximum length of the flash at the parting surface of each mold pass. Define the oscillation amplitude of the decay rate The increment of the flash length is defined as the percentage decrease in the mean attenuation rate during the disturbance relative to the mean attenuation rate of several modes before the disturbance. This represents the absolute increase in the average flash length during the disturbance period relative to the average flash length of the historical normal mode. like Swing amplitude threshold exceeding the decay rate and No more than the threshold for significantly increasing flash. If so, the current module will be remarked as a crack propagation warning, and the swing amplitude will be fed back to correct the dynamic trigger limit of the attenuation rate under the current process scenario; like No more than and Exceed If so, the current module will be locked as an irreversible stage and the cavity deformation warning will be maintained; like Exceed and Exceed It is also locked into an irreversible stage and maintains early warning of cavity deformation.
[0015] Preferably, the step of predicting the remaining available modules based on the confirmed warning type includes: When the warning type is erosion wear, calculate the coupling index. Substitute into the inverse power-exponential model Predict the remaining available modulo operations, where, For the remaining usable modules under erosion and wear warning, The failure threshold for cavity surface roughness. It is a proportionality constant. The coupling index is... For temperature field non-uniformity, To impact energy percentage, To reflect the material-process related index of the coupling index on the accelerating effect of roughness growth, A material-process related index that reflects the time-scale characteristics of roughness as it increases with the number of mold passes; These are the currently serving modules; When the warning type is crack propagation, the frequency of occurrence of the dense multi-peak pattern of statistical impact energy proportion is... Modified Logistic Growth Model The spread rate coefficient in , making the crack feature length Crack length reached the repair limit Solve for the critical module to obtain the remaining module. ,in The saturation crack length, For shape parameters, For cumulative modules; When the warning type is cavity deformation, establish the critical cavity size deviation. With cumulative modulus Power law relationship This causes deviations in the critical cavity dimensions. Tolerance zone half width Solve for the critical module to obtain the remaining module. ,in and The fitting coefficients are denoted as .
[0016] Preferably, the control command includes: The predicted remaining available module counts are compared with the preset safety process requirements, and multiple risk levels are divided. A hierarchical control instruction is generated, which includes the mold number, current module count, warning type, estimated remaining module counts, and suggested handling solutions. This instruction is then pushed to the human-machine interface and the manufacturing execution system. At the same time, the warning results and actual maintenance status are fed back to the process scenario database for rolling optimization of the parameters in the remaining module count prediction model.
[0017] This application also provides a system for defining the stage-based degradation of the lifespan of magnesium alloy automotive die-casting molds, the system comprising: The health status ternary index acquisition module is used to acquire online the peak decay rate of the clamping force, the non-uniformity of the temperature field on the cavity surface, and the proportion of impact energy of vibration extracted by multivariate variational mode decomposition for each mode, and combine them into a health status ternary index. The initial attenuation tendency marking module is used to construct a process scenario database with injection speed, casting pressure and pouring temperature as key values using qualified modules in the early stage of normal service, and store the historical distribution of the three-dimensional health status indicators; it matches the current process scenario online, extracts the historical minimum value and generates a dynamic trigger upper limit according to a preset multiple; when any two indicators exceed the limit at the same time, it marks the initial attenuation tendency. The decay mode step-by-step judgment module is used to initiate a three-level step-by-step judgment after marking: First, if the decay rate exceeds a certain multiple of the dynamic trigger limit, a cavity deformation warning is output; Second, if the impact energy ratio changes from a sparse single-peak mode to a dense multi-peak mode, a crack propagation warning is output; Third, if the temperature field non-uniformity increases rapidly and is positively fed into the impact energy ratio increase, an erosion wear warning is output; If none of these conditions are met, the marking is removed and normal monitoring is resumed. The warning verification and disturbance correction module is used to verify the warning types in sequence; if a single warning is confirmed to be valid if it reaches a threshold in a series of consecutive times; when multiple warnings occur alternately, the process parameter disturbance cycle is started, the low-speed injection is fixed, the warning type is confirmed based on the attenuation rate swing amplitude and flash increment, and the dynamic trigger upper limit is corrected by the swing amplitude. The remaining module prediction and control module is used to predict the remaining available modules based on the confirmed warning type, and output control commands in real time based on the predicted remaining available modules and safety process requirements.
[0018] Compared with existing technologies, the method for determining the stage-based degradation of the lifespan of magnesium alloy automotive die-casting molds provided by this invention has the following advantages: This invention addresses the inherent flaw of fixed thresholds, which struggle to adapt to fluctuations in process parameters such as injection speed and casting pressure, leading to high false alarm rates. It constructs a three-dimensional health index system comprising clamping force attenuation rate, temperature field non-uniformity, and impact energy ratio, and dynamically generates an adaptive trigger limit based on historical minimum values for the process scenario. A three-level progressive judgment logic accurately separates three failure modes: cavity deformation, crack propagation, and erosion wear, filling the gap in existing technologies that cannot distinguish between composite failure modes online. Furthermore, a process parameter disturbance cyclic verification and threshold feedback correction mechanism further enhances the reliability of the early warning system. The false alarm rate was reduced by approximately 40%. Based on this, the remaining usable mold cycles under different modes were predicted using the inverse power-law-exponential model, the logistic growth model, and the power-law model, respectively. The prediction error for crack propagation was less than 5%, the prediction error for cavity deformation was approximately 2.1%, and the measured surface roughness during erosion wear repair was highly consistent with the predicted failure threshold. No unplanned downtime occurred during the entire life cycle of the mold. This achieved online adaptive definition of the life decay stage of magnesium alloy automotive die-casting molds, closed-loop reliable confirmation of failure modes, and accurate prediction of remaining life, ensuring production continuity and mold service safety. Attached Figure Description
[0019] Figure 1 A schematic diagram of the overall method for defining the stage-based degradation of the lifespan of magnesium alloy automotive die-casting molds; Figure 2 S100 Flowchart for the Stage-Based Determination Method of Life Decay of Magnesium Alloy Automotive Die-Casting Molds; Figure 3 S200 flowchart for the staged definition method of life decay of magnesium alloy automotive die-casting molds; Figure 4 S300 flowchart for the staged definition method of life decay of magnesium alloy automotive die-casting molds; Figure 5 S400 flowchart for the staged definition method of service life decay of magnesium alloy automotive die-casting molds; Figure 6 S500 flowchart for the staged definition method of life decay of magnesium alloy automotive die-casting molds. Detailed Implementation
[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0021] The theoretical basis of this invention is built upon the fusion of multi-physics degradation characteristics, the construction of adaptive thresholds for dynamic process scenarios, and the theory of stepwise separation and closed-loop verification of failure modes. Magnesium alloy automotive die-casting molds, under repeated scouring by high-temperature, high-pressure molten metal, simultaneously undergo three physical degradation processes on the cavity surface: material loss, surface cracking, and permanent deformation. These three degradation processes exhibit multidimensional responses with mutually coupled but distinct evolutionary characteristics in the clamping force, temperature field, and vibration signals. By online acquisition of the peak attenuation rate of the clamping force, the non-uniformity of the temperature field, and the proportion of vibration impact energy, a three-dimensional health status index is constructed, which can characterize the global and local features of mold degradation from different physical dimensions. Utilizing a large amount of qualified mold data accumulated during the early stages of normal mold service, a process scenario database is constructed using injection speed, casting pressure, and pouring temperature as combined key values. The historical minimum value under similar process scenarios is used as a dynamic reference benchmark, effectively eliminating the normal drift of characteristic quantities caused by fluctuations in process parameters. Based on this, a progressive attenuation mode determination logic is introduced to sequentially exclude three typical failure modes: cavity deformation, crack propagation, and erosion wear. Closed-loop verification and threshold self-correction are performed in conjunction with process parameter disturbance cycles, which significantly improves the reliability of failure mode identification and the accuracy of remaining mode prediction.
[0022] The derivation of the core formulas includes: Temperature field nonuniformity Defined as the critical area on the surface of the mold cavity Temperature at each measuring point With average temperature The root mean square value of the deviation between them is given by equation (1): In equation (1), For the first The average temperature of each temperature measurement point within the current temperature sampling window. For all The arithmetic mean of the average temperatures at each temperature measurement point. It reflects the dispersion of temperature distribution on the cavity surface; the larger the value, the more uneven the temperature distribution.
[0023] Impact energy percentage The vibration signal from the mold is decomposed into multiple variational modes and then analyzed from the preset impact sensitive frequency band. The ratio of internal mode energy to total energy is calculated as shown in equation (2): In equation (2), This represents the total number of eigenmode functions obtained from multivariable variational mode decomposition. For the first Energy of a mode For the first The center frequency of each mode For indicator functions, when The time value is Otherwise Impact energy percentage It reflects the strength of the impact components caused by changes in the surface condition of the mold cavity during the molten metal filling process.
[0024] Peak clamping force attenuation rate Defined as the peak value of the current mold clamping force. Peak clamping force relative to the reference mold The relative percentage decrease is shown in equation (3): In equation (3), This represents the average of the peak clamping force values recorded during several successful mold runs under standard process conditions during the initial service period of the mold. This represents the peak clamping force of the current module. Attenuation rate. A gradual increase indicates a weakening of the mold cavity's elastic recovery capability in the mold closing direction, which is usually associated with dimensional deviations caused by permanent plastic deformation or severe wear.
[0025] The coupling strength between the rate of increase in temperature field nonuniformity and the rate of increase in the proportion of impact energy Defined as continuous Within each module The intermodal difference sequence and The normalized cross-correlation peak of the inter-modal difference sequence is given by equation (4): In equation (4), For the first The increment of temperature field nonuniformity in each modulus relative to the previous modulus This corresponds to the increase in the proportion of impact energy. and These are the means of the two difference sequences, respectively. This is the delay step number. When... When the preset coupling threshold is exceeded and both sets of incremental sequences show a monotonically increasing trend, it is determined that the growth rate of temperature field nonuniformity and the growth rate of impact energy ratio form a positive feedback coupling.
[0026] The prediction of remaining available modulus based on erosion wear adopts an inverse power-law-exponential model, introducing a coupling exponent. Remaining modules As in equation (5): In equation (5), The failure threshold for cavity surface roughness. It is a proportionality constant. and For materials-process related indices, For the currently in-service modules. Formula (5) uses the coupling index. The synergistic effect of temperature field nonuniformity and impact energy ratio is mapped to the remaining modes, enabling a quantitative assessment of the remaining life during the erosion wear stage.
[0027] Crack propagation residual modulus prediction is based on a logistic growth model, with crack feature length... With cumulative modulus The change is described as shown in equation (6): In equation (6), The saturation crack length, For shape parameters, This is the expansion rate coefficient. When... Crack length reached the repair limit When the critical module is obtained, the solution is then performed. Then the remaining available modules .
[0028] Remaining mold order prediction for cavity deformation based on critical cavity size deviation With cumulative modulus The power-law relationship is shown in equation (7): In equation (7), and These are the fitting coefficients. When the dimensional deviation reaches half the tolerance band... At that time, the critical module Remaining available modules .
[0029] The above formula provides the mathematical basis for quantitative judgment and prediction in this invention.
[0030] The specific embodiments of the present invention will be described in detail below.
[0031] Example 1: Please see Figure 1 , Figure 1 This invention illustrates a method for phased determination of the life decay of magnesium alloy automotive die-casting molds, as provided in an embodiment of the present invention. The method includes: S100: Continuously acquires the peak decay rate of clamping force, temperature field non-uniformity, and impact energy ratio, which are combined into a three-dimensional index of health status.
[0032] S200: Match the process scenario database, extract the historical minimum value and calculate the dynamic trigger limit, and perform preliminary attenuation tendency marking.
[0033] S300: Performs a three-level attenuation mode determination on the initial attenuation tendency mode, and outputs the warning type or removes the flag.
[0034] S400: Verify and correct the final warning type through cyclical verification of the warning sequence and cyclical confirmation of process parameter disturbances.
[0035] S500: Predicts the remaining available modules based on the final warning type and outputs control commands in real time.
[0036] The specific plan includes: In a method for defining the phased degradation of the life of a magnesium alloy automotive die-casting mold, S100 synchronously collects clamping force, multi-point temperature of the cavity, and mold vibration signals online, and extracts a ternary index that can comprehensively reflect the plastic deformation, thermal balance deterioration, and surface impact state of the mold cavity, providing a unified feature input for subsequent phased definition.
[0037] Please refer to Figure 2 , Figure 2 A flowchart of an exemplary method S100 for defining the staged degradation of the lifespan of a magnesium alloy automotive die-casting mold, as described in this application, is shown. The flowchart includes the following: S110: Deploy clamping force, temperature and vibration sensors and perform baseline module calibration.
[0038] In this step, clamping force sensors, multi-point temperature sensors, and broadband vibration sensors are installed on the clamping mechanism of the magnesium alloy automotive die-casting machine, the key areas of the mold cavity surface, and the outer side of the mold fixed mold. During the initial service period of the mold, qualified batches produced continuously under standard process conditions are selected, and the average value of the peak clamping force is recorded as the benchmark peak clamping force. .
[0039] In one possible implementation of this step, a strain gauge type tie rod force sensor is selected as the clamping force sensor and installed on the tie rod of the die-casting machine tail plate. The total clamping force is obtained by summing, and the sampling frequency is set according to the response speed of the die-casting machine. Temperature measurement of key areas on the mold cavity surface is achieved using thermocouples pre-embedded at a certain depth from the cavity surface, arranged along the gate, the end of the runner, and easily eroded areas. There are [number] temperature measurement points, each numbered as follows: The temperature data acquisition frequency is determined based on the thermal response speed. A wideband vibration sensor, an IEPE accelerometer, with a resonant frequency higher than the expected upper limit of the impact frequency, is installed on the flat, machined surface on the back of the mold's fixed mold. The benchmark mold calibration is performed after the mold has been fully preheated and a certain number of mold cycles have been produced continuously and stably. Several consecutive defect-free mold cycles are selected, and the peak clamping force of each cycle is recorded. The average value is calculated as the peak clamping force of the benchmark mold cycle. And store them for later use.
[0040] S120: Real-time acquisition of peak clamping force for each cycle and calculation of attenuation rate.
[0041] In this step, within each die-casting cycle, the clamping force sensor continuously collects the clamping force signal. The peak clamping force during the pressurization and holding phases of each die is extracted using a peak detection algorithm as the peak clamping force for the current die. And calculate in real time relative to formula (3) attenuation rate .
[0042] In one possible implementation of this step, the peak clamping force extraction window is set to a period of time from before the end of injection to the end of the holding pressure period to ensure complete capture of the clamping force response. The peak detection algorithm uses a sliding window maximum search, with the window width set according to the signal characteristics. The obtained peak clamping force for the current molding cycle is then... Peak clamping force of the reference mold Substitute into the formula The attenuation rate of the current mode is calculated. And stored as a percentage. Decay rate The changing trend can sensitively reflect the degree of weakening of the mold cavity's elastic recovery ability in the mold closing direction.
[0043] S130: Obtain the temperature at multiple points in key areas of the mold cavity surface and calculate the temperature field non-uniformity.
[0044] In this step, temperature data from each temperature measurement point before and after the moment the mold opens for each mold cycle are collected simultaneously, and the average value of each temperature measurement point within the temperature sampling window for the current mold cycle is calculated. and all Average temperature of point Then, the temperature field nonuniformity is calculated according to formula (1). .
[0045] In one possible implementation of this step, the temperature sampling window is selected as a period of time after the mold is fully opened, during which the temperature field on the cavity surface is relatively stable and can reflect the thermal state immediately after filling. There are 1 temperature measurement point, and the calculation is performed at each measurement point. Average temperature within the sampling window Then calculate all Arithmetic mean of points .Will and Substitute into the formula The temperature field nonuniformity of the current mode is obtained. Its physical meaning is the degree of dispersion of temperature distribution on the cavity surface. Local erosion, cracking, or sticking can alter heat conduction conditions, thereby causing... Increase, therefore It can serve as an effective indicator of surface condition deterioration.
[0046] S140: Collect mold vibration signals and extract the impact energy ratio using multivariable variational mode decomposition.
[0047] In this step, effective analysis segments are extracted from the mold vibration signals during each injection and filling stage, and the vibration signals are adaptively decomposed into multiple variational mode decomposition. Calculate the center frequency of each mode using one intrinsic mode function. and energy And based on the preset impact sensitive frequency band The impact energy percentage is calculated using formula (2). .
[0048] In one possible implementation of this step, the vibration signal analysis segment is extracted from the period before the injection start trigger signal to the period after the injection ends, to fully cover the high-speed filling and impact process of the molten metal. The signal acquired by the accelerometer is subjected to multivariate variational mode decomposition, and the number of modes is set. And other control parameters. After decomposition, we obtain... Each intrinsic mode function, for each mode Calculate energy The center frequencies of each mode are obtained using the Hilbert transform. Impact-sensitive frequency band The frequency range was selected based on the transient impact characteristics caused by minute defects and crack opening and closing on the cavity surface during magnesium alloy injection. To impact the lower limit frequency of the sensitive frequency band, This is the upper frequency limit of the impact-sensitive frequency band. Indication function. exist The time value is Otherwise, the value is .Will , Substituting the frequency band boundary into the formula The percentage of impact energy obtained When defects such as cracks or erosion pits appear on the cavity surface, more high-frequency transient impacts will be generated during the molten metal filling process, causing... Significant increase.
[0049] S150: Combines ternary indicators and stores them in a normalized manner.
[0050] In this step, the calculated attenuation rate will be... Temperature field nonuniformity and impact energy ratio The data are combined into a three-dimensional vector, which serves as the three-dimensional health status index for the current module, along with the module number and process parameters, and stored in the time series database.
[0051] In one possible implementation of this step, the ternary index is represented as , For the transpose symbol, add back, It is transformed into a 3×1 column vector. Simultaneously, the modulus number of the current modulus is recorded. Injection speed Casting pressure and pouring temperature The storage format uses a dual index of timestamps and module numbers to facilitate subsequent retrieval by process scenario. All data undergoes outlier removal and missing value imputation before being stored in the database to ensure data quality.
[0052] In a phased method for defining the life decay of magnesium alloy automotive die-casting molds, S200 utilizes qualified mold data accumulated during the early stage of normal mold service to construct a process scenario database with injection speed, casting pressure, and pouring temperature as combined key values. It also dynamically generates an adaptive trigger limit through online matching to achieve sensitive and robust marking of the initial decay tendency.
[0053] Please refer to Figure 3 , Figure 3 A flowchart of S200, an exemplary method for defining the staged degradation of the lifespan of a magnesium alloy automotive die-casting mold according to this application, is shown. The flowchart includes the following: S210: Offline construction of process scenario database and storage of historical distribution of ternary indicators.
[0054] In this step, data from continuous production runs during the initial service phase of the mold that have passed quality verification are collected. Each run includes injection speed data. Casting pressure Pouring temperature The corresponding health status ternary index is used to classify process scenarios based on the combined key value of three process parameters. Modular ternary indices with the same or similar key values are grouped into the same scenario to establish a process scenario database.
[0055] In one possible implementation of this step, the initial service life of the mold refers to a continuous series of qualified mold runs after a new mold is put into production. During this stage, the mold cavity shows no significant degradation, and the ternary parameters only reflect normal process fluctuations. The combined key values are discretized for injection speed, casting pressure, and pouring temperature, with the step size set according to the control precision of the process parameters. The ternary parameters for each mold run are... The data is bound to discretized key-value pairs and stored in a relational database. For each process scenario, the decay rate sequence is calculated and stored separately. Temperature field non-uniformity sequence and impact energy percentage sequence The basic statistics, including historical minimum, mean, and standard deviation, provide a data foundation for subsequent dynamic triggering of upper limit calculations.
[0056] S220: Obtain the current process parameters online and match the closest process scenario.
[0057] In this step, after each die-casting cycle, the actual injection speed of the current die is read in real time from the die-casting machine control system. Casting pressure and pouring temperature The Euclidean distance between the combined key values in the process scenario database and the process scenario with the smallest distance is calculated.
[0058] In one possible implementation of this step, since the dimensions and ranges of the parameters are different, the standard deviation of the corresponding process parameters in the database is used for normalization before matching. The formula for calculating the Euclidean distance after normalization is as follows: Choose to make The smallest process scenario is used as the matching scenario for the current module, and all historical ternary index data corresponding to the matching scenario are loaded into memory.
[0059] S230: Extract the historical minimum value of each indicator in the matching scenario and calculate the dynamic trigger limit.
[0060] In this step, the historical minimum value of the attenuation rate is extracted from the matched process scenario database records. The historical minimum value of temperature field nonuniformity and the historical minimum value of impact energy percentage Multiply by the preset multiplier respectively , and This yields the dynamic trigger limits for the three indicators.
[0061] In one possible implementation of this step, the preset multiplier is determined comprehensively based on the mold material, product precision requirements, and normal process fluctuations. The dynamic triggering upper limit of the attenuation rate is... The dynamic triggering upper limit for temperature field nonuniformity is The dynamic triggering upper limit of the impact energy ratio is The aforementioned dynamic trigger limit is automatically adjusted according to the process scenario. The historical minimum value varies under different scenarios. Therefore, even if the process parameters deviate normally, the alarm threshold can adapt accordingly, effectively suppressing false alarms.
[0062] S240: Perform simultaneous over-limit judgment of two indicators and mark the initial attenuation trend.
[0063] In this step, the ternary index of the current module is... With the corresponding dynamic trigger limit Compare them one by one. If any two indicators exceed their respective dynamic trigger limits at the same time, the current module is marked as having an initial attenuation tendency; otherwise, maintain normal monitoring status and return to S100 to continue collecting data for the next module.
[0064] In one possible implementation of this step, the judgment logic is: if or or If the condition is met, the current module's marker position will be set to the initial attenuation state, and the S300 will be triggered to enter the attenuation mode determination process; if none of the above logics are met, the module will continue to be monitored normally.
[0065] In a phased method for determining the life decay of magnesium alloy automotive die-casting molds, S300 takes the initial decay tendency marker generated by S200 as the entry point, and through three progressively advancing judgment levels, sequentially examines the characteristics of three failure modes: cavity deformation, crack propagation, and erosion wear, outputs the corresponding warning type, or removes the marker after eliminating all modes.
[0066] Please refer to Figure 4 , Figure 4 A flowchart of S300, an exemplary method for defining the staged degradation of the lifespan of a magnesium alloy automotive die-casting mold according to this application, is shown. The flowchart includes the following: S310: First-level judgment - Attenuation rate over-limit check and cavity deformation early warning output.
[0067] In this step, for the current module marked as having an initial attenuation tendency, the attenuation rate is first examined separately. Does it exceed the dynamic trigger limit of the decay rate under the current process scenario? Specific multiples If the condition is met, it is directly determined to be a cavity deformation mode, a cavity deformation warning is output, and the determination process of this mode ends.
[0068] In one possible implementation of this step, a specific multiple. The value is set based on the elastic modulus of the mold material and the dimensional tolerances of the cavity; it is greater than 1 to detect abnormal acceleration in the decay rate. The judgment condition is... When the above conditions are met, it indicates that permanent deformation in the mold closing direction has led to significant dimensional deviations, and plastic deformation of the cavity is the current dominant failure mechanism. A cavity deformation warning is then output, and S300 ends, directly entering the S400 loop verification mechanism. If the conditions are not met, the process proceeds to S320 to execute the second-level judgment.
[0069] S320: Second-level judgment - detection of the mode transition from sparse single peak to dense multi-peak impact energy ratio and early warning output of crack propagation.
[0070] In this step, if the cavity deformation warning is not triggered at the first level, the impact energy percentage is checked. In recent consecutive Within each mold cycle, the process checks whether the mold cavity surface cracks have transitioned from a sparse unimodal mode to a dense multimodal mode. If a mode transition is detected, the process determines that the cracks have entered a rapid propagation phase, outputs a crack propagation warning, and ends the current determination.
[0071] In one possible implementation of this step, the successive module window length The timescale for mold thermal fatigue crack propagation is set according to the typical timescale. For each mode within the window, the energy proportion distribution of each mode in the impact band is extracted from the stored multivariate variational mode decomposition results of the vibration signal. A peak search algorithm is used to identify independent peaks in the distribution, i.e., the energy proportion sequence is smoothed, a relative height threshold and a density threshold are set, and all maxima exceeding the height threshold on the smoothed curve are detected as peaks. If only one peak is detected in a mode, it is marked as a sparse single-peak mode; if three or more peaks are detected, and the frequency interval between the center frequencies of adjacent peaks is less than the density threshold, it is marked as a dense multi-peak mode. The mode transition determination condition is: within the window... In the first half, the vast majority of modes are in a sparse unimodal mode; within the window... In the latter half of the signal, the vast majority of modes are in a dense multi-peak mode. When the above conditions are met, it is determined that the impact energy ratio has completed the transition from a sparse single-peak mode to a dense multi-peak mode, outputting a crack propagation warning, ending S300, and proceeding to S400. The appearance of the dense multi-peak mode is closely related to the initiation and propagation of multiple cracks; the increase in the number and size of cracks leads to the appearance of multiple separate energy peaks within the impact response frequency band. If no mode transition is detected, proceeding to S330 to execute the third-level judgment.
[0072] S330: Third-level judgment - positive feedback coupling detection of temperature field nonuniformity and impact energy ratio and output of erosion and wear early warning.
[0073] In this step, if the second level does not output a crack expansion warning, then for the most recent consecutive... Temperature field nonuniformity in each mode Sequence and shock energy ratio Perform coupling analysis on the sequence to determine Does the growth rate show an accelerating upward trend, and is it consistent with... The growth rate forms a positive feedback coupling, and if the coupling condition is met, an erosion and wear warning is output.
[0074] In one possible implementation of this step, a successive module window The selection is based on the evolution rate of erosion wear. First, the time interval between adjacent modules within the window is calculated. Difference sequences and Difference sequences The result was obtained through linear fitting. The slope of the sequence; if the slope is greater than zero and the second difference remains positive, then... The growth rate is accelerating. The coupling strength is then calculated according to formula (4). Delay steps The search is performed within a preset small range, and the maximum value is taken as the final coupling strength. If... Exceeding the preset coupling threshold and and If the Pearson correlation coefficient also exceeds the preset threshold, it is determined that the increase in temperature field non-uniformity and the increase in the proportion of impact energy form a positive feedback coupling. During erosion wear, the loss of material from the cavity surface exacerbates local overheating and flow disturbance, increasing temperature non-uniformity. At the same time, the increase in surface roughness leads to an increase in the proportion of impact energy. The two promote each other, and positive feedback coupling is a significant characteristic that distinguishes erosion wear from other failure modes. If the coupling condition is met, the system outputs an erosion wear warning, ends S300, and transitions to S400.
[0075] S340: Remove the initial attenuation tendency marker when the third level is not satisfied.
[0076] If the coupling condition is not met in the third-level judgment in this step, it indicates that the initial attenuation tendency of the current mode may be caused by random disturbances or mild synergy of multiple factors, which is not enough to be identified as any failure mode. The initial attenuation tendency mark is removed and the mode returns to normal monitoring status.
[0077] In one possible implementation of this step, the initial attenuation marker of the current module is cleared, and the intermediate variables generated during the S300 determination process are cleared. The initial attenuation tendency cancellation event is recorded in the monitoring system log, including the module number and the specific value of the ternary index. Then the process returns to S100 to continue the data collection of the next module and the online matching of S200.
[0078] In a phased definition method for the life decay of magnesium alloy automotive die-casting molds, S400 performs dual verification of the warning type output by S300 based on sequence continuity and process parameter disturbance cycle to ensure the reliability of the warning and the accuracy of failure mode classification, and provides feedback correction for the threshold.
[0079] Please refer to Figure 5 , Figure 5 A flowchart of S400, an exemplary method for defining the staged degradation of the lifespan of a magnesium alloy automotive die-casting mold according to this application, is shown. The flowchart includes the following: S410: Construct an early warning sequence and perform continuity confirmation for a single early warning type.
[0080] In this step, the current module is compared with the previous module. The warning types of each module are arranged in chronological order to form a warning sequence. When only the same warning type appears in the warning sequence and the number of consecutive modules reaches a preset duration threshold, a warning sequence is formed. At that time, directly confirm the validity of the corresponding warning and enter S500.
[0081] In one possible implementation of this step, the preceding window length The alarm sequence length is determined based on the module cycle and alarm response delay requirements. If the warning type identifier for all elements in the sequence is erosion wear, and the number of most recent consecutive erosion wear warnings is not less than [number missing], then [the sequence is incomplete]. If the erosion wear warning is valid, the process directly proceeds to the erosion wear remaining mode prediction branch in S500. Similarly, if the sequence consists entirely of crack propagation and the number of consecutive occurrences reaches the threshold, the crack propagation warning is confirmed to be valid; if the sequence consists entirely of cavity deformation and the number of consecutive occurrences reaches the threshold, the cavity deformation warning is confirmed to be valid. A single consecutive warning indicates that the failure mode is occurring stably and no further verification is required.
[0082] S420: Detects multiple alternating types of disturbances and initiates a process parameter disturbance cycle.
[0083] In this step, when two or three warning types appear alternately in the warning sequence, it indicates that a single logic cannot stably distinguish the failure type. There may be coupling or competition between crack propagation and cavity deformation. Initiate a process parameter disturbance cycle and observe the mold response by actively changing the injection speed level to separate the two failure contributions.
[0084] In one possible implementation of this step, the alternating pattern is defined as follows: in the warning sequence, at least two different warning types appear, and the number of consecutive occurrences of the same warning type is less than a preset alternation judgment value. Once the above alternation pattern is detected, the process parameter disturbance cycle for the current module is immediately activated.
[0085] S430: Perform low-speed injection disturbance and collect response characteristics.
[0086] In this step, based on the current process scenario matched by S200, the injection speed is fixed at a low speed setting, while the holding time and other process parameters remain unchanged for subsequent continuous injection. For each batch of production, the peak clamping force and flash length of each batch are collected, and the swing amplitude of the decay rate is calculated. and the increment of flash length .
[0087] In one possible implementation of this step, the low-speed setting is selected as a low-quantum value within the commonly used injection speed range for the current process scenario, to significantly reduce the impact kinetic energy of the molten metal on the mold cavity. The holding time remains at the original set value. Number of disturbance modes. The method is determined based on statistical stability requirements. In each disturbance mode, the peak closing force is recorded and the attenuation rate is calculated. Simultaneously, a displacement sensor installed at the parting surface is used to measure the maximum length of flash generated in each mode. The swing amplitude of the attenuation rate is also measured. Defined as the percentage decrease in the average attenuation rate during the disturbance period relative to the average attenuation rate of the last few moduli before the disturbance was triggered. (Increase in flash length) This represents the absolute increase in the average length of flash during the disturbance period relative to the average length of flash during the historical normal mode.
[0088] S440: Determines and corrects the warning type based on the combination of sway amplitude and flash increment.
[0089] In this step, based on Swing amplitude threshold with decay rate and Significantly increases threshold with flash The comparison relationship will be used to re-mark the current module as a crack expansion warning or lock it as an irreversible stage, while the swing amplitude will be fed back to S200 for dynamic triggering of upper limit correction.
[0090] In one possible implementation of this step, the determination logic is as follows: If and This indicates that the mold clamping force significantly recovered after reducing the injection speed, while the flash did not worsen. The dominant factor was crack propagation leading to poor adhesion between the moving and fixed molds. Therefore, the current mold run is re-marked as a crack propagation warning. Simultaneously, The value is transmitted to S200 to correct the dynamic trigger limit of the attenuation rate under the current process scenario; the correction magnitude is proportional to the swing amplitude. If and This indicates that flash increases significantly under low-speed injection, and the attenuation rate does not significantly recover as the speed decreases, indicating that irreversible plastic deformation has occurred in the cavity. Therefore, the current mold cycle is locked in the irreversible stage, and a cavity deformation warning is maintained. If... and This indicates that the mold simultaneously exhibits severe cracking and irreversible deformation, thus locking it into an irreversible stage and maintaining a cavity deformation warning. After the disturbance cycle ends, it enters S500 with a re-marked or maintained warning type.
[0091] In a phased definition method for the life decay of magnesium alloy automotive die-casting molds, S500 calls the corresponding degradation model based on the finally confirmed warning type, predicts the remaining usable mold cycles, and generates control instructions in real time based on the prediction results.
[0092] Please refer to Figure 6 , Figure 6 A flowchart of S500, an exemplary method for defining the staged degradation of the lifespan of a magnesium alloy automotive die-casting mold according to this application, is shown. The flowchart includes the following: S510: Prediction of remaining modulus under erosion wear early warning.
[0093] In this step, when the final warning type is erosion wear, the coupling index is calculated. Using an inverse power-law exponential model, combined with pre-calibrated material-process constants and cavity surface roughness failure threshold, Based on formula (5), the coupling index is predicted to reach The remaining available modules at that time.
[0094] In one possible implementation of this step, the roughness failure threshold The model parameters are determined based on the surface roughness requirements of the sealing or load-bearing surfaces for magnesium alloy automotive structural components. , and The current die coupling index is obtained by fitting data from historical erosion and wear tests of the die or accelerated tests of similar dies. From the most recent several modules The average value is obtained. Substitute it into the formula. The remaining usable modes under erosion wear-dominated conditions are calculated. When the number of cycles falls below the preset safety threshold, a control command is issued suggesting that the cavity surface be repaired.
[0095] S520: Remaining module prediction under crack propagation early warning.
[0096] In this step, when the final warning type is crack propagation, the frequency of occurrence of the dense multi-peak pattern of impact energy proportion is statistically analyzed. The frequency of occurrence is used as an acceleration factor to correct the propagation rate coefficient. According to the logistic growth model shown in formula (6), the law of crack feature length changing with the number of modulo cycles is fitted to predict that the crack length will reach the maintenance limit. The remaining available modules at that time.
[0097] In one possible implementation of this step, the frequency of occurrence of dense multimodal patterns Defined as the proportion of modules identified as having a dense multimodal pattern within a recent certain number of modules. The expansion rate coefficient in the logistic growth model Perform linear correction, the correction relationship is as follows: ,in The basic expansion rate coefficient is determined from the initial cracking data of the mold. Acceleration factor weights. Saturation crack length. and shape parameters Determined based on repair records of the same mold model. Repair limit crack length. Based on the enterprise standard. Substitute the above parameters into the formula. ,make The critical module is obtained by solving. And thus the remaining modules .when When the temperature falls below a preset threshold, the system will output control commands suggesting cavity crack repair by welding or insert replacement.
[0098] S530: Prediction of remaining mold cycles under cavity deformation early warning.
[0099] In this step, when the final warning type is cavity deformation, the attenuation rate is directly used. Historical sequence, establish key cavity size deviation With cumulative modulus Based on the power law relationship, formula (7) is used to predict whether the dimensional deviation exceeds the half-width of the tolerance zone. The remaining available modules at that time.
[0100] In one possible implementation of this step, the critical cavity size deviations under several modules are obtained through periodic offline coordinate measuring machine measurements, and a mapping is established with the attenuation rate recorded for the corresponding module, which is then further converted into... and Power-law model Fit coefficients and Determined by least squares fitting. Tolerance band half-width. Provided according to the drawing requirements. The critical module is obtained by solving. Then the remaining modulo .when When the value falls below the preset warning value, an emergency control command is generated indicating that the mold is about to exceed the tolerance and the machine should be stopped immediately for inspection.
[0101] S540: Integrated output and feedback of control commands.
[0102] In this step, the remaining mold cycle prediction value calculated by S510, S520 or S530 is combined with the current production cycle and mold maintenance resources to generate hierarchical control instructions in real time. These instructions are then pushed to relevant personnel through the human-machine interface and manufacturing execution system. At the same time, the prediction results and maintenance records are fed back to the process scenario database for subsequent rolling optimization of model parameters.
[0103] In one possible implementation of this step, the control command is divided into multiple levels based on the remaining number of mold cycles. When there are sufficient remaining mold cycles, only information is recorded; when there are few remaining mold cycles, a maintenance plan is prompted; and when the remaining mold cycles fall below a warning value, a maintenance or mold replacement command is forcibly pushed. The command content includes the mold number, current mold cycle, warning type, estimated remaining mold cycles, and suggested handling plan. All warnings and handling results are recorded and compared with the actual state of the mold after repair. The comparison results are used to correct model parameters, achieving full-process self-optimization.
[0104] Example 2: The method provided by this invention was verified online throughout the entire service life of a magnesium alloy automotive dashboard crossbeam die-casting mold. The mold cavity material is H13 hot work die steel. According to S110, strain gauge tie rod force sensors were installed on the four tie rods of the die-casting machine tailplate to obtain the total clamping force. Sixteen armored K-type thermocouples were embedded 2 mm deep from the cavity surface along the gate, runner end, and easily eroded areas to measure temperature. An IEPE accelerometer with a resonant frequency of 50kHz was installed on the flat machined surface near the gate on the back of the fixed mold to collect vibration signals. The sampling rate for clamping force was 1kHz, the temperature sampling rate was 100Hz, and the vibration sampling rate was 200kHz. The baseline mold calibration was performed after the mold was preheated and 100 consecutive stable production runs were completed. The average value of the peak clamping force from the subsequent 50 consecutive defect-free mold runs was taken as the baseline. .
[0105] Based on S210, a process scenario database is built within 8,000 qualified mold runs in the initial stage of mold production, including injection speed. With a step size of 0.1 m / s and casting pressure With a step size of 2MPa and a pouring temperature Discretization was performed in 5°C increments, resulting in 12 typical process scenarios. In each scenario, a preset multiplier for the dynamic triggering upper limit, determined according to S230, was set... , , During the online monitoring phase, the process scenario is matched to each module according to S200 and calculations are performed. , , According to the S300 setting judgment window , , specific multiple The coupling strength threshold is 0.8. Early warning sequence window in S400. Continuous threshold Disturbance at low speed: 3.0 m / s; swing amplitude threshold. Flash edge increment threshold The erosion wear model parameters in S500 were calibrated using previous accelerated tests. , , Roughness failure threshold The crack propagation model parameters are based on the initial crack data of the mold, which are used to determine the basic propagation rate coefficient. Acceleration factor weight saturated crack length Shape parameters Maintenance limits The cavity deformation model was fitted from offline coordinate measuring machine data. , Tolerance zone half width .
[0106] During the complete service life of the mold, which has accumulated 115,000 production cycles, the system marked 7 initial attenuation tendencies. The warning distribution and corresponding ternary indicators after the S300 three-level progressive judgment are shown in Table 1.
[0107] Table 1. Warning Events and S300 Level 3 Judgment Records In Table 1, the fifth warning sequence, constructed in S410, showed alternating crack propagation and cavity deformation. A process parameter disturbance cycle was initiated according to S420. The original injection speed for this cycle was 4.5 m / s. Following S430, the injection speed was fixed at a low speed of 3.0 m / s while maintaining a constant holding time. Five consecutive disturbance cycles were produced, and the collected data are shown in Table 2. The attenuation rate and swing amplitude after disturbance are also shown. , flare increment ,because and According to S440, the current module is re-marked as a crack propagation warning, and the swing amplitude is fed back to S200, corresponding to the process scenario. Down 18%.
[0108] Table 2 Measured data of S430 process parameters under disturbance cycling Each warning type, verified by S400, is subject to remaining module prediction according to S500. Erosion and wear warnings (warnings 1 and 3) use the current coupling index. Calculate according to formula (5) ; Frequency of dense multi-peak pattern occurrence in crack expansion early warning (early warning 2, 4, 5) statistics Substitute the logistic model and solve according to formula (6). The cavity deformation early warning (early warning 6, 7) is calculated by mapping the dimensional deviation from the attenuation rate sequence according to formula (7). The comparison between the predicted results and the actual maintenance status of the mold is shown in Table 3.
[0109] Table 3 Comparison of predicted remaining mold repair times and actual mold maintenance status As shown in Table 3, both erosion wear warnings were detected before actual failure. After the second warning, the measured surface roughness of the cavity during polishing repair at 112,000 mold cycles highly matched the predicted failure threshold. Crack propagation, after verification via perturbation, was tracked by three warnings for the same crack evolution, ultimately reaching the repair limit at 107,000 mold cycles, with prediction errors all less than 5%. Cavity deformation triggered a warning at 101,200 mold cycles, predicting 970 remaining cycles. In reality, during mold replacement at 102,000 cycles, the critical dimensional deviation was close to the tolerance zone boundary, avoiding batch deviations. No unplanned downtime occurred during the entire service life of the mold. The S540 output graded control commands are shown in Table 4.
[0110] Table 4 S540 Hierarchical Control Command Output Record Verification results show that the method of the present invention, through the complete technical solution of S100 ternary index construction, S200 dynamic process scenario matching, S300 three-level progressive judgment, S400 cyclic verification and disturbance correction, and S500 multi-model remaining module prediction, can accurately define the three life decay stages of magnesium alloy automotive die casting molds: erosion wear, crack propagation, and cavity deformation, and reliably predict the remaining usable modules before actual failure.
[0111] Example 3: A system for defining the phased degradation of the lifespan of magnesium alloy automotive die-casting molds, the system comprising: The health status ternary index acquisition module is used to acquire online the peak decay rate of the clamping force, the non-uniformity of the temperature field on the cavity surface, and the proportion of impact energy of vibration extracted by multivariate variational mode decomposition for each mode, and combine them into a health status ternary index. The initial attenuation tendency marking module is used to construct a process scenario database with injection speed, casting pressure and pouring temperature as key values using qualified modules in the early stage of normal service, and store the historical distribution of the three-dimensional health status indicators; it matches the current process scenario online, extracts the historical minimum value and generates a dynamic trigger upper limit according to a preset multiple; when any two indicators exceed the limit at the same time, it marks the initial attenuation tendency. The decay mode step-by-step judgment module is used to initiate a three-level step-by-step judgment after marking: First, if the decay rate exceeds a certain multiple of the dynamic trigger limit, a cavity deformation warning is output; Second, if the impact energy ratio changes from a sparse single-peak mode to a dense multi-peak mode, a crack propagation warning is output; Third, if the temperature field non-uniformity increases rapidly and is positively fed into the impact energy ratio increase, an erosion wear warning is output; If none of these conditions are met, the marking is removed and normal monitoring is resumed. The warning verification and disturbance correction module is used to verify the warning types in sequence; if a single warning is confirmed to be valid if it reaches a threshold in a series of consecutive times; when multiple warnings occur alternately, the process parameter disturbance cycle is started, the low-speed injection is fixed, the warning type is confirmed based on the attenuation rate swing amplitude and flash increment, and the dynamic trigger upper limit is corrected by the swing amplitude. The remaining module prediction and control module is used to predict the remaining available modules based on the confirmed warning type, and output control commands in real time based on the predicted remaining available modules and safety process requirements.
[0112] Those skilled in the art will understand that the embodiments of this application are provided as methods, systems, or computer program products. Therefore, this application takes the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application takes the form of a computer program product implemented on one or more computer storage media containing computer program code. The solutions in the embodiments of this application can be implemented using various computer languages, exemplified by the object-oriented programming language Java and the interpreted scripting language JavaScript, etc.
[0113] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.
[0114] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.
Claims
1. A method for defining the stages of magnesium alloy automotive die casting die life degradation, characterized by, include: The peak decay rate of the clamping force, the non-uniformity of the temperature field on the cavity surface, and the proportion of impact energy of vibration extracted by multivariate variational mode decomposition are obtained online for each module and combined into a three-dimensional health status index. Using qualified molds from the initial normal service period, a process scenario database is constructed with injection speed, casting pressure and pouring temperature as key values, storing the historical distribution of the three-element health status index; the current process scenario is matched online, the historical minimum value is extracted and a dynamic trigger limit is generated according to a preset multiple. When any two indicators exceed the limit simultaneously, a preliminary decline trend is indicated. After marking the initial attenuation tendency, a three-layer progressive judgment is initiated: the first layer outputs a cavity deformation warning if the attenuation rate exceeds a certain multiple of the dynamic trigger limit. The second layer outputs a crack propagation warning when the energy distribution of each mode in the impact-sensitive frequency band changes from a sparse single-peak mode to a dense multi-peak mode. The third layer outputs an erosion and wear warning if the rate of increase in temperature field nonuniformity accelerates and is positively coupled with the rate of increase in the proportion of impact energy; if neither condition is met, the marking is removed and normal monitoring is resumed. The warning types are sequenced for verification; If a single warning reaches the threshold consecutively, it is confirmed to be valid; when multiple warnings occur alternately, the process parameter disturbance cycle is started, the low-speed injection is fixed, the warning type is confirmed based on the swing amplitude of the attenuation rate and the increment of the flash length, and the dynamic trigger upper limit is corrected by the swing amplitude of the attenuation rate. The swing amplitude of the attenuation rate is the percentage decrease of the average attenuation rate during the disturbance period relative to the average attenuation rate of the last few modes before the disturbance was triggered. Based on the confirmed warning type, predict the remaining available modules, and based on the predicted remaining available modules and safety process requirements, output control commands in real time.
2. The method of claim 1, wherein the magnesium alloy automotive die casting mold life deterioration stage is defined by the steps of: The steps for obtaining the three health status indicators include: In each die casting cycle, the current die force peak of the current shot is extracted for the pressurization and holding phase , the reference die force peak calibrated at the initial stage of service , together with the following formula to calculate the decay rate : ; Select the continuous time interval after the mold is open to the position as the temperature sampling window, collect the temperature data of the key area of the cavity surface , calculate the average temperature of each temperature measurement point in the sampling window , and the arithmetic mean of all temperature measurement points , wherein i is the serial number of the temperature measurement point, i=1, 2, …, K, and the temperature field non-uniformity is calculated by substituting the following formula : ; Vibration signals from the injection and filling stages were extracted and decomposed into multiple variational mode shapes. Calculate the energy of each mode using one eigenmode function. and center frequency In the preset impact-sensitive frequency band The percentage of impact energy is extracted using the following formula. : ;in, For indicator functions, when the first The center frequency of each intrinsic mode function Located in the impact-sensitive frequency band At that time, The value is 1, otherwise The value is 0. To impact the lower limit frequency of the sensitive frequency band, In order to push the upper frequency limit of the sensitive frequency band, For indexing; attenuation rate Temperature field nonuniformity and impact energy ratio The data are combined into a three-dimensional vector, with the mold sequence number, injection speed, casting pressure, and pouring temperature appended, and then stored in a time-series database.
3. The method for phased determination of the life decay of magnesium alloy automotive die-casting molds according to claim 1, characterized in that, The steps for constructing the process scenario database and matching the current process scenario include: Taking qualified dies with no significant degradation in the early stages of mold service as the object, the three process parameters of injection speed, casting pressure, and pouring temperature are used as the combined key values. Each process parameter is discretized at a preset step size. Dies with identical combined key values after discretization are grouped into the same process scenario, and the decay rate sequence under each scenario is stored. Temperature field non-uniformity sequence and impact energy percentage sequence Basic statistics; The injection speed, casting pressure, and pouring temperature of the current mold are obtained online. After normalization using the standard deviation of the corresponding process parameters in the database, the Euclidean distance between the current process parameters and the combined key values of each scenario is calculated, and the process scenario with the smallest distance is selected as the matching scenario.
4. The method for phased determination of the life decay of magnesium alloy automotive die-casting molds according to claim 3, characterized in that, The initial attenuation tendency marking step includes: Extract the historical minimum value of attenuation rate from the matched process scenarios. The historical minimum value of temperature field nonuniformity and the historical minimum value of impact energy percentage Multiply by the attenuation rate, temperature field non-uniformity, and the preset ratio of impact energy, respectively. , , The dynamic trigger limit of the decay rate is obtained. Temperature field nonuniformity dynamic triggering upper limit Dynamic trigger limit for impact energy percentage ; when or or When any one of the following conditions is met, the current mode is marked as having an initial attenuation tendency, where, For attenuation rate, For temperature field non-uniformity, The percentage of impact energy.
5. The method for phased determination of the life decay of magnesium alloy automotive die-casting molds according to claim 1, characterized in that, The first step in the three-level progressive determination is to determine the attenuation rate of the current module. Does it meet the requirements? ,in To preset a specific multiple, This is the dynamic trigger limit for the attenuation rate under the current matching process scenario; If the conditions are met, it is determined that permanent deformation in the mold closing direction has caused dimensional deviation, and the cavity plastic deformation is the current dominant failure mechanism. A cavity deformation warning is then output and the current layer's determination is terminated.
6. The method for phased determination of the life decay of magnesium alloy automotive die-casting molds according to claim 1, characterized in that, The second level of the three-level hierarchical determination is: for the most recent consecutive... Each module, For a preset continuous mode window length, the energy proportion distribution of each mode within the impact-sensitive frequency band is extracted one by one, and the number and spacing of independent peaks are identified. If only one significant peak is detected in a single mode, it is marked as a sparse single-peak mode. If three or more significant peaks are detected and the frequency interval between the center of adjacent peaks is less than the dense threshold, it is marked as a dense multi-peak mode. When the first half of the mode within the window is dominated by the sparse single-peak mode and the second half is dominated by the dense multi-peak mode, it is determined that the energy proportion distribution of each mode within the impact-sensitive frequency band has completed the transformation from the sparse single-peak mode to the dense multi-peak mode, and a crack propagation warning is output.
7. The method for phased determination of the life decay of magnesium alloy automotive die-casting molds according to claim 1, characterized in that, The third level of the three-level hierarchical determination is: For the most recent consecutive The difference sequence for calculating the non-uniformity of the temperature field is obtained by performing multiple operations. The difference sequence of impact energy proportion ; By judgment The slope of the sequence is greater than zero and the second difference remains positive, confirming that the rate of increase in the non-uniformity of the temperature field is accelerating. Calculate the coupling strength using the following formula : ;in, and The difference sequences of temperature field inhomogeneity are respectively The difference sequence of impact energy proportion The mean, For the delay steps, The length of the continuous module window. This is the index of the module within the window. For the first The increase in the proportion of impact energy of each modal relative to the previous modal; when Difference sequences exceeding a preset coupling threshold and exhibiting non-uniformity in the temperature field. The difference sequence of impact energy proportion When the Pearson correlation coefficient exceeds the preset threshold, it is determined that the increase in temperature field nonuniformity and the increase in the proportion of impact energy form a positive feedback coupling, and an erosion wear warning is output.
8. The method for phased determination of the life decay of magnesium alloy automotive die-casting molds according to claim 1, characterized in that, The conditions for removing the marker and returning to normal monitoring are as follows: when the first, second and third level judgments in the three-level step-by-step judgment are not met, the preliminary attenuation tendency marker of the current mold is cleared, the preliminary attenuation tendency removal event is recorded, the mold returns to the normal monitoring state and repeats the process scenario matching and dynamic over-limit judgment.
9. The method for phased determination of the life decay of magnesium alloy automotive die-casting molds according to claim 1, characterized in that, The condition for confirming the validity of a single warning consecutive module reaching a threshold is: comparing the current module with the previous module. The combination of warning types for each module is called a warning sequence. When the warning sequence contains only the same warning type from erosion wear warning, crack propagation warning, or cavity deformation warning, and the number of consecutive occurrences is not less than a preset duration threshold, the warning sequence is considered to be in sequence. When the warning type is confirmed to be valid, the remaining prediction modules will proceed.
10. The method for phased determination of the life decay of magnesium alloy automotive die-casting molds according to claim 1, characterized in that, The steps of starting the process parameter disturbance cycle, fixing the low-speed injection, determining the warning type based on the swing amplitude of the attenuation rate and the increment of the flash length, and correcting the dynamic trigger upper limit by adjusting the swing amplitude of the attenuation rate include: When at least two different warning types appear in the warning sequence and the number of consecutive occurrences of each type is less than the alternation judgment value, based on the currently matched process scenario, the injection speed is fixed at a low speed and the holding time remains unchanged, and continuous production continues. Each module; Collect the peak value of the mold closing force during the disturbance and calculate the attenuation rate; measure the maximum length of the flash at the parting surface of each mold pass. Define the oscillation amplitude of the decay rate The increment of the flash length is defined as the percentage decrease in the mean attenuation rate during the disturbance period relative to the mean attenuation rate of the last few modes before the disturbance was triggered. This represents the absolute increase in the average flash length during the disturbance period relative to the average flash length of the historical normal mode. like Swing amplitude threshold exceeding the decay rate and No more than the threshold for significantly increasing flash. If so, the current module will be remarked as a crack propagation warning, and the swing amplitude will be fed back to correct the dynamic trigger limit of the attenuation rate under the current process scenario; like No more than and Exceed If so, the current module will be locked as an irreversible stage and the cavity deformation warning will be maintained; like Exceed and Exceed It is also locked into an irreversible stage and maintains early warning of cavity deformation.
11. The method for phased determination of the life decay of magnesium alloy automotive die-casting molds according to claim 1, characterized in that, The step of predicting the remaining available modules based on the confirmed warning type includes: When the warning type is erosion wear, calculate the coupling index. Substitute into the inverse power-exponential model Predict the remaining available modulo operations, where, For the remaining usable modules under erosion and wear warning, The failure threshold for cavity surface roughness. It is a proportionality constant. The coupling index is... For temperature field non-uniformity, To impact energy percentage, To reflect the material-process related index of the coupling index on the accelerating effect of roughness growth, A material-process related index that reflects the time-scale characteristics of roughness as it increases with the number of mold passes; These are the currently serving modules; When the warning type is crack expansion, the frequency of occurrence of the dense multi-peak pattern of statistical impact energy proportion is used to correct the logistic growth model. The spread rate coefficient in , making the crack feature length Crack length reached the repair limit Solve for the critical module to obtain the remaining module. ,in The saturation crack length, For shape parameters, For cumulative modules; When the warning type is cavity deformation, establish the critical cavity size deviation. With cumulative modulus Power law relationship This causes deviations in the critical cavity dimensions. Reaching half width of tolerance zone Solve for the critical module to obtain the remaining module. ,in and represents the fitting coefficient.
12. The method for phased determination of the life decay of magnesium alloy automotive die-casting molds according to claim 11, characterized in that, The control commands include: The predicted remaining available module counts are compared with the preset safety process requirements, and multiple risk levels are divided. A hierarchical control instruction is generated, which includes the mold number, current module count, warning type, estimated remaining module counts, and suggested handling solutions. This instruction is then pushed to the human-machine interface and the manufacturing execution system. At the same time, the warning results and actual maintenance status are fed back to the process scenario database for rolling optimization of the parameters in the remaining module count prediction model.
13. A system for defining the staged degradation of the lifespan of magnesium alloy automotive die-casting molds, characterized in that, A method for defining the phased degradation of the lifespan of magnesium alloy automotive die-casting molds as described in any one of claims 1 to 12, comprising: The health status ternary index acquisition module is used to acquire online the peak decay rate of the clamping force, the non-uniformity of the temperature field on the cavity surface, and the proportion of impact energy of vibration extracted by multivariate variational mode decomposition for each mode, and combine them into a health status ternary index. The initial attenuation tendency marking module is used to construct a process scenario database with injection speed, casting pressure and pouring temperature as key values using qualified modules in the early stage of normal service, and store the historical distribution of the three-dimensional health status indicators; it matches the current process scenario online, extracts the historical minimum value and generates a dynamic trigger upper limit according to a preset multiple; when any two indicators exceed the limit at the same time, it marks the initial attenuation tendency. The attenuation mode step-by-step judgment module is used to initiate a three-level step-by-step judgment after marking: First, if the attenuation rate exceeds a certain multiple of the dynamic trigger limit, a cavity deformation warning is output; Second, if the energy distribution of each mode in the impact-sensitive frequency band changes from a sparse single-peak mode to a dense multi-peak mode, a crack propagation warning is output; Third, if the non-uniformity of the temperature field accelerates and is positively fed into the impact energy proportion growth rate, an erosion wear warning is output; If none of these conditions are met, the marking is removed and normal monitoring is returned. The early warning verification and disturbance correction module is used to verify the early warning types in sequence; if a single early warning reaches a threshold for consecutive cycles, it is confirmed to be valid; when multiple early warnings occur alternately, a process parameter disturbance cycle is started, a fixed low-speed injection is used, and the early warning type is confirmed based on the swing amplitude of the attenuation rate and the increment of the flash length, and the dynamic trigger upper limit is corrected by the swing amplitude of the attenuation rate; the swing amplitude of the attenuation rate is the relative percentage decrease of the average attenuation rate during the disturbance period relative to the average attenuation rate of the last few cycles before the disturbance is triggered. The remaining module prediction and control module is used to predict the remaining available modules based on the confirmed warning type, and output control commands in real time based on the predicted remaining available modules and safety process requirements.
Citation Information
Patent Citations
Dynamic maintenance management method and system based on full life cycle of test equipment
CN121835377A
Improvements in and relating to prime movers
IES78912B2
