Mold life prediction system based on data analysis

Abstract

The present application relates to the technical field of equipment health management, in particular to a mold life prediction system based on data analysis, in the present application, through real-time analysis of the mechanical dynamics response and fluid stress variable of the stamping die closing transient, the precise quantification of single processing cycle load impact is realized, and the mechanical parameters of the mold base material and the interface lubrication characteristics are fused, the loss contribution degree operation of multi-factor coupling is executed, and then the physical wear increment of the cavity surface and the fatigue damage probability of the structure inside are deduced, the step-by-step deduction logic from micro-wear to macro-structure failure effectively overcomes the limitations of experience judgment, realizes the dynamic penetration insight of the mold health state, combines the production plan and historical repair data to carry out nonlinear attenuation iteration, accurately predicts the remaining processing life, greatly improves the accuracy of prediction and the foresight of maintenance decision, strengthens the whole life cycle asset management level, ensures the production continuity and optimizes the spare parts inventory cost.

Description

Technical Field

[0001] This invention relates to the field of equipment health management technology, and in particular to a mold life prediction system based on data analysis. Background Technology

[0002] Traditional mold life prediction systems are tools used to assess the wear and tear of molds during processing cycles such as stamping, injection molding, or die casting. Their technical focus is on determining the specific points at which molds fail due to wear, fatigue, or deformation. Traditional mold life prediction systems rely on manual recording of the number of processing cycles, or on technicians visually observing the mold surface roughness and cracks, as well as measuring the dimensional deviations of the finished product using calipers. A fixed upper limit for the service life is set based on empirical values. When the preset processing frequency is reached or significant geometric accuracy deviations are observed, the mold is considered to have reached the end of its lifespan, and offline replacement or repair is performed.

[0003] Traditional mold life prediction relies on manual recording of processing times and the experience-based visual assessment of technicians. This method is highly subjective and lacks precision, making it difficult to capture the actual wear and tear of the mold in real time. Furthermore, judgments based on finished product dimensional deviations are a form of delayed management, where problems are discovered only after products have already been scrapped and production has been interrupted. Setting a fixed upper limit for the service life cannot adapt to dynamic changes under different working conditions, often leading to premature mold disposal or unexpected failure. The lack of quantitative analysis of the evolution of internal stress and fatigue accumulation in the mold makes it impossible to proactively deploy maintenance plans, resulting in low overall management efficiency and uncontrollable risks. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a mold life prediction system based on data analysis.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a mold life prediction system based on data analysis includes: The mold closing load impact quantification module analyzes the stress variation amplitude within a single processing cycle based on the mechanical dynamic response of the stamping equipment actuator during mold closing transient and the fluid pressure holding stress variable, combined with the lateral displacement offset of the mold guide component, and generates a quantitative evaluation value of the mold closing transient dynamic load. The cavity substrate wear degradation analysis module extracts the mechanical parameters of the mold substrate, the physical specifications of the plate to be processed, and the interface lubrication characteristics from the big data storage unit based on the quantitative evaluation value of the transient dynamic load of the mold closing. It performs loss contribution matrix operation and weighted cumulative calculation to deduce the physical wear increment of the mold cavity surface and construct the cavity multi-factor coupled wear degradation prediction value. The mold internal structure fatigue damage prediction module is based on the multi-factor coupled wear degradation prediction value of the cavity, correlates the stress distribution characteristics of the mold support structure with the preload state of the elastic element, calculates the geometric coordinate offset vector of the stress concentration area, and generates the structural stress concentration fatigue damage failure probability. The nonlinear decay remaining processing quantity calculation module, based on the structural stress concentration fatigue damage failure probability, matches the dynamic parameter settings in the production plan with the original mold repair ledger, performs nonlinear exponential decay iterative calculation, and outputs the nonlinear remaining processing cycle life prediction quantity. The life-end and physical maintenance intervention module identifies the mold scrapping reference date based on the nonlinear remaining processing cycle life prediction, combined with the mold's unique identification code and stamping rated load schedule, reverse-engineers the spare parts logistics sequence, and outputs a mold full life-cycle maintenance intervention management table.

[0006] As a further aspect of the present invention, the mold closing load impact quantification module includes: The transient impact mechanics analysis submodule for mold closing analyzes the mechanical impact frequency and amplitude envelope during mold closing based on the mechanical dynamics feedback signal of the slide of the stamping equipment at the bottom dead center position, and extracts the peak impact force data. The fluid pressure holding stress peak calibration submodule, based on the impact force peak data, extracts the hydraulically driven pressure transmitter variable, extracts the pressure fluctuation change curve during the mold closing and pressure holding stage, and identifies and locks the fluid pressure holding stress peak. The load impact integral quantization submodule calls the peak value of the fluid pressure holding stress, combines it with the physical lateral force displacement fed back from the mold guide post, and performs dynamic time integral calculation to obtain the quantitative evaluation value of the transient dynamic load when the mold is closed.

[0007] As a further aspect of the present invention, the pressure fluctuation curve refers to the pressure fluctuation curve obtained by real-time acquisition of the voltage analog signal output by the pressure transmitter in the hydraulic drive circuit after the slide of the stamping equipment reaches the die closing dead point, converting the voltage analog signal into a standard pressure value sequence, and calculating the rate of change of the derivative of the standard pressure value sequence during the pressure holding duration.

[0008] As a further aspect of the present invention, the cavity substrate wear degradation analysis module includes: The multi-source medium loss contribution calculation submodule, based on the quantitative evaluation value of the mold closing transient dynamic load, retrieves the mold steel hardness index, stamping plate thickness specification and lubricating medium viscosity value from the big data storage unit, constructs a three-dimensional loss feature vector, performs element product operation, and obtains the independent contribution parameters of the participating items to the physical loss of the mold cavity surface. The micro wear depth increment extrapolation submodule, based on the independent contribution parameter and combined with the quantitative evaluation value of the mold closing transient dynamic load, assigns a preset mechanical weighting coefficient, compares it with the preset wear critical safety threshold, and calculates the change in wear depth on the mold cavity surface under the current processing batch. The cavity geometric accuracy degradation modeling submodule calls the change in wear depth on the mold cavity surface, combines it with the cumulative stamping frequency recorded by the mold mechanical controller, calculates the geometric accuracy degradation rate of the mold cavity surface, and establishes a multi-factor coupled wear degradation prediction value for the cavity.

[0009] As a further aspect of the present invention, the geometric accuracy degradation rate of the mold cavity surface refers to the change in the cumulative stamping frequency per unit time, and the change in the wear depth of the mold cavity surface is divided by the change per unit time to obtain the geometric accuracy degradation rate of the mold cavity surface.

[0010] As a further aspect of the present invention, the fatigue damage prediction module for the internal structure of the mold includes: The static stress distortion spatial mapping submodule calls the predicted value of multi-factor coupled wear degradation of the cavity, and combines it with the stress distribution variables of the bottom support component of the mold base plate to map the static stress distortion distribution parameters of the overall mold structure. The thermo-coupling stress concentration offset calculation submodule, based on the static stress distortion distribution parameters, combined with the unloading spring preload value and the thermodynamic temperature range inside the mold cavity, analyzes the structural space variation under thermo-coupling action and obtains the stress concentration point offset vector. The material fatigue limit fitting attenuation submodule calls the stress concentration point offset vector, matches the preset mold substrate fatigue limit attenuation curve, performs fatigue life distribution fitting calculation, and generates the structural stress concentration fatigue damage failure probability.

[0011] As a further aspect of the present invention, the matching of the preset fatigue limit decay curve of the mold substrate refers to retrieving a pre-stored chart in the storage unit showing the relationship between the stress amplitude decay of the mold substrate and the increase in the number of stress cycles, and comparing the local stress value corresponding to the stress concentration point offset vector with the stress amplitude decay relationship chart to extract the upper limit value of the allowable stress of the material under the current state, thereby obtaining the matching result of the preset fatigue limit decay curve of the mold substrate.

[0012] As a further aspect of the present invention, the nonlinear attenuation remaining processing amount calculation module includes: The nonlinear exponential decay benchmark configuration submodule calls the structural stress concentration fatigue damage failure probability, extracts the machining speed setting value and the mold rated maintenance interval duration from the current production instruction set, and obtains the nonlinear exponential decay initial parameters. The repair variable is introduced into the multi-round iterative submodule. Based on the nonlinear exponential decay initial parameter, the original number of physical repair and welding of the mold is introduced as a correction variable. Multi-round iterative decay calculation is performed to obtain the coordinates of the mold performance degradation cutoff point. The standard machining cycle allowance conversion submodule maps the difference between the current physical timestamp and the expected mechanical failure timestamp based on the coordinates of the mold performance degradation cutoff point, converts it into the number of times the standard stamping machine is executed, and obtains the nonlinear remaining machining cycle life prediction.

[0013] As a further aspect of the present invention, the life-end and physical maintenance intervention module includes: The rated load decay and scrapping benchmark scheduling submodule calls the nonlinear remaining processing cycle life prediction, associates the mold unique code with the stamping production plan, performs dimensionality reduction mapping based on the rated frequency per unit cycle, calculates the remaining physical working life of the mold, performs time-series accumulation in combination with the crane mold time scale, and performs dynamic correction by coupling the material fatigue load decay coefficient to generate the mold scrapping benchmark date. The spare parts circulation and maintenance timeline reverse calculation submodule calls the mold scrapping baseline date, extracts the standard quotas for physical circulation and safety stock consumption of spare parts in the stamping workshop, sets intervention action nodes for mold disassembly and physical maintenance, classifies the response level based on the real-time deviation between the node and the current date, and generates a physical intervention timeline for maintenance. The full-life structural loss matrix grid module, based on the maintenance physical intervention time axis, maps the original cumulative number of punches, cavity polishing records and repair geometric increments associated with the mold coding, constructs a multi-dimensional metal loss evolution matrix, embeds the maintenance intervention time axis at the end of the matrix, and integrates the frequency distribution characteristics of mechanical faults for probability weighting, generating a mold full life cycle maintenance intervention management table.

[0014] As a further aspect of the present invention, the intervention action node for mold disassembly and physical maintenance refers to the procurement lead time for vulnerable parts in the standard quota for physical circulation and safety stock consumption of spare parts in the stamping workshop. Before the mold scrapping benchmark date, the intervention action node for mold disassembly and physical maintenance is obtained by reverse time offset calculation based on the procurement lead time and the mold shutdown and maintenance cycle. The multidimensional metal loss evolution matrix refers to the extraction of the original cumulative number of strokes associated with the mold code as the time dimension coordinate, and the subtraction depth value in the cavity polishing record and the additive height value in the geometric increment of the repair welding as the spatial dimension variables, which are arranged in time sequence to generate a numerical matrix describing the changes in the mold surface morphology, thus obtaining the multidimensional metal loss evolution matrix.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, the mechanical dynamics response and fluid stress variables of the stamping die closing transient are analyzed in real time to accurately quantify the impact of a single processing cycle load. By integrating the mechanical parameters of the die substrate and the interface lubrication characteristics, a multi-factor coupled loss contribution calculation is performed to deduce the physical wear increment on the cavity surface and the probability of fatigue damage inside the structure. This step-by-step deduction logic from micro wear to macro structural failure effectively overcomes the limitations of experience-based judgment and achieves a dynamic and penetrating insight into the health status of the die. Combined with production plans and historical repair data, nonlinear decay iteration is performed to accurately predict the remaining processing life. Finally, an intervention plan including spare parts logistics and maintenance nodes is generated in reverse, which greatly improves the accuracy of prediction and the foresight of maintenance decisions, strengthens the level of full life cycle asset management, ensures production continuity, and optimizes spare parts inventory costs. Attached Figure Description

[0016] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0018] Please see Figure 1 The data analysis-based mold life prediction system includes: The mold closing load impact quantification module analyzes the stress variation amplitude within a single processing cycle based on the mechanical dynamic response of the stamping equipment actuator during mold closing transient and the fluid pressure holding stress variable, combined with the lateral displacement offset of the mold guide component, and generates a quantitative evaluation value of the mold closing transient dynamic load. The cavity substrate wear degradation analysis module extracts the mechanical parameters of the mold substrate, the physical specifications of the sheet material to be processed, and the interface lubrication characteristics from the big data storage unit based on the quantitative evaluation value of the transient dynamic load of the mold closing. It then performs loss contribution matrix operation and weighted cumulative calculation to deduce the physical wear increment of the mold cavity surface and construct the cavity multi-factor coupled wear degradation prediction value. The mold internal structure fatigue damage prediction module is based on the multi-factor coupled wear degradation prediction value of the cavity, correlates the stress distribution characteristics of the mold support structure with the preload state of the elastic element, calculates the geometric coordinate offset vector of the stress concentration area, and generates the structural stress concentration fatigue damage failure probability. The nonlinear decay remaining processing quantity calculation module calculates the nonlinear decay remaining processing quantity based on the probability of structural stress concentration fatigue damage failure, matches the dynamic parameter settings in the production plan with the original mold repair ledger, performs nonlinear exponential decay iterative calculation, and outputs the nonlinear remaining processing cycle life prediction quantity. The end-of-life and physical maintenance intervention module is based on the nonlinear remaining processing cycle life prediction, combined with the mold's unique identification code and stamping rated load schedule, to identify the mold's scrapping reference date, reverse the spare parts logistics sequence, and output a mold full life cycle maintenance intervention management table.

[0019] The mold closing load impact quantification module includes: The transient impact mechanics analysis submodule for mold closing analyzes the mechanical impact frequency and amplitude envelope during mold closing based on the mechanical dynamics feedback signal of the slide of the stamping equipment at the bottom dead center position, and extracts the peak impact force data. The voltage time-series analog signal output from the piezoelectric force sensor installed at the slider connecting rod of the stamping equipment is extracted, and the analog signal within the sampling time window is obtained by setting a sampling frequency. The voltage time-series analog signal is input to the front-end digital signal conditioning component, and a first-order median filtering algorithm is called to remove interference components in the original analog signal that are higher than the set threshold, thus extracting the impact voltage sequence. The voltage-to-mechanical conversion coefficient calibrated in the storage unit is called, and the discrete voltage values ​​in the impact voltage sequence are multiplied with the voltage-to-mechanical conversion coefficient to generate a mechanical dynamics feedback signal sequence in the time domain. The initial mechanical force state parameters are obtained by extracting the discrete voltage values ​​and substituting them into the conversion coefficient for multiplication. The signal sequence in the time domain is input to a frequency domain analyzer based on a fast Fourier transform architecture to extract the energy distribution bands within the frequency band, identify the center frequency value corresponding to the band, and calibrate it as the mechanical impact frequency. The Hilbert transform component is called to calculate the analytical signal components at discrete time points for the mechanical dynamics feedback signal sequence, and the set of absolute values ​​of the analytical signal components is extracted as the amplitude envelope. Traverse the numerical nodes in the amplitude envelope, execute the maximum difference comparison logic of adjacent nodes, lock the highest discrete point where the slope of the numerical mutation changes from positive to negative, and confirm the mechanical value corresponding to the highest discrete point as the peak impact force data.

[0020] The fluid pressure holding stress peak calibration submodule extracts the hydraulically driven pressure transmitter variable based on the impact force peak data, extracts the pressure fluctuation change curve during the mold closing and pressure holding stage, and identifies and locks the fluid pressure holding stress peak. The pressure fluctuation curve refers to the pressure fluctuation curve obtained by real-time acquisition of the voltage analog signal output by the pressure transmitter in the hydraulic drive circuit after the slide of the stamping equipment reaches the die closing dead point, converting the voltage analog signal into a standard pressure value sequence, and calculating the rate of change of the derivative of the standard pressure value sequence during the pressure holding duration. The analog electrical signal output by the pressure transmitter in the hydraulic drive circuit after the mold closing dead point is acquired. The DC voltage component of the analog electrical signal is extracted and identified as the pressure transmission variable. The calibration table is retrieved to obtain the range conversion ratio parameter and zero-point drift compensation reference value of the pressure transmitter. The DC voltage component is multiplied by the range conversion ratio parameter to obtain the conversion value. The conversion value is then summed with the zero-point drift compensation reference value to obtain the standard pressure value. Applying the above derivation logic to input analog parameters and performing multiplication and summation operations, the required reference pressure parameter can be output. The calculation is applied to the acquisition nodes within the pressure holding time period to generate a standard pressure value sequence. Adjacent pressure data in the standard pressure value sequence are extracted sequentially. The pressure change is obtained by subtracting the pressure data of the preceding time node from the pressure data of the following time node. The pressure change is then divided by the time step to obtain the derivative rate of change within the time period. The rate of change of the derivative is connected and spliced ​​along the time axis to generate a pressure fluctuation curve. The pressure data corresponding to the local maximum points in the pressure fluctuation curve is read and the arithmetic mean is calculated. The arithmetic mean is summed with the set error tolerance value, and finally the peak value of the fluid pressure holding stress is extracted.

[0021] The load impact integration quantization submodule calls the peak value of fluid pressure holding stress, combines it with the physical lateral force displacement fed back from the side of the mold guide post, performs dynamic time integration calculation, and obtains the quantitative evaluation value of the transient dynamic load when the mold is closed. The peak impact force data and the peak fluid holding stress are extracted. The lateral force displacement output from the displacement sensor on the side of the mold guide pillar is read. The cross-sectional area parameter of the mold hydraulic cylinder is retrieved. The peak fluid holding stress and the cross-sectional area parameter are multiplied to obtain the equivalent support force. After substituting the extracted peak parameters and cross-sectional area data into the product operation, the equivalent support force variable value can be directly calculated. The peak impact force data is extracted, and the peak impact force data and the equivalent support force are summed to generate the transient comprehensive force of mold closing. The transient comprehensive force of mold closing and the lateral force displacement are discretely multiplied and accumulated. The lateral force displacement is divided into discrete displacement intervals. Within the discrete displacement intervals, the transient comprehensive force of mold closing is multiplied by the length of the displacement interval to output the single-step work. The physical parameters of the single-step work can be directly obtained by multiplying the displacement interval length parameter and the comprehensive force variable. The single-step work is summed and accumulated to complete the dynamic time integration operation and obtain the quantitative evaluation value of the transient dynamic load of mold closing.

[0022] The cavity substrate wear degradation analysis module includes: The multi-source medium loss contribution calculation submodule, based on the quantitative evaluation value of the transient dynamic load of mold closing, retrieves the hardness index of mold steel, the thickness specification of stamping plate and the viscosity value of lubricating medium from the big data storage unit, constructs a three-dimensional loss feature vector, performs element-wise product operation, and obtains the independent contribution parameters of the participating items to the physical loss of the mold cavity surface. The transient dynamic load quantification evaluation value of the mold closing mechanism is invoked, and a data retrieval command is sent to the storage unit. The attribute parameters corresponding to the processing task are received from the storage unit, and the hardness index of the mold steel, the thickness specification of the stamping sheet, and the viscosity value of the lubricating medium are extracted to obtain the reference constants. The hardness index of the mold steel is divided by the first reference constant to obtain the dimensionless hardness scalar; the thickness specification of the stamping sheet is divided by the second reference constant to obtain the dimensionless thickness scalar; and the viscosity value of the lubricating medium is divided by the third reference constant to obtain the dimensionless viscosity scalar. Extracting the hardness index value and dividing it by the corresponding reference constant yields the dimensionless scalar result. The dimensionless hardness scalar, thickness scalar, and viscosity scalar are combined in the correct order to construct a three-dimensional loss feature vector. The weight vector is extracted, and element-wise multiplication is performed on the corresponding elements in the three-dimensional loss feature vector. Multiplying the elements of each dimension yields an array of independent contribution parameters to the physical loss of the mold cavity surface. Multiplying the weight components by their corresponding feature elements outputs the independent contribution parameters.

[0023] The micro wear depth increment extrapolation submodule, based on the independent contribution parameter and combined with the quantitative evaluation value of the mold closing transient dynamic load, assigns a preset mechanical weighting coefficient, compares it with the preset wear critical safety threshold, and calculates the change in wear depth on the mold cavity surface under the current processing batch. The system receives an array of independent contribution values, calls the quantified evaluation value of the transient dynamic load during mold closing, and sums the values ​​in the independent contribution array to obtain the physical loss contribution factor. Extracting array elements and performing addition processing yields the required physical loss contribution factor. The system calls the weighting coefficient matrix, matches the mechanical weighting coefficients according to the numerical range of the physical loss contribution factor, and sequentially multiplies the quantified evaluation value of the transient dynamic load during mold closing with the physical loss contribution factor and the mechanical weighting coefficients to calculate the theoretical change in wear depth on the mold cavity surface. Substituting the quantified load evaluation value, relevant contribution factors, and weighting coefficients into the multiplication formula outputs the theoretical change in wear depth parameter. The system extracts the critical safety threshold for wear and compares the calculated theoretical change in wear depth with this critical safety threshold for verification. Once the comparison confirms that the theoretical change in wear depth is less than the safety threshold, the output is the change in wear depth on the mold cavity surface.

[0024] The cavity geometric accuracy degradation modeling submodule calls the change in wear depth on the mold cavity surface, combines it with the cumulative stamping frequency recorded by the mold mechanical controller, calculates the geometric accuracy degradation rate of the mold cavity surface, and establishes a multi-factor coupled wear degradation prediction value for the cavity. The rate of degradation of the geometric accuracy of the mold cavity surface refers to the change in the cumulative stamping frequency per unit time. The change in the wear depth of the mold cavity surface is divided by the change per unit time to obtain the rate of degradation of the geometric accuracy of the mold cavity surface. The process involves extracting the change in wear depth on the mold cavity surface, reading the cumulative stamping frequency recorded in the mold mechanical controller, using the internal clock component to obtain the duration per unit time, calculating the change in cumulative stamping frequency per unit time, and subtracting the recorded historical stamping frequency from the current cumulative stamping frequency to obtain the new stamping frequency. Substituting the current cumulative value with the historical value and performing a subtraction operation yields the new frequency per unit time. Dividing the change in wear depth on the mold cavity surface by the change in new stamping frequency yields the average single impact wear value of the mold cavity surface, which is defined as the geometric accuracy degradation rate. Extracting the depth change parameter and dividing it by the new frequency calculates the geometric accuracy degradation rate. A degradation correction coefficient is extracted, and the geometric accuracy degradation rate, degradation correction coefficient, and cumulative stamping frequency are multiplied together to establish a multi-factor coupled wear degradation prediction value for the mold cavity.

[0025] The mold internal structure fatigue damage prediction module includes: The static stress distortion spatial mapping submodule calls the cavity multi-factor coupled wear degradation prediction value, and combines it with the stress distribution variables of the bottom support component of the mold base plate to map the static stress distortion distribution parameters of the overall mold structure. The multi-factor coupled wear degradation prediction value of the cavity is called, and the force distribution variables fed back by the sensor array at the four corners of the bottom of the mold base plate are extracted. The force values ​​of the support points obtained by the array are summed to calculate the total load of the bottom support. The force value of each support point is divided by the total load of the bottom support to obtain the azimuth static off-center load coefficient. The off-center load coefficient is obtained by performing a division operation between the mechanical data of the support points and the total load parameter. The off-center load coefficient is multiplied by the reciprocal of the elastic modulus to obtain the initial elastic deformation scalar. The multi-factor coupled wear degradation prediction value is converted into metric units, and summed with the initial elastic deformation scalar of the azimuth. The degradation amount is superimposed on the structural deformation to output a vector set containing the azimuth distortion variable, which is mapped to the static force distortion distribution parameters of the overall mold structure.

[0026] The thermo-coupling stress concentration offset calculation submodule, based on static stress distortion distribution parameters, combined with the preload force of the unloading spring and the thermodynamic temperature range inside the mold cavity, analyzes the structural space variation under thermo-coupling action and obtains the stress concentration point offset vector. The system reads static stress distortion distribution parameters, calls the pre-stored unloading spring preload value in the controller, extracts the upper and lower limits of the thermodynamic temperature range, and subtracts the lower limit from the upper limit to obtain the internal temperature fluctuation amplitude. It then extracts the boundary temperature parameters and performs a difference operation to obtain the temperature fluctuation amplitude variable. Next, it retrieves the linear thermal expansion constant of the mold substrate from the material thermal expansion coefficient library, and performs a multiplication operation on the temperature fluctuation amplitude, the linear thermal expansion constant, and the characteristic length of the mold cavity to obtain the thermodynamic spatial expansion variation. Finally, it inputs the characteristic length and expansion constant along with the temperature amplitude into the multiplication formula to obtain the thermal expansion variation. It then extracts the component values ​​of the stress distortion distribution parameters and multiplies them with the unloading spring preload value to calculate the mechanical distortion equivalent. Finally, it performs vector superposition calculation on the mechanical distortion equivalent and the thermodynamic spatial expansion variation, extracting the superposition vector magnitude and azimuth angle to obtain the stress concentration point offset vector.

[0027] The material fatigue limit fitting attenuation submodule calls the stress concentration point offset vector, matches the preset mold substrate fatigue limit attenuation curve, performs fatigue life distribution fitting calculation, and generates the structural stress concentration fatigue damage failure probability. Matching the preset fatigue limit decay curve of the mold substrate refers to retrieving the pre-stored chart in the storage unit about the relationship between the stress amplitude decay of the mold substrate and the increase of the number of stress cycles. By comparing the local stress value corresponding to the offset vector of the stress concentration point with the stress amplitude decay relationship chart, the upper limit value of the allowable stress of the material under the current state is extracted, and the matching result of the preset fatigue limit decay curve of the mold substrate is obtained. The system retrieves the stress amplitude decay chart of the mold substrate as a function of temperature fluctuations and stress cycle count from the storage unit by calling the stress concentration point offset vector. The modulus value of the offset vector is extracted and multiplied with the substrate deformation resistance coefficient to obtain the local stress value. Multiplying the extracted vector modulus variable with the deformation resistance coefficient directly maps and outputs the local stress value. Using the local stress value and the cumulative number of stamping cycles as conditions, a comparison logic is executed in the stress amplitude decay chart to extract the upper limit of the allowable stress of the material, obtaining the matching result of the fatigue limit decay curve of the mold substrate. Combining the parameters obtained from the lookup table, the allowable stress limit of the material can be located. The local stress value is divided by the upper limit of the allowable stress obtained from the lookup table to obtain the local stress load rate. The Weibull distribution fitting algorithm is called, and the local stress load rate is input into the model parameter dimension to calculate the cumulative failure distribution function value, generating the probability of structural stress concentration fatigue damage failure.

[0028] The nonlinear attenuation remaining processing amount calculation module includes: The nonlinear exponential decay benchmark configuration submodule calls the structural stress concentration fatigue damage failure probability, extracts the machining speed setting value and the mold rated maintenance interval duration from the current production instruction set, and obtains the nonlinear exponential decay initial parameters. The system receives the probability of fatigue damage failure due to structural stress concentration, extracts the machining speed setpoint from the production instruction set and the mold maintenance interval duration from the maintenance record, extracts the probability of fatigue damage failure due to structural stress concentration, and performs a difference operation with a numerical constant to obtain the basic scalar value of the mold's remaining health. Subtracting the failure probability parameter from the setpoint yields the basic scalar value of the remaining health. Multiplying the machining speed setpoint and the mold maintenance interval duration yields the total full-load stamping processing volume within the maintenance cycle. Substituting the speed setpoint and duration parameter into the product, the total processing volume is output. Multiplying the total full-load stamping processing volume with the decay acceleration constant yields the nonlinear basic base factor. This basic base factor and the basic scalar value of the remaining health are combined into a configuration vector, calibrated as the initial parameter for nonlinear exponential decay, and saved to the cache queue.

[0029] The repair variable is introduced into the multi-round iterative submodule. Based on the nonlinear exponential decay initial parameter, the original number of physical repair and welding of the mold is introduced as the correction variable. Multi-round iterative decay calculation is performed to obtain the coordinates of the mold performance degradation cutoff point. The process involves reading the initial parameters of nonlinear exponential decay, extracting the physical repair and welding count records corresponding to the mold code, setting the physical repair and welding count as a correction variable, multiplying it with the welding strength weakening coefficient to obtain the repair loss rate, and performing a difference operation between the constant value and the repair loss rate to obtain the structural robustness coefficient after repair. Substituting the repair count variable and the weakening coefficient into the product rule, the loss rate parameter is calculated. The remaining health baseline scalar is extracted and multiplied with the structural robustness coefficient to obtain the corrected health scalar. The health parameter and robustness coefficient are multiplied to output the corrected scalar result. An iterative loop architecture is constructed, extracting the health scalar of the current cycle and performing an exponential operation with the baseline factor for estimation. Iteration stops when the estimated simulated degradation slope exceeds the scrap slope threshold. The iteration step number that triggers the stopping condition is recorded. The iteration step number is used as the x-axis and the corrected health scalar as the y-axis to synthesize the coordinates of the mold performance degradation cutoff point in two-dimensional space.

[0030] The standard machining cycle allowance conversion submodule maps the difference between the current physical timestamp and the expected mechanical failure timestamp based on the coordinates of the mold performance degradation cutoff point, converts it into the number of executions of the standard stamping machine, and obtains the nonlinear remaining machining cycle life prediction. The system receives the coordinates of the mold performance degradation cutoff point, retrieves the current physical timestamp data from the local clock, extracts the time conversion ratio constant based on the x-coordinate value of the mold performance degradation cutoff point coordinates to calculate the expected mechanical failure timestamp, and subtracts the current physical timestamp from the expected mechanical failure timestamp to obtain the absolute physical survival time difference. It then extracts the x-coordinate parameter and combines it with the conversion ratio constant to obtain the timestamp variable, before performing subtraction to obtain the survival time difference. It reads the hourly stamping cycle time parameter and the equipment comprehensive efficiency index value of the workshop machine tool, performs a product operation on the hourly stamping cycle time parameter and the equipment comprehensive efficiency index value to calculate the effective processing rate. Combining the cycle time parameter and efficiency index, it performs a product calculation logic operation to directly output the effective processing rate parameter. Finally, it performs a product conversion between the absolute physical survival time difference and the effective processing rate, converting the time dimension to the machine execution count dimension to obtain the nonlinear remaining processing cycle life prediction.

[0031] The end-of-life and physical maintenance intervention module includes: The rated load decay and scrapping benchmark scheduling submodule calls the nonlinear remaining processing cycle life prediction, associates the mold unique code with the stamping production plan, performs dimensionality reduction mapping based on the rated frequency per unit cycle, calculates the remaining physical working life of the mold, performs time-series accumulation by combining the time scale of the upper mold on the crane, and performs dynamic correction by coupling the material fatigue load decay coefficient to generate the mold scrapping benchmark date. The process involves calling the nonlinear remaining machining cycle life prediction, associating it with the stamping production schedule task book bound to the unique mold code, summarizing the daily planned production frequency in the task book, and performing a dimensionality reduction mapping by dividing the nonlinear remaining machining cycle life prediction by the daily planned production frequency to obtain the uncorrected remaining physical working days of the mold. The uncorrected working days parameter is immediately calculated by dividing the life prediction by the daily average production frequency variable. The upper die time stamp date of the mold installation on the stamping equipment is extracted, and a time-series accumulation operation is performed on this upper die time stamp date and the uncorrected working days to obtain the preliminary scheduling date. The fatigue load attenuation coefficient is extracted based on the mold material, and the attenuation coefficient is multiplied by the uncorrected working days to obtain the negative attenuation time-day compensation amount. This time-series compensation amount is subtracted from the preliminary scheduling date for time-series correction. Finally, time-series accumulation is performed using the time stamp date and working days, combined with backtracking processing using the attenuation compensation amount, to generate the mold scrapping baseline date.

[0032] The spare parts circulation and maintenance sequence reverse calculation submodule calls the mold scrapping baseline date, extracts the standard quotas for the physical circulation of spare parts and the consumption of safety stock in the stamping workshop, sets the intervention action nodes for mold disassembly and physical maintenance, classifies the response level based on the real-time deviation between the node and the current date, and generates a maintenance physical intervention timeline. The intervention action nodes for mold disassembly and physical maintenance refer to the procurement lead time for vulnerable parts in the standard quota for the physical circulation of spare parts and the consumption of safety stock in the stamping workshop. Before the mold scrapping benchmark date, the intervention action nodes for mold disassembly and physical maintenance are obtained by reverse time offset calculation based on the procurement lead time and the mold downtime maintenance cycle. The process involves reading the mold scrapping baseline date, extracting the standard quota document for spare parts physical circulation and inventory consumption, extracting the lead time parameters for core vulnerable parts procurement in days and the mold downtime maintenance cycle parameters recorded in the quota document, and summing these parameters to obtain the total maintenance lead time redundancy. The lead time parameters and downtime cycle variables are then combined and added to output the total redundancy value. Using the mold scrapping baseline date as the baseline anchor point, a reverse time offset calculation is performed. The total maintenance lead time redundancy is calculated by subtracting the total maintenance lead time redundancy from the calendar value of the mold scrapping baseline date. The intervention action node date for mold disassembly and physical maintenance is then calculated. Using the scrapping date data, a time rollback operation is performed to deduct the redundancy and deduce the node date. The current real-time date is retrieved, and the intervention action node date is subtracted from the current real-time date to obtain the deviation in days. This deviation is compared with the response level division interval array to calibrate the early warning response level. The node date and response level label are then concatenated with the mold code to generate a maintenance physical intervention timeline.

[0033] The full life-cycle structural loss matrix grid module, based on the maintenance physical intervention time axis, maps the original cumulative number of punches, cavity polishing records and repair geometric increments associated with mold coding, constructs a multi-dimensional metal loss evolution matrix, embeds the maintenance intervention time axis at the end of the matrix, and integrates the frequency distribution characteristics of mechanical failures for probability weighting to generate a mold full life-cycle maintenance intervention management table. The multidimensional metal loss evolution matrix refers to extracting the original cumulative number of strokes associated with the mold code as the time dimension coordinate, and taking the subtraction depth value in the cavity polishing record and the additive height value in the geometric increment of the repair welding as the spatial dimension variables, and arranging them in time sequence to generate a numerical matrix describing the changes in the mold surface morphology, thus obtaining the multidimensional metal loss evolution matrix. The process involves retrieving the mold's unique code from the maintenance physical intervention timeline, sending this code to the manufacturing execution end, extracting the original cumulative stamping count sequence as a one-dimensional coordinate vector, extracting the subtraction depth value sequence from historical cavity polishing records, and extracting the additive height value sequence from repair welding records. The subtraction depth value is defined as a negative spatial variable, and the additive height value as a positive spatial variable. By establishing subtraction data as a negative parameter and additive data as a positive parameter, a two-way calibration is achieved. A multi-dimensional metal loss evolution matrix is ​​generated using the original cumulative stamping count as the row index and the positive and negative spatial variables as the column values. The maintenance physical intervention timeline is embedded as the data record row at the end of the multi-dimensional metal loss evolution matrix. The proportion of historical fault counts to the total number of work batches is calculated to obtain the mechanical fault frequency distribution characteristic constant. The spatial variables and fault characteristic constants in the multi-dimensional matrix are multiplied and weighted to complete the risk ratio scaling transformation of the matrix elements. The processed matrix data is then converted into a report format and output as a mold lifecycle maintenance intervention management table.

[0034] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A mold life prediction system based on data analysis, characterized in that, The system includes: The mold closing load impact quantification module analyzes the stress variation amplitude within a single processing cycle based on the mechanical dynamic response of the stamping equipment actuator during mold closing transient and the fluid pressure holding stress variable, combined with the lateral displacement offset of the mold guide component, and generates a quantitative evaluation value of the mold closing transient dynamic load. The cavity substrate wear degradation analysis module extracts the mechanical parameters of the mold substrate, the physical specifications of the plate to be processed, and the interface lubrication characteristics from the big data storage unit based on the quantitative evaluation value of the transient dynamic load of the mold closing. It performs loss contribution matrix operation and weighted cumulative calculation to deduce the physical wear increment of the mold cavity surface and construct the cavity multi-factor coupled wear degradation prediction value. The mold internal structure fatigue damage prediction module is based on the multi-factor coupled wear degradation prediction value of the cavity, correlates the stress distribution characteristics of the mold support structure with the preload state of the elastic element, calculates the geometric coordinate offset vector of the stress concentration area, and generates the structural stress concentration fatigue damage failure probability. The nonlinear decay remaining processing quantity calculation module, based on the structural stress concentration fatigue damage failure probability, matches the dynamic parameter settings in the production plan with the original mold repair ledger, performs nonlinear exponential decay iterative calculation, and outputs the nonlinear remaining processing cycle life prediction quantity. The life-end and physical maintenance intervention module identifies the mold scrapping reference date based on the nonlinear remaining processing cycle life prediction, combined with the mold's unique identification code and stamping rated load schedule, reverse-engineers the spare parts logistics sequence, and outputs a mold full life-cycle maintenance intervention management table.

2. The mold life prediction system based on data analysis according to claim 1, characterized in that, The mold closing load impact quantification module includes: The transient impact mechanics analysis submodule for mold closing analyzes the mechanical impact frequency and amplitude envelope during mold closing based on the mechanical dynamics feedback signal of the slide of the stamping equipment at the bottom dead center position, and extracts the peak impact force data. The fluid pressure holding stress peak calibration submodule, based on the impact force peak data, extracts the hydraulically driven pressure transmitter variable, extracts the pressure fluctuation change curve during the mold closing and pressure holding stage, and identifies and locks the fluid pressure holding stress peak. The load impact integral quantization submodule calls the peak value of the fluid pressure holding stress, combines it with the physical lateral force displacement fed back from the mold guide post, and performs dynamic time integral calculation to obtain the quantitative evaluation value of the transient dynamic load when the mold is closed.

3. The mold life prediction system based on data analysis according to claim 2, characterized in that, The pressure fluctuation curve refers to the pressure fluctuation curve obtained by real-time acquisition of the voltage analog signal output by the pressure transmitter in the hydraulic drive circuit after the slide of the stamping equipment moves to the die closing dead point, converting the voltage analog signal into a standard pressure value sequence, and calculating the rate of change of the derivative of the standard pressure value sequence during the pressure holding duration.

4. The mold life prediction system based on data analysis according to claim 2, characterized in that, The cavity substrate wear degradation analysis module includes: The multi-source medium loss contribution calculation submodule, based on the quantitative evaluation value of the mold closing transient dynamic load, retrieves the mold steel hardness index, stamping plate thickness specification and lubricating medium viscosity value from the big data storage unit, constructs a three-dimensional loss feature vector, performs element product operation, and obtains the independent contribution parameters of the participating items to the physical loss of the mold cavity surface. The micro wear depth increment extrapolation submodule, based on the independent contribution parameter and combined with the quantitative evaluation value of the mold closing transient dynamic load, assigns a preset mechanical weighting coefficient, compares it with the preset wear critical safety threshold, and calculates the change in wear depth on the mold cavity surface under the current processing batch. The cavity geometric accuracy degradation modeling submodule calls the change in wear depth on the mold cavity surface, combines it with the cumulative stamping frequency recorded by the mold mechanical controller, calculates the geometric accuracy degradation rate of the mold cavity surface, and establishes a multi-factor coupled wear degradation prediction value for the cavity.

5. The mold life prediction system based on data analysis according to claim 4, characterized in that, The geometric accuracy degradation rate of the mold cavity surface refers to the change in the cumulative stamping frequency per unit time, which is obtained by dividing the change in the wear depth of the mold cavity surface by the change per unit time.

6. The mold life prediction system based on data analysis according to claim 4, characterized in that, The fatigue damage prediction module for the internal structure of the mold includes: The static stress distortion spatial mapping submodule calls the predicted value of multi-factor coupled wear degradation of the cavity, and combines it with the stress distribution variables of the bottom support component of the mold base plate to map the static stress distortion distribution parameters of the overall mold structure. The thermo-coupling stress concentration offset calculation submodule, based on the static stress distortion distribution parameters, combined with the preload force of the unloading spring and the thermodynamic temperature range inside the mold cavity, analyzes the structural space variation under thermo-coupling and obtains the stress concentration point offset vector. The material fatigue limit fitting attenuation submodule calls the stress concentration point offset vector, matches the preset mold substrate fatigue limit attenuation curve, performs fatigue life distribution fitting calculation, and generates the structural stress concentration fatigue damage failure probability.

7. The mold life prediction system based on data analysis according to claim 6, characterized in that, The matching of the preset fatigue limit decay curve of the mold substrate refers to retrieving the stress amplitude decay relationship chart of the mold substrate with temperature fluctuation and stress cycle number increase from the storage unit, comparing the local stress value corresponding to the stress concentration point offset vector with the stress amplitude decay relationship chart, extracting the upper limit value of the allowable stress of the material under the current state, and obtaining the matching result of the preset fatigue limit decay curve of the mold substrate.

8. The mold life prediction system based on data analysis according to claim 6, characterized in that, The nonlinear attenuation remaining processing amount calculation module includes: The nonlinear exponential decay benchmark configuration submodule calls the structural stress concentration fatigue damage failure probability, extracts the machining speed setting value and the mold rated maintenance interval duration from the current production instruction set, and obtains the nonlinear exponential decay initial parameters. The repair variable is introduced into the multi-round iterative submodule. Based on the nonlinear exponential decay initial parameter, the original number of physical repair and welding of the mold is introduced as a correction variable. Multi-round iterative decay calculation is performed to obtain the coordinates of the mold performance degradation cutoff point. The standard machining cycle allowance conversion submodule maps the difference between the current physical timestamp and the expected mechanical failure timestamp based on the coordinates of the mold performance degradation cutoff point, converts it into the number of times the standard stamping machine is executed, and obtains the nonlinear remaining machining cycle life prediction.

9. The mold life prediction system based on data analysis according to claim 8, characterized in that, The life-end and physical maintenance intervention module includes: The rated load decay and scrapping benchmark scheduling submodule calls the nonlinear remaining processing cycle life prediction, associates the mold unique code with the stamping production plan, performs dimensionality reduction mapping based on the rated frequency per unit cycle, calculates the remaining physical working life of the mold, performs time-series accumulation in combination with the crane mold time scale, and performs dynamic correction by coupling the material fatigue load decay coefficient to generate the mold scrapping benchmark date. The spare parts circulation and maintenance timeline reverse calculation submodule calls the mold scrapping baseline date, extracts the standard quotas for physical circulation and safety stock consumption of spare parts in the stamping workshop, sets intervention action nodes for mold disassembly and physical maintenance, classifies the response level based on the real-time deviation between the node and the current date, and generates a physical intervention timeline for maintenance. The full-life structural loss matrix grid module, based on the maintenance physical intervention time axis, maps the original cumulative number of punches, cavity polishing records and repair geometric increments associated with the mold coding, constructs a multi-dimensional metal loss evolution matrix, embeds the maintenance intervention time axis at the end of the matrix, and integrates the frequency distribution characteristics of mechanical faults for probability weighting, generating a mold full life cycle maintenance intervention management table.

10. The mold life prediction system based on data analysis according to claim 9, characterized in that, The intervention action nodes for mold disassembly and physical maintenance refer to the procurement lead time for vulnerable parts in the standard quota for physical circulation and safety stock consumption of spare parts in the stamping workshop. Before the mold scrapping benchmark date, the intervention action nodes for mold disassembly and physical maintenance are obtained by reverse time offset calculation based on the procurement lead time and the mold downtime maintenance cycle. The multidimensional metal loss evolution matrix refers to the extraction of the original cumulative number of strokes associated with the mold code as the time dimension coordinate, and the subtraction depth value in the cavity polishing record and the additive height value in the geometric increment of the repair welding as the spatial dimension variables, which are arranged in time sequence to generate a numerical matrix describing the changes in the mold surface morphology, thus obtaining the multidimensional metal loss evolution matrix.

Images