A computer-aided design-based injection mold structure optimization method and system

Abstract

The application relates to the technical field of computer-aided design, and provides a computer-aided design-based injection mold structure optimization method and system, which comprises the following steps: performing preliminary structure stress analysis on a geometric feature region in a mold structure; in the preliminary structure stress analysis process, the calculation progress and result stability of each region of the mold structure are monitored, so as to give each region a quantitative reliability score, and regions with a reliability score lower than a preset threshold value are marked as high-uncertainty regions; for the high-uncertainty regions, local excitation is injected to stimulate the dynamic response of the regions, and the dynamic response is monitored to diagnose the sensitivity of the regions to micro-vibration fatigue; according to the diagnosis result, an optimization suggestion for the region is generated in combination with a structured design guiding principle and partial convergence simulation data of the region; and for regions not marked as high-uncertainty regions, an optimization suggestion is generated according to simulation analysis results. The application improves the reliability and optimization efficiency of mold design.

Description

Technical Field

[0001] This application relates to the field of computer-aided design technology, and more specifically, to a method and system for optimizing the structure of injection molds based on computer-aided design. Background Technology

[0002] Current injection mold development, especially for structures with complex shapes and drastic wall thickness variations, relies excessively on repeated testing with physical prototypes. This not only consumes significant time and money but also makes it difficult to accurately predict the stress conditions in actual use. Even with the introduction of computer-aided design tools, automated mesh generation programs often struggle to handle complex areas such as abrupt transitions between thin and thick walls, easily producing poorly shaped mesh elements. These mesh defects severely affect the computational stability of subsequent structural stress analysis, leading to calculation interruptions. This prevents engineers from obtaining reliable data to evaluate mold performance, forcing them to rely on experience for adjustments, thus hindering the development process and preventing effective verification of the design's safety and reliability. Summary of the Invention

[0003] This application provides a method and system for optimizing the structure of injection molds based on computer-aided design, in order to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above objectives, this application adopts the following technical solution: In a first aspect, this application discloses a method for optimizing the structure of injection molds based on computer-aided design, including: Obtain the structural information of the injection mold, identify the geometric feature areas in the mold structure, and perform preliminary structural stress analysis on the geometric feature areas; During the initial structural stress analysis, the calculation progress and result stability of each region of the mold structure are monitored to assign a quantitative reliability score to each region, and regions with reliability scores below a preset threshold are marked as high uncertainty regions. For regions with high uncertainty, local excitation is injected to stimulate the dynamic response of the region, and the energy dissipation mode, stress wave propagation characteristics, and high-frequency stress fluctuation spectrum of the dynamic response are monitored to diagnose the sensitivity of the region to micro-vibration fatigue. Based on the diagnostic results, and combined with the structured design guidelines and partial convergence simulation data of the region, optimization suggestions are generated for the region. For regions not marked as high uncertainty areas, optimization suggestions are generated based on their simulation analysis results.

[0005] Secondly, this application also discloses a computer-aided design-based injection mold structure optimization system, comprising: The information acquisition module is used to acquire the structural information of the injection mold, identify the geometric feature areas in the mold structure, and perform preliminary structural stress analysis on the geometric feature areas. The reliability assessment module is used to monitor the calculation progress and result stability of each region of the mold structure during the initial structural stress analysis process, so as to assign a quantitative reliability score to each region and mark the region with a reliability score below a preset threshold as a high uncertainty region. The optimization suggestion generation module is used to inject local excitation into high uncertainty regions to stimulate the dynamic response of the region, and monitor the energy dissipation mode, stress wave propagation characteristics, and high-frequency stress fluctuation spectrum of the dynamic response to diagnose the sensitivity of the region to micro-vibration fatigue. Based on the diagnostic results, combined with the structured design guidelines and the partial convergence simulation data of the region, optimization suggestions are generated for the region. For regions not marked as high uncertainty regions, optimization suggestions are generated based on their simulation analysis results. Compared with the prior art, this application has at least the following beneficial effects: This application effectively solves the problems of unstable calculations, lack of reliable data support, and inability to effectively assess micro-fatigue risks in the optimization of complex injection mold structures in the prior art by introducing a reliability scoring mechanism, micro-vibration fatigue diagnosis, and targeted optimization suggestions. Attached Figure Description

[0006] Figure 1 A flowchart illustrating a computer-aided design-based method for optimizing the structure of injection molds provided in this application; Figure 2 This application provides a schematic diagram of a computer-aided design-based injection mold structure optimization system. Detailed Implementation

[0007] The technical solutions 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.

[0008] like Figure 1 As shown in the embodiments of this application, a method for optimizing the structure of injection molds based on computer-aided design is proposed, including: Obtain the structural information of the injection mold, identify the geometric feature areas in the mold structure, and perform preliminary structural stress analysis on the geometric feature areas; During the initial structural stress analysis, the calculation progress and result stability of each region of the mold structure are monitored to assign a quantitative reliability score to each region, and regions with reliability scores below a preset threshold are marked as high uncertainty regions. For regions with high uncertainty, local excitation is injected to stimulate the dynamic response of the region, and the energy dissipation mode, stress wave propagation characteristics, and high-frequency stress fluctuation spectrum of the dynamic response are monitored to diagnose the sensitivity of the region to micro-vibration fatigue. Based on the diagnostic results, and combined with the structured design guidelines and partial convergence simulation data for the region, optimization suggestions are generated for the region. For regions not marked as high uncertainty areas, optimization suggestions are generated based on their simulation analysis results.

[0009] The structural information of an injection mold can include its three-dimensional geometric model, material properties, boundary conditions, and expected load conditions. This information forms the basis for structural analysis. Geometric feature regions refer to areas in the mold structure with specific geometric shapes or functional characteristics, such as runners, reinforcing ribs, cooling holes, and cavity walls. These regions are often the focus of structural stress analysis. Preliminary structural stress analysis refers to the initial assessment of the mold's stress, strain, deformation, and other mechanical behaviors under typical operating conditions using simulation tools such as finite element analysis (FEA) during the early stages of mold design.

[0010] Computational progress and result stability refer to the real-time monitoring of the convergence of computational iterations, the changing trends of key parameters, and the fluctuation range of results during structural stress analysis. For example, it can monitor the changes in the number of iterations, residual norm, and displacement or stress values ​​of key nodes with the iteration step size. Reliability scoring is a quantitative assessment of the credibility of computational results; a higher value indicates greater reliability. This score can comprehensively consider factors such as computational convergence, mesh quality, and the accuracy of the material model. A preset threshold is a standard used to judge the reliability of computational results; regions below this threshold are considered to have high uncertainty. High uncertainty regions refer to those areas where the preliminary structural stress analysis results are unstable due to factors such as geometric complexity, poor mesh quality, complex load conditions, or material nonlinearity.

[0011] Local excitation refers to external forces applied to a region of high uncertainty that can elicit a dynamic response within it, such as virtual heat flow pulses, virtual pressure waves, or mechanical vibrations. Dynamic response refers to the transient or steady-state vibrations, deformations, and stress changes produced in a region of high uncertainty under local excitation. Energy dissipation modes describe how energy is absorbed and converted into heat or other forms of energy during vibration, reflecting the damping characteristics of the material and the energy absorption capacity of the structure. Stress wave propagation characteristics refer to the speed, attenuation, and reflection of stress waves propagating in a material, revealing internal defects and discontinuities. High-frequency stress wave spectra are obtained through spectral analysis of stress response signals, revealing the characteristic frequencies and amplitudes of microscopic vibration fatigue. Microscopic vibration fatigue sensitivity refers to the tendency of a material or structure to develop fatigue damage under small vibration loads.

[0012] Structured design guidelines refer to a set of proven design rules, best practices, and engineering experience used to guide the optimized design of mold structures, such as avoiding stress concentration, optimizing material distribution, and improving stiffness. Partially convergent simulation data refers to simulation results from preliminary structural stress analyses in high-uncertainty regions, which, although not fully converged, still provide some trend or local information. Optimization recommendations are improvement schemes proposed for the mold structure, aiming to improve the mold's performance, reliability, and service life.

[0013] The method for optimizing injection mold structure based on computer-aided design provided in this application will be described in further detail below: There are several ways to obtain structural information about injection molds. For example, geometric information can be obtained by importing 3D model files of the mold generated by computer-aided design software (such as STEP, IGES, Parasolid, etc.). Simultaneously, mechanical property parameters of the materials used in the mold, such as elastic modulus, Poisson's ratio, and yield strength, can be extracted from a material database. Furthermore, the types and magnitudes of loads the mold may bear in actual production, as well as environmental parameters such as temperature and humidity, can be manually entered or selected from a preset working condition library. Identifying geometric feature regions in the mold structure can be done automatically using geometric processing algorithms. For example, based on curvature analysis, features such as sharp corners, rounded corners, and holes can be identified; or based on topology, flow channels and reinforcing ribs can be identified. Alternatively, engineers can manually select and mark areas of interest on the 3D model through interactive methods. Preliminary structural stress analysis of the geometric feature regions can be performed using finite element analysis software. For example, the mold model can be discretized into a finite element mesh, boundary conditions and loads can be applied, and then the equilibrium equations can be solved to obtain the stress, strain, and deformation distribution of each region.

[0014] During the initial structural stress analysis, monitoring the computational progress and result stability of each region of the mold structure allows for real-time tracking of the finite element solver's iterative convergence curve, observing whether the residual norm continuously decreases and tends to stabilize. Simultaneously, several monitoring points in key areas of the mold can be selected to record stress or displacement values ​​at different iteration steps, and their fluctuation range can be analyzed. Assigning a quantified reliability score to each region involves a weighted average of factors such as computational convergence speed, mesh quality indicators (e.g., Jacobian ratio, aspect ratio), the rationality of material model selection, and the accuracy of boundary condition application, resulting in a comprehensive score. For example, regions with fast convergence speed, high mesh quality, and reasonable material model and boundary condition settings will have higher reliability scores. Regions with reliability scores below a preset threshold are marked as high-uncertainty regions. This can be achieved by setting a value, such as 0.7; any region with a reliability score below 0.7 will be automatically marked as a high-uncertainty region by the system and subjected to subsequent special processing.

[0015] For regions with high uncertainty, injecting local excitation to stimulate the region's dynamic response can be achieved in various ways. For example, a high-frequency vibrator can be simulated in a simulation environment to apply a periodic force load of a specific frequency and amplitude to the region with high uncertainty. Alternatively, a local heating source can be simulated to apply a transient heat flux pulse to the region. Monitoring the energy dissipation mode, stress wave propagation characteristics, and high-frequency stress fluctuation spectrum of the dynamic response to diagnose the region's sensitivity to microscopic vibration fatigue can be achieved through post-processing analysis of the simulation results. For example, energy dissipation can be assessed by calculating the area of ​​the local stress-strain hysteresis loop, stress wave propagation characteristics can be analyzed by tracing the propagation path and attenuation of stress waves within the region, and the high-frequency stress fluctuation spectrum can be obtained by performing a Fast Fourier Transform (FFT) on the local stress response signal.

[0016] Based on the diagnostic results, combined with structured design guidelines and partially convergent simulation data for the region, optimization suggestions are generated for high-uncertainty areas. These suggestions can be based on the diagnosed microscopic vibration fatigue sensitivity; for example, if the diagnostic results show that a certain area is sensitive to high-frequency vibration fatigue, optimization suggestions might include increasing the wall thickness of that area, changing the material, introducing damping structures, or optimizing the geometry to disperse stress. Structured design guidelines provide general design rules, such as avoiding sharp corners, using rounded transitions, and optimizing the layout of stiffeners. While partially convergent simulation data is not entirely reliable, it can provide some trend information, such as the approximate location of potential stress concentrations, which can serve as a reference for optimization suggestions. For areas not marked as high-uncertainty areas, optimization suggestions are generated based on their simulation analysis results. Since the simulation results for these areas are reliable, corresponding optimization suggestions can be directly generated based on their stress, strain, and deformation results, combined with design specifications and performance requirements. Examples include adjusting dimensions to meet strength requirements or optimizing the shape to reduce weight.

[0017] This application assigns a quantitative reliability score to each region by monitoring computational progress and result stability in real time, and identifies regions with high uncertainty. This step enables the system to accurately pinpoint the problem.

[0018] Furthermore, for these highly uncertain regions, this application no longer relies on traditional macroscopic mechanical analysis. Instead, it injects local excitation to stimulate the dynamic response of the region and monitors energy dissipation modes, stress wave propagation characteristics, and high-frequency stress fluctuation spectra. This microscopic diagnostic method can reveal the region's sensitivity to microscopic vibration fatigue more deeply, thus providing a more accurate basis for optimization.

[0019] In some embodiments, the step of monitoring the calculation progress and result stability of each region of the mold structure during the preliminary structural stress analysis process, assigning a quantified reliability score to each region, and marking regions with reliability scores below a preset threshold as high uncertainty regions further includes: During the preliminary structural stress analysis, characteristic parameters of the current operating environment of the mold are extracted; Obtain the runtime environment parameters of historical design cases; Based on the differences between the characteristic parameters of the current operating environment and the operating environment parameters of historical design cases, the environmental condition difference index is calculated, and the reference weight of historical design cases when assigning quantitative reliability scores to each region is adjusted according to the environmental condition difference index. Based on the monitored computational progress and result stability of each region, and combined with the adjusted reference weights, a quantitative reliability score is assigned to the computational results of each region, and regions with reliability scores below a preset threshold are marked as high uncertainty regions.

[0020] Specifically, when evaluating the reliability scores of various regions of the mold structure, it is first necessary to extract the characteristic parameters of the current operating environment of the mold. These characteristic parameters may include, but are not limited to, ambient temperature, humidity, pressure, type of cooling medium, batch differences in injection molding materials, and injection molding process parameters (such as injection speed, holding pressure, and cooling time). Simultaneously, it is also necessary to obtain the operating environment parameters of historical design cases. This historical data comes from feedback on completed mold design, manufacturing, and actual operation, and includes information such as the mold's performance under different environmental conditions, failure modes, and deviations between simulation results and actual conditions.

[0021] The environmental condition difference index is calculated by comparing the characteristic parameters of the current mold operating environment with those of historical design cases. For example, multidimensional vector distance, weighted Euclidean distance, or machine learning-based similarity assessment methods can be used to quantify this difference. A higher environmental condition difference index indicates a lower similarity between the current operating conditions and historical cases, and its reference value may need to be adjusted accordingly. Based on this difference index, the reference weight of historical design cases in evaluating the reliability of calculation results can be adjusted. For example, when the environmental condition difference index is high, the weight of historical cases with lower similarity can be reduced, while when the difference index is low, their weight can be increased to ensure the effective use of historical experience.

[0022] After comprehensively considering the computational progress and result stability of each monitored region, and combining the adjusted reference weights mentioned above, a quantitative reliability score was assigned to the computational results of each region. Computational progress reflects the convergence of the simulation, while result stability reflects the volatility of the simulation results. By combining these factors with the environmental condition difference index and the adjusted historical case reference weights, a more comprehensive and accurate reliability score can be obtained. Finally, regions with reliability scores below a preset threshold are marked as high-uncertainty regions for subsequent targeted optimization.

[0023] This application incorporates consideration of current operating environment characteristic parameters and historical design case operating environment parameters, and calculates an environmental condition difference index to adjust the reference weights of historical cases, making the reliability score generation process more intelligent and contextualized. Therefore, when evaluating the reliability of various regions of the mold structure, it not only relies on the current simulation calculation state but also fully utilizes historical experience and actual operating environment information, thereby more accurately identifying regions that may have potential risks or uncertainties under complex operating conditions. This method effectively overcomes the limitations of solely relying on simulation results for evaluation, improving the accuracy and reliability of identifying high-uncertainty regions.

[0024] Suppose an injection mold is undergoing structural optimization. Preliminary stress analysis shows that the calculated stability of certain areas is only moderate. Based solely on the current calculated stability, these areas might be assigned a medium reliability score. However, the approach proposed in this application first extracts the operating environment parameters of the current mold. For example, the mold will operate in a high-temperature, high-humidity environment and use a novel composite material for injection molding. Simultaneously, the system acquires operating environment parameters from a large number of historical design cases. In some of these cases, molds using traditional materials exhibited fatigue cracks under similar high-temperature, high-humidity conditions. By comparing the current environmental parameters with those of historical cases, an environmental condition difference index is calculated. Due to the similarity between the current environment and some historical failure cases, and the use of a novel material, the system adjusts the reference weights of historical cases based on the difference index, assigning higher weights to cases that failed in similar environments. Ultimately, after comprehensively considering the computational progress, result stability, and the adjusted reference weights of historical cases, even if the calculated stability of certain areas is acceptable, their reliability scores may be significantly reduced due to the similarity of their environment to historical failure cases, thus being more accurately labeled as high-uncertainty areas. This allows optimization suggestions to focus on these potential weaknesses earlier, thereby effectively improving the overall reliability of the mold.

[0025] In some embodiments, the steps described above for injecting local excitation to stimulate the dynamic response of a high-uncertainty region, and monitoring the energy dissipation mode, stress wave propagation characteristics, and high-frequency stress fluctuation spectrum of the dynamic response to diagnose the sensitivity of the region to micro-vibration fatigue include: Based on the characteristic parameters of the current operating environment of the mold and the material properties of the high uncertainty region, the characteristic frequency range and stress reversal amplitude of the micro-vibration fatigue that may be induced in the high uncertainty region are predicted. Based on the predicted characteristic frequency range, the duration of the virtual heat flow pulse is adjusted so that the spectrum of the virtual heat flow pulse matches the characteristic frequency range. Based on the predicted stress reversal amplitude, the waveform of the virtual pressure wave is adjusted so that the virtual pressure wave produces an enhancement effect at the stress reversal amplitude. Adjusted virtual heat flow pulses and virtual pressure waves are injected into regions of high uncertainty as local excitation, and the local stress response in these regions is monitored in real time. Based on the frequency components and amplitude of the local stress response, the parameters of the local excitation are iteratively corrected until the local excitation can excite the micro vibration mode in the high uncertainty region. Monitoring the energy dissipation mode, stress wave propagation characteristics, and high-frequency stress fluctuation spectrum of the dynamic response during iterative correction is used to diagnose the region's sensitivity to micro-vibration fatigue.

[0026] Specifically, before applying local excitation to a region with high uncertainty, it is necessary to predict the characteristic frequency range and stress reversal amplitude of the micro-vibration fatigue that may be induced in the region, based on the characteristic parameters of the current mold's operating environment, such as temperature, pressure cycles, and cooling rate, as well as the material properties of the region with high uncertainty, such as elastic modulus, Poisson's ratio, and fatigue limit. This prediction process can be based on methods such as finite element analysis, machine learning models, or empirical formulas, aiming to provide accurate targeting information for subsequent local excitation. The duration of the virtual heat flow pulse is adjusted to match its spectrum with the predicted characteristic frequency range. This means that by controlling the frequency components of the thermal excitation, it can resonate with the microstructure within the material, thereby more effectively stimulating the fatigue-related thermomechanical response. For example, this can be achieved by adjusting the width or repetition frequency of the virtual laser pulse. Simultaneously, the waveform of the virtual pressure wave is adjusted to produce an enhancement effect at the predicted stress reversal amplitude. This ensures that the mechanical excitation can simulate or amplify the stress levels that may lead to fatigue in actual operation, thus more realistically reflecting the region's sensitivity to micro-vibration fatigue. For example, this can be achieved by adjusting the amplitude or modulating the waveform of the virtual ultrasonic wave.

[0027] Subsequently, adjusted virtual heat flux pulses and virtual pressure waves are injected as local excitations into the high-uncertainty region. During the injection process, the local stress response in the high-uncertainty region is monitored in real time. This monitoring can be performed using virtual sensors or simulation analysis tools to obtain dynamic behavior data of the region under excitation. Further, the parameters of the local excitation are iteratively corrected based on the frequency components and amplitudes of the real-time monitored local stress response. This iterative process aims to gradually optimize the excitation parameters until the local excitation can stably and effectively stimulate the micro-vibration modes in the high-uncertainty region. This means that through feedback control, the excitation is ensured to achieve optimal matching with the inherent vibration characteristics of the region, thereby accurately simulating fatigue-inducing conditions. During the iterative correction process, the energy dissipation mode, stress wave propagation characteristics, and high-frequency stress fluctuation spectrum of the dynamic response are continuously monitored. The energy dissipation mode can reveal the damping and plastic deformation behavior within the material; the stress wave propagation characteristics can reflect the damage accumulation and crack initiation of the material; and the high-frequency stress fluctuation spectrum can capture early signals of micro-crack propagation or material structural changes. By comprehensively analyzing these monitoring data, the sensitivity of the region to micro-vibration fatigue can be accurately diagnosed.

[0028] This application overcomes the potential blindness of local excitation in the basic scheme by introducing a prediction mechanism for the characteristic frequency range and stress reversal amplitude of micro-vibration fatigue, and based on this, making targeted adjustments to virtual heat flow pulses and virtual pressure waves. Because the excitation parameters can accurately match the potential fatigue characteristics of high-uncertainty regions, the injected local excitation can more effectively stimulate dynamic responses directly related to micro-vibration fatigue. Furthermore, by monitoring the local stress response in real time and performing iterative corrections, it ensures that the excitation can stably stimulate the target micro-vibration mode, avoiding diagnostic biases caused by improper excitation. This refined excitation and response analysis mechanism makes the monitoring of energy dissipation modes, stress wave propagation characteristics, and high-frequency stress fluctuation spectra more targeted and accurate, thereby revealing the fatigue damage mechanism of materials at the micro-level more deeply.

[0029] Suppose the gate region of an injection mold is identified as a high-uncertainty area. First, by analyzing the injection pressure cycle frequency, cooling water temperature fluctuations, and fatigue performance data of the mold steel used in the gate region during actual production, the characteristic frequency range of micro-vibration fatigue that may be induced in this region is predicted to be 500Hz to 1500Hz, with a stress reversal amplitude potentially reaching 200MPa. Based on this prediction, the duration of the virtual heat flow pulse is adjusted, for example, by simulating a pulsed laser, setting its pulse width and repetition frequency so that its spectral energy is mainly concentrated in the 500Hz to 1500Hz range. Simultaneously, the waveform of the virtual pressure wave is designed to produce a resonant enhancement effect when the simulated stress reaches 200MPa, for example, by simulating a high-frequency piezoelectric exciter and adjusting its driving voltage and frequency. Subsequently, these adjusted virtual excitations are injected into the simulation model of the gate region, and the local stress response in this region is monitored in real time. If the initial response fails to clearly demonstrate the expected micro-vibration mode, the duration of the virtual laser pulse or the driving parameters of the virtual piezoelectric actuator are iteratively corrected based on the monitored frequency and amplitude data of the stress response, for example, by analyzing its spectrum through Fourier transform. For instance, if the response peak deviates from the predicted characteristic frequency, the excitation frequency is fine-tuned; if the response amplitude is insufficient to induce a fatigue mode, the excitation intensity is increased. This iterative process continues until the gate region in the simulation model exhibits stable micro-vibrations consistent with the predicted fatigue mode. During this iterative correction process, the energy dissipation rate of the gate region, the propagation path and attenuation of stress waves within the material, and specific harmonic components in the high-frequency stress wave spectrum are continuously monitored. For example, if a significant increase in energy dissipation rate is observed at a specific frequency, or if abnormal reflection or scattering of stress waves occurs at certain microstructural interfaces, and characteristic frequencies associated with microcrack initiation appear in the high-frequency stress wave spectrum, it can be diagnosed that the gate region is highly sensitive to micro-vibration fatigue.

[0030] In some embodiments, the step of iteratively correcting the parameters of the local excitation based on the frequency components and amplitude of the local stress response until the local excitation can excite micro-vibration modes in the high-uncertainty region includes: Perform a frequency scan by gradually adjusting the frequency of the local excitation within a preset frequency range and monitoring the local stress response in high uncertainty regions in real time to identify all peak frequencies and their corresponding amplitudes that have significant local stress responses. Based on the identified peak frequencies and their corresponding amplitudes, the characteristic parameters of the current operating environment of the mold, and the material properties of the high uncertainty region, the correlation between the vibration mode corresponding to each peak frequency and structural fatigue damage is evaluated to determine the microscopic vibration modes and their characteristic frequencies that play a dominant role in structural fatigue. Adjust the frequency of the local excitation to lock it at the characteristic frequency of the dominant micro-vibration mode; Adjust the amplitude of the local excitation to produce a resonant enhancement effect at the characteristic frequency.

[0031] Specifically, frequency scanning refers to gradually changing the frequency of the injected local excitation within a preset frequency range, such as from low to high frequencies, with a certain step size or an adaptive step size. During this process, sensors deployed in high-uncertainty regions monitor the local stress response in real time, including its amplitude and phase information. By performing spectral analysis on the monitored response data, all frequency points that cause significant local stress responses during the frequency scanning process can be identified, i.e., peak frequencies, along with the corresponding stress response amplitudes at these peak frequencies. These peak frequencies represent the inherent vibration modes that may exist in this region.

[0032] Based on the identified peak frequencies and their corresponding amplitudes, characteristic parameters of the current operating environment of the mold (e.g., temperature, humidity, type of external load), and material properties in high-uncertainty regions (e.g., elastic modulus, Poisson's ratio, fatigue limit), the correlation between the vibration modes corresponding to each peak frequency and structural fatigue damage is assessed. This assessment can be performed using a pre-established fatigue damage model, finite element analysis, or machine learning algorithms. The aim is to distinguish which vibration modes, among numerous others, have a dominant effect on structural fatigue under actual operating conditions. For example, some vibration modes may have high stress amplitudes but their frequencies do not match the fatigue-sensitive frequencies of the material; while other modes may have relatively low stress amplitudes but their frequencies closely match the fatigue-sensitive frequencies of the material, thus making them more prone to fatigue damage. Through this assessment, the microscopic vibration modes and their characteristic frequencies that have a dominant effect on structural fatigue can be accurately determined.

[0033] The frequency of the local excitation is adjusted and locked to the characteristic frequency of the identified dominant micro-vibration mode. This means precisely matching the excitation frequency to the natural frequency most likely to cause fatigue damage. Subsequently, the amplitude of the local excitation is adjusted to produce a resonance enhancement effect at this characteristic frequency. The resonance enhancement effect refers to the significant increase in the vibration amplitude of the structure when the excitation frequency matches the natural frequency of the structure, thereby more effectively exciting and amplifying the micro-vibration mode, making it more apparent in simulations or tests, and facilitating subsequent monitoring and diagnosis.

[0034] This application addresses the potential blindness and inefficiency of iterative correction by introducing a frequency scanning and fatigue correlation assessment mechanism. First, by performing a frequency scan, the response characteristics of high-uncertainty regions under different excitation frequencies are systematically detected, thereby comprehensively identifying all potential micro-vibration modes and their characteristic frequencies. This avoids the risk of missing critical vibration modes. Second, by combining operating environment parameters and material properties, the correlation between the identified vibration modes and structural fatigue damage is assessed, enabling the system to intelligently screen out micro-vibration modes that have a dominant effect on structural fatigue. This assessment mechanism ensures that subsequent excitation adjustments are targeted at the most critical fatigue sources, rather than just arbitrary vibration modes. Finally, by precisely locking the local excitation frequency to the characteristic frequency of the dominant micro-vibration mode and adjusting the amplitude to produce a resonance enhancement effect, this scheme can most effectively excite and amplify the target micro-vibration mode with minimal energy input. This provides clear and amplified signals for subsequent monitoring of energy dissipation modes, stress wave propagation characteristics, and high-frequency stress wave spectra, greatly improving the accuracy and efficiency of diagnosing the region's sensitivity to micro-vibration fatigue.

[0035] Suppose we are analyzing a high-uncertainty region of an injection mold, which may pose a potential risk of micro-vibration fatigue due to its complex geometry or material inhomogeneity. First, the system performs a frequency scan, injecting virtual heat flux pulses and virtual pressure waves as local excitations in 50Hz increments within a frequency range of 100Hz to 10kHz. During this process, a micro-strain sensor array monitors the local stress response in real time within this region. By performing Fast Fourier Transform (FFT) analysis on the acquired stress response data, significant stress response peaks are identified, for example, at 500Hz, 1200Hz, and 3500Hz, and their corresponding amplitudes are recorded.

[0036] Subsequently, combining the typical operating temperature and pressure cycle characteristics of the mold during injection molding production, as well as the specific alloy material properties (e.g., its fatigue SN curve data) in this high-uncertainty region, a fatigue correlation assessment is performed on the vibration modes corresponding to these three peak frequencies. For example, using a pre-defined fatigue damage accumulation model, the influence weights of the 500Hz, 1200Hz, and 3500Hz modes on the material's fatigue life are calculated under the current operating environment. The assessment results may show that although the 3500Hz mode has a higher stress amplitude, its frequency does not match the material's fatigue-sensitive frequency well; while the 1200Hz mode, although having a moderate stress amplitude, has a frequency that closely matches the material's fatigue-sensitive frequency and is prone to inducing fatigue crack initiation under the current operating environment. Therefore, the system determines that the 1200Hz mode is the dominant micro-vibration mode for structural fatigue, with a characteristic frequency of 1200Hz.

[0037] Finally, the system precisely adjusted and locked the frequency of the local excitation at 1200Hz. Simultaneously, by gradually increasing the amplitudes of the virtual heat flux pulse and virtual pressure wave, a significant resonance enhancement effect was detected at 1200Hz in this high-uncertainty region, meaning the local stress response amplitude reached its maximum and stabilized. This method ensured the effective excitation of the most critical fatigue-related micro-vibration modes, laying the foundation for subsequent precise monitoring of energy dissipation modes, stress wave propagation characteristics, and high-frequency stress wave spectra, as well as fatigue sensitivity diagnosis.

[0038] In some embodiments, the step of iteratively correcting the parameters of the local excitation based on the frequency components and amplitude of the local stress response until the local excitation can excite micro-vibration modes in the high-uncertainty region includes: Based on the geometric characteristics, material properties, and current operating environment parameters of the high uncertainty region, the nonlinear response characteristics and coupling effects of the high uncertainty region under different excitation parameters are predicted. Based on the predicted nonlinear response characteristics and coupling effects, a response surface is constructed, which describes the relationship between local excitation parameters and the degree of excitation of micro-vibration modes in the high uncertainty region. During the iterative correction process, the search direction and step size of the local excitation parameters are dynamically adjusted on the response surface based on the real-time monitored local stress response. A convergence judgment mechanism is introduced, which determines whether the target micro-vibration mode has been stably excited based on the stability of the local stress response, the purity of the frequency components, and the convergence trend of the amplitude, and stops iterating when the convergence condition is met.

[0039] Specifically, predicting the nonlinear response characteristics and coupling effects of a highly uncertain region under different excitation parameters, based on its geometric features, material properties, and current operating environment parameters, involves using numerical simulation, finite element analysis, or machine learning models to estimate the mechanical behavior of the region under different local excitations (such as the frequency, amplitude, and duration of virtual heat flow pulses and virtual pressure waves) before actual excitation. Nonlinear response characteristics may include plastic deformation, geometric nonlinearity (e.g., large deformation), and contact nonlinearity, while coupling effects may refer to the interactions between multiple physical fields, such as thermo-mechanical coupling and fluid-structure interaction. Predicting these characteristics helps to more comprehensively understand the complex behavior of the region and provides a theoretical basis for subsequent adjustment of excitation parameters.

[0040] Based on the predicted nonlinear response characteristics and coupling effects, a response surface is constructed. This surface describes the relationship between local excitation parameters and the excitation degree of micro-vibration modes in the high-uncertainty region. It can be understood as establishing a multidimensional function model, which takes local excitation parameters (e.g., excitation frequency, amplitude) as input and the excitation degree of the target micro-vibration modes (e.g., stress response amplitude, energy density at a specific frequency) as output. The response surface can be fitted using experimental design methods combined with numerical simulation data, such as polynomial regression, radial basis functions, or neural networks. The aim is to provide a continuous, differentiable functional representation so that the optimal excitation parameters can be efficiently searched during iterative correction.

[0041] During the iterative correction process, the search direction and step size of the local excitation parameters are dynamically adjusted on the response surface based on the real-time monitored local stress response. This refers to using the gradient information of the response surface to guide the update of the excitation parameters. When local stress response data is monitored in real time, it is compared with the response surface, and the local gradient of the response surface at the current excitation parameter point is calculated. This gradient indicates the direction of the fastest increase in excitation intensity. Simultaneously, based on the local curvature information of the response surface, the search step size can be adaptively adjusted to avoid overshoot or slow convergence, thereby achieving more efficient and accurate parameter optimization.

[0042] A convergence judgment mechanism is introduced, which determines whether the target micro-vibration mode has been stably excited based on the stability of the local stress response, the purity of the frequency components, and the convergence trend of the amplitude. It stops iteration when the convergence conditions are met. Specifically, stability refers to whether the fluctuation range of the local stress response is less than a preset threshold over a period of time; the purity of the frequency components refers to whether the characteristic frequency of the target micro-vibration mode dominates in the spectral analysis, and whether the amplitudes of other irrelevant frequency components are sufficiently small; the convergence trend of the amplitude refers to whether the stress response amplitude at the target frequency has reached stability or its maximum value. When these conditions are simultaneously met, the target micro-vibration mode is considered to have been stably excited, thus stopping the iteration and avoiding unnecessary computational resource consumption.

[0043] This application transforms the complex problem of excitation parameter optimization into an optimization search problem on the response surface by introducing predictions of nonlinear response characteristics and coupling effects and constructing a response surface. Because the response surface provides a quantitative relationship between the excitation parameters and the degree of excitation of micro-vibration modes, the search direction and step size of the local excitation parameters can be dynamically and directionally adjusted during iterative correction based on the gradient and curvature information of the response surface, according to real-time monitored local stress responses. This model-guided optimization strategy significantly improves the efficiency and accuracy of parameter optimization compared to blind trial and error or fixed-step iteration. Furthermore, the introduction of a convergence judgment mechanism ensures that iteration stops only when the target micro-vibration mode is stably and effectively excited, avoiding premature stopping or invalid iterations, thus guaranteeing the reliability of the diagnostic results.

[0044] Suppose we need to diagnose the micro-vibration fatigue sensitivity of a high-uncertainty region within an injection mold. First, based on the region's geometric characteristics (e.g., acute angles, thin-walled structure), material properties (e.g., viscoelasticity of polymer materials), and current operating environment parameters (e.g., injection temperature, pressure cycle frequency), multiphysics coupling simulation software is used to predict the nonlinear stress-strain response and thermo-mechanical coupling effects that may occur in this region under different virtual heat flux pulses and virtual pressure wave parameters (e.g., frequency, amplitude). For example, it is predicted that under specific high-frequency excitation, the material may exhibit significant viscoelastic damping and localized plastic deformation.

[0045] Next, a response surface is constructed based on these predicted data. This response surface can be a three-dimensional graph, where two axes represent the frequency of the virtual heat flux pulse and the amplitude of the virtual pressure wave, and the third axis represents the excitation degree of the target micro-vibration mode (e.g., the local stress peak at a specific frequency). This surface is built by fitting a large number of simulation data points, thus visually displaying the influence of different excitation parameter combinations on the excitation effect of micro-vibration modes.

[0046] During the iterative correction process, the system injects a set of initial local excitation parameters and monitors the local stress response in high-uncertainty regions in real time. For example, data is acquired using a micro-strain sensor array. If the monitored stress response fails to reach the stable excitation state of the target micro-vibration mode, the system calculates the local gradient and curvature at the current parameter point on a pre-constructed response surface based on the current local stress response data. For instance, if adjusting the frequency and amplitude in a certain direction is found to improve the stress response more quickly, the excitation parameters are adjusted along that gradient direction. Simultaneously, based on curvature information, if the response surface changes gently in this region, the step size is appropriately increased; if the change is drastic, the step size is decreased to avoid overshoot.

[0047] Finally, the introduced convergence judgment mechanism continuously evaluates the local stress response. For example, if the amplitude fluctuation of the stress response at the target frequency is less than 1% over 10 consecutive sampling periods, and the spectral analysis shows that the energy proportion of the target characteristic frequency exceeds 90%, while the amplitude has reached more than 95% of its maximum value and tends to stabilize, then it is determined that the target micro-vibration mode has been stably excited, and the iteration stops. In this way, it ensures that the target micro-vibration mode can be efficiently, accurately, and stably excited under complex nonlinear conditions, providing a solid foundation for subsequent fatigue sensitivity diagnosis.

[0048] In some embodiments, the step of dynamically adjusting the search direction and step size of the local excitation parameters on the response surface based on the real-time monitored local stress response during the iterative correction process includes: Obtain local stress response data under the current local excitation parameters; Calculate the local gradient of the response surface at the current parameter point based on the local stress response data; The local gradient is corrected based on the local material properties and geometry of the high uncertainty region to determine the updated search direction; Adjust the search step size based on the local curvature information of the response surface at the current parameter point; Based on the updated search direction and adjusted search step size, determine the local excitation parameters for the next iteration.

[0049] Specifically, obtaining local stress response data under the current local excitation parameters means that, during each iteration of adjusting the local excitation parameters, the stress, strain, vibration, and other response data of the high-uncertainty region under the current excitation parameters are collected in real time through numerical simulation or in combination with actual test data. This data forms the basis for evaluating the current excitation effect and guiding subsequent parameter adjustments.

[0050] Calculating the local gradient of the response surface at the current parameter point based on local stress response data can be understood as analyzing the relationship between the current local stress response data and the local excitation parameters using numerical methods (such as the finite difference method, adjoint method, etc.) or based on machine learning models. This allows us to determine the trend of the response surface at that parameter point and the steepest direction of ascent or descent. The local gradient indicates how fine-tuning the excitation parameters near the current parameter point can most effectively change the excitation level of the micro-vibration modes.

[0051] The local gradient is corrected based on the local material properties and geometry of regions with high uncertainty to determine the updated search direction. The aim is to make the search direction more consistent with actual physical processes and structural characteristics. For example, some materials may exhibit anisotropic or nonlinear behavior under specific stress states, or complex geometries may lead to stress concentration effects. By incorporating these physical constraints and properties into the gradient correction, the search direction can be avoided from pointing to physically infeasible or inefficient regions, thereby improving the effectiveness and robustness of the search. The correction process may involve introducing material constitutive models, geometric constraint functions, or weighting factors based on empirical knowledge.

[0052] Adjusting the search step size based on the local curvature information of the response surface at the current parameter point aims to optimize the convergence speed and stability of the iteration. Curvature information reflects the degree of bending of the response surface near the current point. For example, in flat regions with low curvature, a larger step size can be used to accelerate the search; while in steep or curved regions with high curvature, a smaller step size should be used to avoid overshoot or oscillations, ensuring the stability of the search and accurately approximating the optimal solution. This curvature information can be obtained and utilized through the Hessian matrix or its approximations.

[0053] Based on the updated search direction and adjusted search step size, the local excitation parameters for the next iteration are determined, thereby enabling intelligent and adaptive adjustment of the local excitation parameters to more efficiently stimulate the target's micro-vibration modes.

[0054] This application acquires local stress response data in real time and calculates the local gradient of the response surface based on this data, thus providing clear directional guidance for parameter adjustment. Furthermore, by combining the local material properties and geometric structure of high-uncertainty regions to correct the gradient, the physical rationality and effectiveness of the search direction are ensured, avoiding blind searching. Simultaneously, the search step size is dynamically adjusted using the local curvature information of the response surface, allowing the iterative process to adaptively select an appropriate step distance based on the local morphology of the response surface, thereby significantly improving search efficiency while ensuring convergence stability. This refined parameter adjustment mechanism enables the system to more accurately and quickly approximate the local excitation parameters that can effectively stimulate the target's micro-vibration modes, overcoming the problems of slow convergence and susceptibility to local optima inherent in traditional methods.

[0055] Suppose that in a certain iteration, the system acquires local stress response data of a high-uncertainty region under the current local excitation parameters (e.g., the duration of the virtual heat flux pulse, the amplitude of the virtual pressure wave). First, based on this response data, the local gradient of the response surface at the current parameter point is calculated through numerical simulation or data fitting. This gradient indicates the rate of change of the excitation degree of micro-vibration modes in the direction of the excitation parameters. For example, if the gradient is positive in the duration direction, it means that increasing the duration will enhance vibration excitation. Next, considering that the material in this high-uncertainty region may exhibit nonlinear elasticity at high temperatures, and that its geometry may have sharp corners leading to stress concentration, the system will correct the calculated gradient based on these local material properties and geometry. For example, in stress concentration regions, even if the gradient indicates a certain direction, the search direction may be adjusted due to the risk of local material failure, avoiding high-risk regions or prioritizing the exploration of parameter combinations that can alleviate stress concentration. Simultaneously, local curvature information is obtained by analyzing the Hessian matrix of the response surface near the current point. If the curvature is small, it indicates that the response surface is relatively flat in this region, and the system will use a larger step size for exploration; conversely, if the curvature is large, it indicates that the response surface changes drastically, and the system will use a smaller step size for a more refined search. Finally, by combining the corrected search direction and the adjusted search step size, the local excitation parameters for the next iteration are determined, thereby guiding the optimization process to converge efficiently and stably to the optimal excitation parameters that can effectively excite micro-vibration modes.

[0056] In some embodiments, the step of dynamically adjusting the search direction and step size of the local excitation parameters on the response surface based on the real-time monitored local stress response during the iterative correction process includes: When correcting the local excitation parameters in each iteration, local stress response data of the high uncertainty region is obtained; Based on local stress response data, the material damping characteristics and elastic modulus changes in high uncertainty regions are detected in real time. Based on the detected material damping characteristics and elastic modulus changes, the material property parameters in the high uncertainty region are dynamically updated. Based on the updated material property parameters, recalculate the local gradient and curvature of the response surface at the current parameter point; Based on the recalculated local gradient and curvature, the search direction and step size of the local excitation parameters are adjusted.

[0057] Specifically, during each iteration of correcting the local excitation parameters, the system acquires local stress response data in the high-uncertainty region. This data can originate from sensors deployed in the region, such as micro-strain sensors, or from surface strain field data acquired through image processing techniques. Real-time detection of material damping characteristics and elastic modulus changes in the high-uncertainty region refers to inferring real-time changes in the material's internal damping coefficient and elastic modulus by analyzing the acquired local stress response data, such as characteristics like amplitude attenuation, phase hysteresis, and resonant frequency drift. For example, damping characteristics can be evaluated by analyzing the phase difference between the excitation and response signals, and the elastic modulus can be evaluated by analyzing changes in the resonant frequency. The aim is to obtain the most accurate material mechanical behavior in the high-uncertainty region at the current moment.

[0058] Based on the detected changes in material damping characteristics and elastic modulus, the material property parameters in high-uncertainty regions are dynamically updated. This means that in the simulation model, the material parameters corresponding to high-uncertainty regions are no longer fixed but are adjusted according to real-time detection results. For example, if an increase in damping is detected, the damping coefficient is updated in the model; if a decrease in elastic modulus is detected, the elastic modulus is updated. The aim is to enable the simulation model to more accurately reflect the actual state of the material under dynamic excitation.

[0059] Based on the updated material property parameters, the local gradient and curvature of the response surface at the current parameter point are recalculated. The response surface describes the relationship between local excitation parameters and the excitation degree of micro-vibration modes in the high-uncertainty region. The shape of this response surface changes as material properties change. Therefore, by updating the material property parameters and recalculating the local gradient and curvature, more accurate optimization direction and step size information can be obtained. The purpose is to ensure that the iterative correction process is always based on the most accurate physical model.

[0060] Based on the recalculated local gradient and curvature, the search direction and step size of the local excitation parameters are adjusted. This means that during the iteration process, the adjustment of the local excitation parameters no longer relies solely on preset or static response surface information, but can adaptively adjust according to real-time changes in the material. For example, if the recalculated gradient indicates a new optimal direction, the search direction will be adjusted accordingly; if the curvature information indicates that the response surface becomes flatter or steeper in that region, the step size will also increase or decrease accordingly. The aim is to improve the efficiency and accuracy of iterative correction, ensuring the stable and rapid excitation of the target's microscopic vibration modes.

[0061] This application detects changes in material damping characteristics and elastic modulus in high-uncertainty regions in real time during each iteration of local excitation parameter correction, and dynamically updates material property parameters based on these changes, thereby recalculating the local gradient and curvature of the response surface. It is precisely this dynamic adaptive mechanism that ensures the search direction and step size of the local excitation parameters remain consistent with the actual physical state of the high-uncertainty region, effectively avoiding model mismatch problems caused by changes in material properties, and thus ensuring the accuracy and robustness of the iterative correction process.

[0062] Suppose a region of high uncertainty in an injection mold, such as a thin-walled structure, is marked as such in the initial stress analysis. During the iterative correction process of injecting local excitation to elicit its dynamic response, this thin-walled structure may suffer microscopic damage due to local stress concentration, leading to subtle changes in its material damping properties and elastic modulus.

[0063] First, during each iteration to correct the local excitation parameters, the local stress response data is collected in real time by an array of micro-strain sensors arranged on the thin-walled structure.

[0064] Secondly, based on the collected stress response data, such as by analyzing the phase difference between the excitation frequency and the response frequency, as well as the attenuation rate of the response amplitude, the changes in the material damping coefficient and elastic modulus of the thin-walled structure can be detected in real time. For example, if an increase in the phase hysteresis of the response signal is found, it may indicate enhanced damping characteristics; if the resonant frequency drifts, it may indicate a change in the elastic modulus.

[0065] Next, based on the detected material damping characteristics and elastic modulus changes, the material property parameters of the thin-walled structure in the simulation model are dynamically updated.

[0066] Then, using the updated material property parameters, the local gradient and curvature of the response surface describing the relationship between the local excitation parameters and the degree of excitation of the micro-vibration modes of the thin-walled structure are recalculated at the current parameter point.

[0067] Finally, based on the recalculated local gradient and curvature, the local excitation parameters for the next iteration are adjusted, such as fine-tuning the duration of the virtual heat flux pulse or the waveform of the virtual pressure wave, to more accurately guide the iterative process toward the target micro-vibration mode. In this way, even if the material properties of the thin-walled structure change during the excitation process, the system can adaptively adjust the excitation strategy to ensure efficient and accurate excitation of its micro-vibration modes.

[0068] In some embodiments, the step of obtaining local stress response data in the high-uncertainty region during each iteration of correcting the local excitation parameters includes: Multiple micro-strain sensor arrays are deployed within a region of high uncertainty to collect multi-point local strain data in the region of high uncertainty in real time. Acquire surface strain field data in regions with high uncertainty; By integrating multi-point local strain data and surface strain field data, a three-dimensional strain field in a region with high uncertainty is constructed. In a three-dimensional strain field, identify regions where the strain gradient increases anomalously; Local perturbation excitation was applied to the region with an abnormally large strain gradient, and the dynamic response under local perturbation excitation was monitored; Based on the dynamic response, the stress concentration characteristics and energy dissipation mode of the region with abnormally increased strain gradient are determined to distinguish between the local abnormal response caused by micro-defects and the real response caused by dynamic changes in material properties. The local stress response data is corrected based on the differentiation results and used as the local stress response data for the high uncertainty region.

[0069] Specifically, a micro-strain sensor array can be understood as a group of tiny, highly sensitive sensors strategically deployed within regions of high uncertainty, such as through embedding or surface bonding. Its purpose is to acquire strain information at different locations within this region, thereby providing refined data on the internal mechanical response. Surface strain field data can be obtained through non-contact measurement methods, such as digital image correlation (DIC) or laser speckle interferometry, to provide overall macroscopic strain distribution information for the region, supplementing and verifying the internal local strain data. In practical applications, data fusion can be achieved through finite element model updates, data assimilation algorithms, or machine learning methods, combining discrete internal point data with continuous surface field data to form a comprehensive and accurate three-dimensional strain distribution model. This aims to provide a complete and high-resolution view of the mechanical state for subsequent analysis. Specifically, regions with abnormally large strain gradients typically indicate stress concentration, material inhomogeneity, or potential microscopic defects, such as crack initiation points or pores. Identifying these regions can be accomplished by calculating the spatial derivative of the strain field and comparing it with a preset threshold, aiming to precisely pinpoint potential problem areas requiring further diagnosis. In this process, local perturbation excitation can be a low-energy, high-frequency mechanical vibration, ultrasonic pulse, or micro-thermal pulse, precisely applied to an identified high-gradient region. Monitoring its dynamic response, for example using local vibration sensors or high-speed cameras, aims to detect the unique behavior of this region under minute perturbations. Specifically, anomalous responses caused by microscopic defects (such as cracks) typically exhibit specific nonlinear behavior, high energy dissipation rates, and unique stress wave reflection / scattering modes; while true responses caused by dynamic changes in material properties (such as plastic deformation and phase transformation) may exhibit different frequency responses, damping characteristics, and stress-strain hysteresis loops. By analyzing these differences, the source of the response can be accurately determined, avoiding misinterpreting local effects caused by defects as changes in overall material properties. In practical applications, the correction process may include filtering, eliminating, or compensating for anomalous data caused by microscopic defects to ensure that the final local stress response data used for iterative correction is pure and accurate, providing reliable input for subsequent material property detection and excitation parameter adjustment.

[0070] This application addresses the challenge of accurately acquiring local stress response data in complex mold structures through multi-source data fusion, refined anomaly region identification, and a targeted perturbation diagnostic mechanism. First, a high-resolution three-dimensional strain field is constructed by internally arranging a micro-strain sensor array and combining it with surface strain field data, providing a comprehensive and accurate view of the mechanical state. Second, identifying regions with abnormally increased strain gradients within the three-dimensional strain field allows for precise location of potential stress concentrations or defects. Furthermore, by applying local perturbation excitation to these anomaly regions and monitoring their dynamic response, this application enables in-depth analysis of their stress concentration characteristics and energy dissipation modes. This refined diagnostic approach allows the system to effectively distinguish between local anomaly responses caused by microscopic defects and true responses caused by dynamic changes in material properties. Thus, by correcting the original local stress response data, interference factors are eliminated, ensuring that the acquired data accurately reflects the material damping characteristics and elastic modulus changes in high-uncertainty regions, providing a reliable basis for subsequent iterative correction of local excitation parameters.

[0071] After performing a preliminary structural stress analysis on a high-uncertainty region of an injection mold, it is necessary to obtain its local stress response data for iterative correction of local excitation parameters. First, within this high-uncertainty region, such as near specific weak points in the mold cavity wall, multiple miniature fiber Bragg grating (FBG) strain sensor arrays are arranged to collect multi-point local strain data in real time. Simultaneously, a digital image correlation (DIC) system is used to acquire the full-field strain data of the surface in this region. Subsequently, these internal discrete-point strain data are fused with the continuous surface strain field data, and a precise three-dimensional strain field for this region is constructed using finite element model update technology. In the constructed three-dimensional strain field, the system identifies a point with a significantly higher strain gradient than the surrounding area, which may indicate a small internal crack or material inhomogeneity. To diagnose this anomaly, the system applies a high-frequency, low-amplitude local perturbation excitation to this high-gradient region using a miniature piezoelectric ceramic (PZT) exciter, and monitors its dynamic response using an ultrasonic sensor. By analyzing the attenuation characteristics and spectral changes of the monitored ultrasonic signals, the system determined that the abnormal response was caused by a microscopic defect (e.g., a 0.1 mm microcrack), rather than a dynamic change in the overall material properties. Based on this distinction, the system corrected the original local stress response data, for example, through local data interpolation or by removing outlier data, thus obtaining more accurate local stress response data that eliminated defect interference. This corrected data was then used to detect changes in the material's damping characteristics and elastic modulus in that region, thereby guiding the precise adjustment of local excitation parameters.

[0072] like Figure 2As shown in the embodiments, this application also discloses a computer-aided design-based injection mold structure optimization system, including: Information acquisition module 1 is used to acquire the structural information of the injection mold, identify the geometric feature areas in the mold structure, and perform preliminary structural stress analysis on the geometric feature areas; The reliability assessment module 2 is used to monitor the calculation progress and result stability of each region of the mold structure during the initial structural stress analysis process, so as to assign a quantitative reliability score to each region and mark the region with a reliability score lower than a preset threshold as a high uncertainty region. The optimization suggestion generation module 3 is used to inject local excitation to stimulate the dynamic response of high uncertainty regions, and monitor the energy dissipation mode, stress wave propagation characteristics and high-frequency stress fluctuation spectrum of the dynamic response to diagnose the sensitivity of the region to micro-vibration fatigue. Based on the diagnosis results, combined with the structured design guidelines and the partial convergence simulation data of the region, optimization suggestions are generated for the region. For regions not marked as high uncertainty regions, optimization suggestions are generated based on their simulation analysis results.

[0073] The system of this application can automatically optimize the structure of injection molds, effectively solving the problems of excessive manual intervention, low efficiency and poor reliability in traditional methods, and significantly improving the level of intelligence in mold design.

[0074] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for optimizing the structure of injection molds based on computer-aided design, characterized in that, include: Obtain the structural information of the injection mold, identify the geometric feature areas in the mold structure, and perform preliminary structural stress analysis on the geometric feature areas; During the initial structural stress analysis, the calculation progress and result stability of each region of the mold structure are monitored to assign a quantitative reliability score to each region, and regions with reliability scores below a preset threshold are marked as high uncertainty regions. For regions with high uncertainty, local excitation is injected to stimulate the dynamic response of the region, and the energy dissipation mode, stress wave propagation characteristics, and high-frequency stress fluctuation spectrum of the dynamic response are monitored to diagnose the sensitivity of the region to micro-vibration fatigue. Based on the diagnostic results, and combined with the structured design guidelines and partial convergence simulation data of the region, optimization suggestions are generated for the region. For regions not marked as high uncertainty areas, optimization suggestions are generated based on their simulation analysis results.

2. The method for optimizing the structure of an injection mold based on computer-aided design according to claim 1, characterized in that, The step of monitoring the calculation progress and result stability of each region of the mold structure during the preliminary structural stress analysis, assigning a quantified reliability score to each region, and marking regions with reliability scores below a preset threshold as high uncertainty regions further includes: During the preliminary structural stress analysis, characteristic parameters of the current operating environment of the mold are extracted; Obtain the runtime environment parameters of historical design cases; Based on the differences between the characteristic parameters of the current operating environment and the operating environment parameters of historical design cases, the environmental condition difference index is calculated, and the reference weight of historical design cases when assigning quantitative reliability scores to each region is adjusted according to the environmental condition difference index. Based on the monitored computational progress and result stability of each region, and combined with the adjusted reference weights, a quantitative reliability score is assigned to the computational results of each region, and regions with reliability scores below a preset threshold are marked as high uncertainty regions.

3. The method for optimizing the structure of an injection mold based on computer-aided design according to claim 1, characterized in that, The steps for injecting local excitation to stimulate the dynamic response of a region with high uncertainty, and monitoring the energy dissipation mode, stress wave propagation characteristics, and high-frequency stress fluctuation spectrum of the dynamic response to diagnose the sensitivity of the region to micro-vibration fatigue, include: Based on the characteristic parameters of the current operating environment of the mold and the material properties of the high uncertainty region, the characteristic frequency range and stress reversal amplitude of the micro-vibration fatigue that may be induced in the high uncertainty region are predicted. Based on the predicted characteristic frequency range, the duration of the virtual heat flow pulse is adjusted so that the spectrum of the virtual heat flow pulse matches the characteristic frequency range. Based on the predicted stress reversal amplitude, the waveform of the virtual pressure wave is adjusted so that the virtual pressure wave produces an enhancement effect at the stress reversal amplitude. Adjusted virtual heat flow pulses and virtual pressure waves are injected into regions of high uncertainty as local excitation, and the local stress response in these regions is monitored in real time. Based on the frequency components and amplitude of the local stress response, the parameters of the local excitation are iteratively corrected until the local excitation can excite the micro vibration mode in the high uncertainty region. Monitoring the energy dissipation mode, stress wave propagation characteristics, and high-frequency stress fluctuation spectrum of the dynamic response during iterative correction is used to diagnose the region's sensitivity to micro-vibration fatigue.

4. The method for optimizing the structure of an injection mold based on computer-aided design according to claim 3, characterized in that, The step of iteratively correcting the parameters of the local excitation based on the frequency components and amplitude of the local stress response until the local excitation can excite micro-vibration modes in the high-uncertainty region includes: By gradually adjusting the frequency of local excitation within a preset frequency range and monitoring the local stress response in high uncertainty regions in real time, the peak frequencies and their corresponding amplitudes of all regions with significant local stress responses can be identified. Based on the identified peak frequencies and their corresponding amplitudes, the characteristic parameters of the current operating environment of the mold, and the material properties of the high uncertainty region, the correlation between the vibration mode corresponding to each peak frequency and structural fatigue damage is evaluated to determine the microscopic vibration modes and their characteristic frequencies that play a dominant role in structural fatigue. Adjust the frequency of the local excitation to lock it at the characteristic frequency of the dominant micro-vibration mode; Adjust the amplitude of the local excitation to produce a resonant enhancement effect at the characteristic frequency.

5. The method for optimizing the structure of an injection mold based on computer-aided design according to claim 3, characterized in that, The step of iteratively correcting the parameters of the local excitation based on the frequency components and amplitude of the local stress response until the local excitation can excite micro-vibration modes in the high-uncertainty region includes: Based on the geometric characteristics, material properties, and current operating environment parameters of the high uncertainty region, the nonlinear response characteristics and coupling effects of the high uncertainty region under different excitation parameters are predicted. Based on the predicted nonlinear response characteristics and coupling effects, a response surface is constructed, which describes the relationship between local excitation parameters and the degree of excitation of micro-vibration modes in the high uncertainty region. During the iterative correction process, the search direction and step size of the local excitation parameters are dynamically adjusted on the response surface based on the real-time monitored local stress response. A convergence judgment mechanism is introduced, which determines whether the target micro-vibration mode has been stably excited based on the stability of the local stress response, the purity of the frequency components, and the convergence trend of the amplitude, and stops iterating when the convergence condition is met.

6. The method for optimizing the structure of an injection mold based on computer-aided design according to claim 5, characterized in that, The step of dynamically adjusting the search direction and step size of the local excitation parameters on the response surface based on the real-time monitored local stress response during the iterative correction process includes: Obtain local stress response data under the current local excitation parameters; Calculate the local gradient of the response surface at the current parameter point based on the local stress response data; The local gradient is corrected based on the local material properties and geometry of the high uncertainty region to determine the updated search direction; Adjust the search step size based on the local curvature information of the response surface at the current parameter point; Based on the updated search direction and adjusted search step size, determine the local excitation parameters for the next iteration.

7. The method for optimizing the structure of an injection mold based on computer-aided design according to claim 6, characterized in that, The step of dynamically adjusting the search direction and step size of the local excitation parameters on the response surface based on the real-time monitored local stress response during the iterative correction process includes: When correcting the local excitation parameters in each iteration, local stress response data of the high uncertainty region is obtained; Based on local stress response data, the material damping characteristics and elastic modulus changes in high uncertainty regions are detected in real time. Based on the detected material damping characteristics and elastic modulus changes, the material property parameters in the high uncertainty region are dynamically updated. Based on the updated material property parameters, recalculate the local gradient and curvature of the response surface at the current parameter point; Based on the recalculated local gradient and curvature, the search direction and step size of the local excitation parameters are adjusted.

8. The method for optimizing the structure of an injection mold based on computer-aided design according to claim 7, characterized in that, The step of obtaining local stress response data in the high-uncertainty region during each iteration of correcting the local excitation parameters includes: Multiple micro-strain sensor arrays are deployed within a region of high uncertainty to collect multi-point local strain data in the region of high uncertainty in real time. Acquire surface strain field data in regions with high uncertainty; By integrating multi-point local strain data and surface strain field data, a three-dimensional strain field in a region with high uncertainty is constructed. In a three-dimensional strain field, identify regions where the strain gradient increases anomalously; Local perturbation excitation was applied to the region with an abnormally large strain gradient, and the dynamic response under local perturbation excitation was monitored; Based on the dynamic response, the stress concentration characteristics and energy dissipation mode of the region with abnormally increased strain gradient are determined to distinguish between the local abnormal response caused by micro-defects and the real response caused by dynamic changes in material properties. The local stress response data is corrected based on the differentiation results and used as the local stress response data for the high uncertainty region.

9. The method for optimizing the structure of an injection mold based on computer-aided design according to claim 8, characterized in that, The surface strain field data of the high uncertainty region is obtained by a non-contact measurement method.

10. A computer-aided design-based injection mold structure optimization system, characterized in that, include: The information acquisition module is used to acquire the structural information of the injection mold, identify the geometric feature areas in the mold structure, and perform preliminary structural stress analysis on the geometric feature areas. The reliability assessment module is used to monitor the calculation progress and result stability of each region of the mold structure during the initial structural stress analysis process, so as to assign a quantitative reliability score to each region and mark the region with a reliability score below a preset threshold as a high uncertainty region. The optimization suggestion generation module is used to inject local excitation to stimulate the dynamic response of high uncertainty regions, and monitor the energy dissipation mode, stress wave propagation characteristics and high-frequency stress fluctuation spectrum of the dynamic response to diagnose the sensitivity of the region to micro-vibration fatigue; based on the diagnosis results, combined with the structured design guidelines and the partial convergence simulation data of the region, optimization suggestions are generated for the region. For regions not marked as high uncertainty areas, optimization suggestions are generated based on their simulation analysis results.

Images