A predictive maintenance system for an automotive interior panel injection mold

Abstract

The present application relates to injection mold technical field, disclose a kind of automobile interior board injection mold predictive maintenance system, the system includes the acoustic pulse injection and receiving device configured in the exit and entrance of mold cooling water path, and a signal processing and decision module, the module takes cooling water path as acoustic diagnostic waveguide, determines a deterioration index by analyzing the difference between acoustic echo signal and reference signal, and when the index is abnormal, fault qualitative is carried out in conjunction with macroscopic process parameters, the present application uses the cooling water path of mold itself, in non-invasive manner realizes early identification and state perception to micro-damage inside mold, so that maintenance decision can be based on the real health status of mold, avoid the lag and resource waste of traditional maintenance mode.

Description

Technical Field

[0001] This invention relates to a predictive maintenance system for automotive interior panel injection molds, belonging to the field of injection mold technology. Background Technology

[0002] In the field of injection molding of automotive interior panels and other products, the health of the mold is a fundamental factor affecting production efficiency and product quality. Therefore, the industry generally adopts periodic preventive maintenance based on a fixed number of injections or responsive maintenance after a failure to maintain the operation of the mold.

[0003] The effectiveness of the above maintenance methods is limited when dealing with progressive micro-damage such as micro-cracks or early wear inside the mold. Since these damages usually do not directly cause changes in macroscopic process parameters such as injection pressure or mold temperature in the early stages of their evolution, the internal health status of the mold is actually in an unknown state. This state creates conditions for unplanned production stoppages that may occur during the production process. On the other hand, it also makes it impossible for fixed-cycle maintenance strategies to match the actual wear and tear of different molds, resulting in unnecessary disassembly, maintenance, and resource consumption.

[0004] While existing technologies include methods such as embedding fiber optic sensors inside the mold to acquire its internal state, this approach, which adds extra hardware, is difficult to widely adopt in the cost-effective injection molding industry due to factors such as deployment costs, the need for invasive mold modifications, and operational stability in industrial environments. In summary, existing technologies mainly exhibit the following characteristics in application: 1. A lack of technical means to perceive the evolution of microscopic damage inside the mold without increasing hardware costs or intruding into the mold body; 2. An information gap exists between existing maintenance logic and the actual health state of the mold, leading to a mismatch between maintenance practices and actual needs. Therefore, how to utilize existing components of the injection molding system to establish a monitoring method that can reflect the real-time health state of the mold's internal components is the technical problem this invention aims to solve. Summary of the Invention

[0005] This invention provides a predictive maintenance system for automotive interior panel injection molds. Its main purpose is to solve the problem of early real-time monitoring of microscopic damage inside the mold without changing the hardware configuration of the injection molding system or intruding into the mold body.

[0006] To achieve the above objectives, the present invention provides a predictive maintenance system for automotive interior panel injection molds, characterized in that the system comprises:

[0007] An acoustic pulse injection device is configured at the inlet of the cooling water channel of the mold for injecting an acoustic pulse signal with a reference waveform into the cooling medium circulating in the cooling water channel.

[0008] An acoustic echo receiver is configured at the outlet of the cooling water circuit to receive the acoustic echo signal after the water flows through the mold;

[0009] A signal processing and decision module, connected to an acoustic pulse injection device and an acoustic echo receiving device, is configured to store a reference acoustic echo signal measured by the acoustic echo receiving device when the mold is in a healthy state. The acoustic echo signal received by the acoustic echo receiver will be transmitted in real time. Compared with the reference acoustic echo signal Perform time-domain cross-correlation calculations and determine a characterization of the acoustic echo signal based on the calculation results. Relative to the reference acoustic echo signal Degradation index of the degree of distortion on the waveform The degradation index Compared with the degradation threshold, in the degradation index When the degradation threshold is exceeded, a warning signal is generated. Furthermore, after generating the warning signal, two macroscopic parameters related to the injection molding process—mold temperature and injection pressure—are retrieved, based on the stored degradation index. The correlation rules between anomalies and changes in macroscopic parameters are used to qualitatively determine potential fault types.

[0010] Preferably, the signal processing and decision module is configured to determine the degradation index using the following mathematical relationship. , ,in, For real-time received acoustic echo signals, As the reference acoustic echo signal, The maximum value of the cross-correlation function of the two signals. The energy of the reference acoustic echo signal.

[0011] Preferably, the signal processing and decision module is further configured to perform reference adaptive compensation. Upon receiving a signal from the injection molding machine control system indicating that the injection molding material has been replaced, the signal processing and decision module controls the injection molding machine's clamping system to apply a perturbation pressure pulse with a defined amplitude and duration to the mold during subsequent injection cycles. Furthermore, the signal processing and decision module utilizes the injection molding machine's pressure sensor to acquire the attenuation response of the melt pressure within the mold cavity caused by the perturbation pressure pulse, and based on the characteristics of the pressure attenuation response, it analyzes the reference acoustic echo signal. Dynamic compensation is performed to generate a compensated reference acoustic echo signal that adapts to the physical properties of the newly replaced material.

[0012] Preferably, the signal processing and decision module is configured to calculate the time taken for the pressure decay response to decay from its peak value to half its peak value as the pressure half-life, and determine a relative viscosity coefficient related to the melt viscosity of the new material based on the pressure half-life; and the signal processing and decision module, based on the relative viscosity coefficient, applies a stored nonlinear mapping function to the reference acoustic echo signal. The time-domain axis is adjusted to achieve dynamic compensation.

[0013] Preferably, the signal processing and decision module is further configured to perform closed-loop self-purification of the cooling water channel. When the signal processing and decision module determines that there are fluctuation characteristics in the acoustic echo signal that conform to the noise interference model, it controls the acoustic pulse injection device to send a swept-frequency acoustic signal. Furthermore, the signal processing and decision module analyzes the signal spectrum received by the acoustic echo receiving device to identify the resonant absorption peak caused by bubbles in the cooling medium.

[0014] Preferably, when the signal processing and decision module detects a resonant absorption peak, it controls the acoustic pulse injection device to send sound waves corresponding to the frequency of the resonant absorption peak to remove air bubbles adhering to the inner wall of the cooling water channel. After removal, it switches back to the original operating mode and re-determines the degradation index. .

[0015] Preferably, the signal processing and decision module is further configured to perform accompanying sensing of material mixing uniformity. The signal processing and decision module processes the injection pressure signal to separate high-frequency pressure fluctuation residual signals. Based on the energy characteristics of the pressure fluctuation residual signals, the signal processing and decision module determines the mixing uniformity index of the plastic material in the current injection cycle. When the mixing uniformity index is worse than the process threshold, it generates control instructions for adjusting the plasticizing parameters of the injection molding machine.

[0016] Preferably, both the acoustic pulse injection device and the acoustic echo receiving device are piezoelectric ceramic oscillators.

[0017] Preferably, the association rule includes: when the degradation index When the degradation threshold is exceeded and the injection pressure shows an upward trend, the potential failure type is judged as a flow anomaly related to overflow or microcracks.

[0018] Preferably, the association rule includes: when the degradation index When the degradation threshold is exceeded and the rate of temperature drop in the mold slows down after product ejection, the potential failure type is identified as a heat dissipation anomaly related to carbon buildup or reduced cooling efficiency.

[0019] Compared with the prior art, the beneficial effects of the present invention are:

[0020] 1. This invention transforms the original cooling water circuit in the mold from a single-function temperature control loop into an active acoustic diagnostic network that extends throughout the mold. By injecting acoustic pulses into the cooling medium and analyzing their echo signals, microscopic damage hidden deep inside the mold and difficult to detect by traditional macroscopic process parameters can be monitored externally in the form of acoustic fingerprint distortion. This approach avoids invasive modifications to the mold body or the addition of expensive dedicated sensors. By utilizing the functional reconstruction of existing components, it achieves transparent perception of the internal health status of the mold, enabling maintenance decisions to be based on the actual condition of the mold rather than a fixed production cycle.

[0021] 2. This invention establishes a hierarchical diagnostic logic from microscopic state changes to macroscopic fault characterization. The system first calculates a degradation index characterizing the deviation of the overall health status of the mold through high-sensitivity acoustic echo comparison. This can capture the weak acoustic characteristics generated by physical damage in its infancy, thus issuing an early warning before changes in traditional parameters such as pressure or temperature. Only when this degradation index is abnormal will the system further retrieve and correlate these macroscopic process parameters. This condition-triggered diagnostic process avoids continuous and complex data analysis, ensuring early warning capabilities while taking into account the economy of computing resources and the efficiency of decision-making.

[0022] 3. This invention solves the key bottleneck faced by acoustic diagnostic methods in real industrial environments through an adaptive and self-cleaning mechanism. When changing plastic raw materials with different physical properties, the system can actively apply clamping force perturbation and analyze pressure attenuation, sense changes in material viscosity online, and dynamically compensate the reference acoustic signal. When there are air bubbles in the cooling water circuit, the system can switch to frequency sweep mode to identify the resonant frequency and actively remove the air bubbles. This self-adjustment capability enables the core acoustic diagnostic function to maintain the continuity of monitoring and the reliability of results under changing production conditions and undesirable media environments. Attached Figure Description

[0023] Figure 1 This is a functional architecture diagram of a predictive maintenance system for automotive interior panel injection molds according to the present invention.

[0024] Figure 2 This is a comparison chart of the degradation index of this invention and the monitoring sensitivity of traditional injection pressure to micro-damage.

[0025] Figure 3 This is a core diagnostic flowchart of a predictive maintenance system for automotive interior panel injection molds according to the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are only some embodiments of this invention, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0027] This invention discloses a predictive maintenance system for automotive interior panel injection molds. The technical solution is based on a non-invasive acoustic diagnostic logic. The system utilizes the existing cooling water channels of the mold as waveguides for acoustic diagnosis. Through an acoustic pulse injection and receiving device located at the inlet and outlet of these channels, and a signal processing and decision-making module, a monitoring system capable of real-time sensing of the mold's internal health status is formed. After capturing the acoustic characteristics of the mold's internal state, the system further performs condition-triggered correlation analysis with process parameters during the injection molding process, thereby achieving layered diagnosis from early damage warning to fault characterization. In a specific application scenario, such as for a large automotive door panel interior component injection mold, where high production cycle requirements are needed and modified polypropylene material with added recycled materials is used, early damage warning or stress concentration caused by minor wear on the mold is crucial. Predicting failure risks is crucial for ensuring continuous production. While relying on monitoring process parameters such as mold temperature or injection pressure, this method is limited in its ability to detect latent damage, such as microcrack initiation or localized carbon buildup, which occurs before these parameters show significant drift. To address this challenge, this system is configured to execute a mold condition monitoring procedure based on acoustic time-domain reflectometry. This procedure begins with an acoustic pulse injection device, which can be a piezoelectric ceramic resonator installed in the mold's cooling water inlet pipe. This device injects an acoustic pulse signal with a reference waveform into the circulating cooling water medium. This acoustic pulse signal propagates along the internal cooling water network of the mold and is received by an acoustic echo receiver, also a piezoelectric ceramic resonator, located at the cooling water outlet. This results in an acoustic echo signal carrying the mold's current complete acoustic response information. When the mold is confirmed to be in a healthy initial state, such as during the first production shift after a new mold is put into service, the signal processing and decision-making module will collect and store the acoustic echo signal at this moment, and define it as the reference acoustic echo signal. In each subsequent injection molding cycle, this module transmits the latest received acoustic echo signal in real time. With the stored reference acoustic echo signal Perform time-domain cross-correlation calculations and apply a degradation index characterizing the degree of waveform distortion. The mathematical relation, namely To quantify the deviation of the mold's health status, where the denominator term... The numerator represents the energy of the reference signal, while the numerator... This is the maximum value of the cross-correlation function of the two signals after optimal time-domain alignment. This ratio reflects the waveform similarity of the two signals. When any microscopic damage occurs inside the mold that changes the acoustic impedance, such as microcracks, it will cause changes in the acoustic echo signal. The waveform is distorted, which reduces the similarity. The value increases accordingly. In this way, the system can externally monitor the microscopic damage inside the mold that cannot be directly perceived by process parameters in the form of a quantifiable deterioration index.

[0028] When the system outputs only a scalar degradation index At that time, on-site maintenance personnel still lacked direct evidence for targeted repairs. Therefore, the system was configured to detect deterioration indices... The process of retrieving and correlating relevant process parameters is only triggered after a preset degradation threshold is exceeded. The degradation threshold can be determined according to a deterministic procedure, namely, recording all parameters within the first 1000 production cycles after the new mold or health status is confirmed. The statistical distribution of the values ​​is analyzed, and the mean of this distribution plus three times the standard deviation is set as the initial degradation threshold. The process parameters mainly include the mold temperature and injection pressure obtained from the injection molding machine control system. Based on the preset association rules stored internally, the system qualitatively judges the potential fault types. For example, an association rule can be defined as when the degradation index... When the degradation threshold is exceeded, and the signal processing and decision-making module analyzes the peak values ​​of the injection pressure curve over multiple consecutive cycles and finds that it exhibits a unidirectional upward trend, the system classifies the potential fault type as related to overflow or abnormal melt flow caused by microcracks; another correlation rule can be defined as: when the degradation index... If the degradation threshold is exceeded, and the system analyzes the temperature sensor readings of the mold after the product is ejected and finds that the rate of temperature drop is slower than the historical average, then the potential fault type is judged to be a fault related to the decrease in heat dissipation efficiency caused by local carbon buildup or blockage in the cooling water circuit. In this way, through a diagnostic process that begins with acoustic condition perception and ends with process parameter verification, the system provides early warning capabilities and provides qualitative judgments with engineering guidance value for subsequent maintenance decisions.

[0029] In flexible manufacturing operations, it is common to change plastics using different grades or colors of masterbatches. The different physical properties of these materials, such as melt viscosity, directly alter the reflection and transmission characteristics of sound waves at the mold-melt interface, thus affecting the original baseline acoustic echo signal. To address the challenges posed by the state transition due to failure and false alarms, the system is configured to execute a baseline adaptive compensation procedure. This procedure is triggered when the system receives a signal from the PLC of the injection molding machine control system indicating that the injection material has been changed; this is the preferred information acquisition path. As a backup path, if a material change signal cannot be directly obtained, the system is also configured to trigger this compensation procedure when a systematic jump of more than 10% is detected in the integral or peak value of the injection pressure curve over more than five consecutive injection cycles. After the procedure is initiated, at the end of the holding pressure phase of a subsequent injection cycle, the signal processing and decision module controls the injection molding machine's clamping system to apply a perturbation pressure pulse with a defined amplitude and duration to the mold. For example, a pressure pulse with an amplitude of 0.5% of the total clamping force and a duration of 0.1 seconds. The system uses the injection molding machine's built-in pressure sensor to collect the attenuation response curve of the melt pressure in the mold cavity caused by this perturbation pressure pulse. The system calculates the time taken for the pressure attenuation response to decay from the peak value to half of the peak value, i.e., the pressure half-life. The system obtains a relative viscosity coefficient related to the melt viscosity of the new material, and based on this coefficient, it uses a nonlinear mapping function stored within the system to process the original reference acoustic echo signal. The time-domain axis is dynamically adjusted to generate a reference acoustic echo signal that has been compensated for the physical properties of the newly replaced material. This mechanism enables the acoustic diagnostic function to maintain the continuity and reliability of its monitoring results under changing production conditions.

[0030] In industrial environments, tiny bubbles may precipitate in the circulating cooling medium and adhere to the inner walls of the cooling water channels. These bubbles act as acoustic scatterers, and the noise they generate can interfere with the weak signals caused by damage to the actual mold. To eliminate the influence of this unclean measurement medium, the system is further configured to execute a closed-loop self-cleaning procedure for the cooling water channels. This procedure is automatically triggered when the system determines that the acoustic echo signal contains fluctuation characteristics that conform to a preset noise interference model, such as when the short-term variance of the signal is consistently higher than a statistical threshold. At this time, the system controls the operating mode of the acoustic pulse injection device, switching from sending a single pulse to sending an acoustic signal that performs a linear frequency sweep within a specific frequency band, such as 20-100kHz. The signal processing and decision-making module then simultaneously... The system analyzes the signal spectrum received by the acoustic echo receiver to identify the presence of a resonance absorption peak generated at its resonant frequency by bubbles of a specific size in the cooling medium. If such a resonance absorption peak is detected in the spectrum, the system confirms the presence of bubble interference and immediately switches the operating mode of the acoustic pulse injection device to send a continuous sound wave with appropriately enhanced power corresponding to the resonant absorption peak frequency. This specific frequency sound wave energy can cause the bubbles attached to the inner wall of the water channel to vibrate, thereby detaching and being carried away by the flowing cooling water. After performing a brief acoustic wave clearing operation, the system automatically switches back to the original pulse operating mode and re-confirms the reference, thus completing a self-cleaning closed-loop operation of the acoustic channel. This cleanup mechanism ensures the high signal-to-noise ratio input required by the diagnostic algorithm.

[0031] In addition to monitoring the health of the mold body, the quality of the injection molded parts is also related to the uniformity of plastic mixing, especially when using recycled and virgin materials for injection molding. Therefore, the system is also configured to sense the uniformity of material mixing without adding hardware. This function is achieved by processing the injection pressure signal acquired by the signal processing and decision-making module. Specifically, the module applies a high-pass digital filter to the injection pressure signal of each cycle to separate the high-frequency pressure fluctuation residual signal. A uniformly mixed melt tends to have a smooth pressure curve during flow, while an unevenly mixed melt will generate small, high-frequency resistance fluctuations during flow. These fluctuations are reflected in the pressure fluctuation residual signal. The system calculates the energy characteristics of this residual signal within an injection molding cycle, such as its root mean square value or variance, to determine the mixing uniformity index of the plastic material in a current cycle. When the mixing uniformity index is continuously worse than a preset process threshold, the system can generate control instructions to adjust the plasticizing parameters of the injection molding machine, such as suggesting adjustments to the screw back pressure or melt temperature, in order to improve the mixing effect of the material in subsequent production. This expands the monitoring dimension of the system from the health of the mold equipment to the online perception of the state of the processed material, thereby providing data support for achieving more comprehensive quality control of the injection molding process.

[0032] Example 1: In a car interior panel production unit that provides just-in-time supply to OEMs, an injection mold used to produce the center console side trim panel has been running for 300,000 production cycles. According to its routine maintenance plan, it will undergo disassembly and maintenance at the 400,000th cycle. However, during a data review at the 320,000th cycle, the predictive maintenance system deployed on this mold calculated a degradation index in its signal processing and decision-making module. The value exhibited a low-slope but continuous linear growth trend over approximately 500 consecutive cycles. Its absolute value had not yet reached the preset degradation threshold, therefore no warning signal was generated. During this period, the closed-loop self-purification procedure of the cooling water channel configured in the system was automatically triggered several times. After switching to frequency sweep mode, the acoustic pulse injection device identified the resonance absorption peak caused by bubbles in the cooling medium in the spectrum of the received signal and immediately sent sound waves of the corresponding frequency to purify the channel. This process confirmed the aforementioned degradation index. The continuous changes in the physical source are not noise interference from the acoustic channel, but rather point to the state evolution of the mold itself. When the mold has run for 335,000 cycles, the degradation index... When the value exceeded the preset degradation threshold for the first time, the system immediately generated an early warning signal. At this moment, the values ​​of the two process parameters, mold temperature and injection pressure, retrieved from the injection molding machine control system, did not show any identifiable abnormalities compared to the historical fluctuation range.

[0033] The generation of the warning signal triggered the system's built-in hierarchical diagnostic logic. The signal processing and decision-making module was then configured to perform high-resolution trend analysis on the injection pressure curves of the last one hundred cycles. The analysis showed a slight increase in the peak injection pressure with an average amplitude of 0.2%. The system then matched this with its internally stored correlation rules, i.e., when the degradation index... When both conditions—exceeding the threshold and an upward trend in injection pressure—are met simultaneously, the potential fault type is identified as early damage related to abnormal melt flow. This qualitative judgment and warning signal are output to the maintenance terminal and serve as the basis for targeted inspection of the mold cavity during the next planned material change interval. During the inspection, maintenance personnel used dye penetrant testing and discovered a surface microcrack less than one millimeter long at the root of a deep rib in the cavity. This crack is difficult to detect under conventional visual inspection. It was this microcrack that caused the early overflow of a small amount of melt and a slight increase in injection pressure. The acoustic monitoring method of this system detects the pressure change while it is still within the instrument noise range through acoustic echo signals. The distortion captured this change in physical state. The microcrack was repaired through localized micro-area welding and polishing. The entire inspection and repair process was completed within the planned downtime. After the mold resumed production, the degradation index monitored by the system was... The value also returned to the normal baseline level, and a mold failure event that could have caused an unplanned production stoppage was avoided. At the same time, the original 400,000-cycle disassembly and maintenance plan also had the objective conditions to be further optimized and extended because information on the internal health status of the mold was obtained.

[0034] Example 2: To objectively verify the degradation index in the technical solution of the present invention To assess the sensitivity of monitoring the evolution of microscopic damage within the mold and its early warning capability relative to traditional process parameters, the following comparative verification experiment was designed and executed. The purpose of this experiment is to quantify the degradation index. The correlation between the test and controlled, progressively developing internal physical damage of a mold was investigated. A standard injection molding machine and a test mold were used. A replaceable metal insert was embedded in the mold cavity. Micron-sized scratches of different sizes were pre-machined onto the insert to simulate early wear or microcracks that occur during long-term mold service. Five sets of inserts were prepared, corresponding to a healthy state (no scratches), and four damage states with scratch depths of 50, 100, 150, and 200 micrometers. The acoustic pulse injection and receiving device used in the test was consistent with the piezoelectric ceramic oscillator in the aforementioned specific embodiment. The signal acquisition was performed using… The sampling frequency was set to 1MHz. This setting was chosen to balance signal fidelity with data processing load. Given that the spectral energy of the injected acoustic pulse signal is mainly concentrated below 100kHz, in order to avoid signal aliasing according to the Nyquist sampling theorem and to reserve margin for capturing high-frequency distortion components caused by damage, a sampling frequency of ten times the upper limit of the signal bandwidth was selected. During the experiment, except for changing the plug-in to change the damage state, all other injection molding process parameters, including material batch, melt temperature, injection speed and pressure, were kept constant. This was used as a control condition to eliminate the interference of process fluctuations on the monitoring results.

[0035] The test procedure is as follows: First, the healthy plug-in is installed in the mold, and it runs continuously for one thousand injection cycles. During this stage, the system collects data to establish a stable reference acoustic echo signal. And record the deterioration index under this healthy state. The baseline fluctuation range was then determined; subsequently, four sets of damaged inserts with scratch depths ranging from 50 micrometers to 200 micrometers were sequentially installed. Each set of inserts was continuously run for two thousand injection cycles, and the degradation index was calculated in real time during each cycle. The value was recorded simultaneously with the peak data from the injection pressure sensor. During the experiment, it was observed that when the mold switched from a healthy state to a 50-micron damaged state, the degradation index increased. The average value showed a step jump that deviated from the healthy baseline fluctuation range, while the average injection pressure during the same period still fluctuated within the statistical noise range of the healthy state. As the scratch depth gradually increased, the degradation index... The average value shows a monotonically increasing trend that is positively correlated with the scratch depth. However, the average value of the injection pressure value only begins to show an increase that deviates from its normal fluctuation range when the scratch depth increases to 150 micrometers. See Table 1 for relevant core data.

[0036] Table 1: Comparison of deterioration index and injection pressure under different damage conditions.

[0037]

[0038] The phenomenon presented in the above experimental data is due to the degradation index. The calculation is based on the entire acoustic echo signal. Waveform and reference signal The time-domain cross-correlation comparison shows that even a tiny scratch with a depth of 50 micrometers, as a newly added point of acoustic impedance discontinuity, is sufficient to produce identifiable scattering or reflection of propagating sound waves, thereby causing an echo signal. The distortion in the time-domain waveform is reflected through cross-correlation operations. Regarding the change in value; in contrast, the magnitude of injection pressure mainly depends on the overall resistance of the melt flow in the cavity. When the scratch depth is not sufficient to have a substantial impact on the macroscopic flow of the melt, the change in injection pressure will be submerged in the noise of process fluctuations. Its response does not deviate from the statistical fluctuation range in the initial stage of damage evolution, showing hysteresis.

[0039] Example 3: This example combines Figures 1 to 3 This document describes a predictive maintenance system for automotive interior panel injection molds, such as... Figure 1 As shown, an acoustic pulse injection device injects an acoustic pulse into an injection mold containing cooling water channels. After propagating through the mold, an acoustic echo receiver receives the echo carrying damage information. The system calculates a degradation index characterizing the degree of waveform distortion by comparing the acoustic echoes. And perform a degradation index threshold judgment; if the If the value does not exceed the preset threshold, the process returns. If it exceeds the threshold, the system generates an early warning signal and initiates a fault characterization process. This process correlates macroscopic process parameters such as mold temperature and injection pressure, and makes a qualitative judgment on potential fault types, such as abnormal flow or abnormal heat dissipation, based on preset correlation rules. The system also includes two closed-loop adaptive modules: one is a reference adaptive compensation module, which can sense the material change signal issued by the injection molding machine control system and actively compensate the reference acoustic signal to adapt to the physical characteristics of the new material; the other is a closed-loop self-purification module for the cooling water channel, which can identify and actively remove air bubble interference in the cooling medium, thereby ensuring the signal-to-noise ratio of the core diagnostic signal.

[0040] like Figure 2 As shown, the graph uses scratch depth units. The horizontal axis represents the dimensionless degradation index. The left vertical axis represents the peak injection pressure. The right vertical axis represents the degradation index, and the solid line connecting the circular data points in the graph represents the degradation index. The change in scratch depth is represented by the dashed line connecting the triangular data points, which in turn represents the change in injection pressure. It can be clearly observed from the graph that when the scratch depth increases from 0 (healthy state) to 50... At that time, the degradation index The value of the scratch depth exhibits a significant step increase, while the injection pressure value remains essentially unchanged during the same period, until the scratch depth increases to 150. Only then did the injection pressure begin to show a identifiable upward trend, a comparison that clearly confirms the degradation index. The sensitivity of this indicator for monitoring early microscopic damage in molds is much higher than that of traditional injection pressure parameters.

[0041] like Figure 3 As shown, the process begins with the acquisition of acoustic signals from the cooling water circuit of the injection mold via module 1.0, thereby obtaining real-time acoustic echo signals. Compared with the reference acoustic echo signal stored in the D1 reference acoustic echo signal library under the mold health condition. Subsequently, Module 2.0 is based on and Calculate the degradation index ,Should The value is passed to module 3.0 for diagnosing potential faults. Module 3.0 combines the association rules retrieved from the D2 fault-parameter association rule library with the macroscopic process parameters obtained from the injection molding machine control system to qualitatively characterize the fault and generate a judgment report for maintenance personnel. In addition, when the injection molding machine control system issues a material change signal, module 4.0 is triggered to perform compensation for the reference signal and update the compensated reference signal to the D1 library, thereby ensuring the continuous accuracy of the diagnosis.

[0042] Example 4: In a newly built automotive dashboard air vent assembly injection molding production line, before the new mold and the predictive maintenance system of this invention are put into mass production, a debugging and calibration procedure is performed to match the system's internal diagnostic model and control thresholds with the physical characteristics of the specific mold and the production materials. This procedure first calibrates the baseline adaptive compensation function for material changes. The on-site engineer prepared three types of plastic particles planned for the production of this product: high-gloss ABS, PC / ABS alloy, and TPE soft plastic, each with a known melt flow index. The engineer sequentially fed these three materials into the injection molding machine for trial production. During the stable injection molding stage of each material, a clamping force micro-perturbation pressure pulse as described in the aforementioned specific embodiment was triggered. The system used the injection molding machine pressure sensor to record the melt pressure decay response caused by the pulse and calculated the pressure half-life corresponding to each material. By comparing these three known melt flow indices with three measured pressure half-lives... Numerical values ​​were fitted, and a second-order polynomial curve was generated and stored. This curve was used as a nonlinear mapping function to convert pressure half-life to relative viscosity coefficient, and was subsequently used as a reference acoustic echo signal for material changes during production. Dynamic compensation.

[0043] Subsequently, the procedure entered the stage of constructing a rule base for qualitative correlation of faults. Engineers, ensuring the mold was in a healthy state and using ABS material for stable production, artificially introduced fault simulation. This was achieved by placing a 0.05 mm thick metal shim on the mold parting surface to simulate early overflow or micro-cracks causing flow abnormalities. The system was run continuously for one hundred cycles under this condition, and the degradation index that consistently exceeded the degradation threshold during this period was recorded. The system collects the value of the degradation index and the injection pressure data, which shows a slight upward trend. This set of data is stored in the association rule base, defining the correlation between the degradation index exceeding the threshold and the increase in injection pressure. Then, the gasket is removed to restore the mold to its normal condition. Next, an abnormal heat dissipation is simulated by partially blocking one of the cooling water outlets. This process is repeated for one hundred cycles, and the system records the degradation index under this condition. The system detects the phenomenon of exceeding the threshold and the phenomenon of a slowdown in the rate of temperature drop in the local mold after the product is ejected. This data is stored in the rule base. In this way, by physically simulating and collecting data on several typical fault modes, the system establishes a set of data-supported fault qualitative judgment logic that ranges from acoustic fingerprint anomalies to changes in specific process parameters.

[0044] Finally, the procedure calibrated the threshold for the accompanying sensing function of material mixing uniformity. Engineers first performed 200 injection molding cycles using 100% pure virgin ABS material and recorded the statistical distribution of the mixing uniformity index calculated from the pressure fluctuation residual signal, with a mean of 0.08. Next, the material was replaced with a mixture of virgin and recycled material in a 7:3 ratio, and the cycle was repeated for another 200 cycles. At this point, the mean mixing uniformity index rose to 0.25, and some of the produced products showed flow marks on their appearance. Combined with offline quality inspection of the products, engineers used this to trigger commands to adjust the plasticizing parameters. The process threshold for the mixing uniformity index is set to 0.18. This value lies between the statistical upper limit of the index distribution under the pure virgin material production state and the statistical lower limit of the index distribution when appearance defects occur in the mixed material production. By executing the above complete debugging and calibration procedures, all algorithm models, association rules, and process thresholds related to specific molds and materials within the predictive maintenance system have been assigned definite values ​​from on-site measurements. All functional modules of the system have been transformed from the initial general configuration state to a dedicated state for this specific production task, providing status monitoring and process control assurance for the upcoming mass production.

[0045] Example 5: In the actual operating environment of an injection molding workshop, the circulating water supply temperature of its cooling water system deviates by more than 15 degrees Celsius between winter and summer due to seasonal temperature differences. This change in the physical properties of the cooling water, which serves as an acoustic diagnostic medium, will affect the reference acoustic echo signal of a healthy mold. A drift occurs, leading to a degradation index. The system generates indications unrelated to mold damage. It is configured to continuously monitor the temperature readings at the cooling water inlet in the background and calculate their 24-hour sliding average. This sliding average is then compared with the current reference acoustic echo signal stored within the system. If the temperature difference between the acquired and received temperatures exceeds 5 degrees Celsius, the system generates a prompt to the operator to update the acoustic reference during the next planned shutdown. After receiving confirmation from the operator that the mold is in normal condition, the system automatically executes a new reference acoustic echo signal. The data was collected and stored, and the water temperature at the time of collection was used as the new reference temperature.

[0046] The cooling water channel closed-loop self-purification procedure is triggered based on a dual-condition judgment logic. The execution of this logic involves the signal processing and decision-making modules simultaneously evaluating two dimensions of signal characteristics. The first dimension is the stability of the short-term degradation index, determined by judging the degradation index over the last ten injection molding cycles. The first dimension is determined by whether the standard deviation of the numerical value is greater than a first statistical threshold; the second dimension is the frequency domain characteristics of the signal, determined by analyzing the received acoustic echo signal. Spectrum analysis is performed to determine whether the signal energy integral in the high-frequency noise observation band is greater than a second energy threshold. When both conditions are met, the system determines that there is channel noise interference caused by bubbles and initiates the subsequent frequency sweep identification and resonance removal process.

[0047] Example 6: When deploying the predictive maintenance system of the present invention for the first time on an injection mold with multiple parallel cooling circuits for producing automotive B-pillar interior trim panels, a pre-engineering configuration procedure needs to be performed to determine the physical installation and signal parameters of the system. This procedure provides the basis for subsequent system debugging and calibration. The first step of this procedure is to determine the installation positions of the acoustic pulse injection device and the acoustic echo receiving device. Based on the cooling system design drawings of the mold, the engineers determine the installation positions according to a selection procedure. This procedure specifies that two piezoelectric ceramic transducers are installed on the main water supply manifold and the main water return manifold, respectively. This layout allows the injected acoustic pulses to traverse all parallel cooling circuits inside the mold, thereby obtaining a global acoustic echo signal reflecting the overall state of the mold. The procedure also indicates that if it is necessary to perform enhanced monitoring of a known high-risk area in the future, a pair of transducers can be added to the branch of that specific circuit for local diagnosis.

[0048] After the location was determined, the procedure entered the calibration phase of the core parameters of the acoustic pulse. The engineers performed a channel attenuation test, sending a sweep signal with a frequency linearly scanning from 10kHz to 150kHz into the selected main channel through the acoustic pulse injection device. The acoustic echo receiving device simultaneously recorded the signal amplitude at different frequencies. Based on the test results, the engineers selected the 20kHz to 80kHz frequency band as the working frequency band, where the signal attenuation was lower than a certain preset decibel value and the signal-to-noise ratio was higher than a certain preset ratio. Within this frequency band, the center frequency of the acoustic pulse signal used for diagnosis was set to 50kHz, and its pulse width was set to 50 microseconds. This setting was intended to balance signal energy and time resolution. The pulse injection amplitude was determined through a feedback adjustment program, which started from a low initial voltage and gradually increased the driving voltage of the injection device while monitoring the spectrum of the received signal. When the second harmonic component caused by the nonlinear effect of the acoustic wave began to appear in the spectrum, 90% of the driving voltage at this time was set as the working voltage.

[0049] Finally, the procedure specifies the algorithm for generating qualitative association rules for faults. After obtaining the field dataset with fault labels in the above manner, the signal processing and decision module uses the C4.5 decision tree algorithm to process the dataset to automatically generate association rules. The input feature of this algorithm is the quantized degradation index. The output of the rating, injection pressure change trend, and mold temperature change trend is used to determine the specific fault type. By adopting this well-known algorithm, the construction path from raw data to diagnostic logic is ensured to be reproducible. After completing the above-mentioned pre-engineering configuration procedures, the hardware installation, signal parameters, and core algorithms of the predictive maintenance system on the current B-pillar interior cover mold have been adapted, providing certain initial conditions for subsequent system debugging and calibration.

[0050] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0051] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A predictive maintenance system for automotive interior panel injection molds, characterized in that, The system includes: An acoustic pulse injection device is configured at the inlet of the cooling water channel of the mold for injecting an acoustic pulse signal with a reference waveform into the cooling medium circulating in the cooling water channel. An acoustic echo receiver is configured at the outlet of the cooling water circuit to receive the acoustic echo signal after the water flows through the mold; A signal processing and decision module, connected to an acoustic pulse injection device and an acoustic echo receiving device, is configured to store a reference acoustic echo signal measured by the acoustic echo receiving device when the mold is in a healthy state. The acoustic echo signal received by the acoustic echo receiver will be transmitted in real time. Compared with the reference acoustic echo signal Perform time-domain cross-correlation calculations and determine a characterization of the acoustic echo signal based on the calculation results. Relative to the reference acoustic echo signal Degradation index of the degree of distortion on the waveform The degradation index Compared with the degradation threshold, in the degradation index When the degradation threshold is exceeded, a warning signal is generated. Furthermore, after generating the warning signal, two macroscopic parameters related to the injection molding process—mold temperature and injection pressure—are retrieved, based on the stored degradation index. The correlation rules between anomalies and changes in macroscopic parameters are used to qualitatively determine potential fault types; The signal processing and decision module is configured to determine the degradation index using the following mathematical formula. , ,in, For real-time received acoustic echo signals, As the reference acoustic echo signal, The maximum value of the cross-correlation function of the two signals. The energy of the reference acoustic echo signal; The signal processing and decision-making module is also configured to perform reference adaptive compensation. Upon receiving a signal from the injection molding machine control system indicating that the injection material has been changed, this module controls the injection molding machine's clamping system to apply a perturbation pressure pulse with a defined amplitude and duration to the mold during subsequent injection cycles. Furthermore, the module utilizes the injection molding machine's pressure sensors to acquire the attenuation response of the melt pressure within the mold cavity caused by the perturbation pressure pulse, and based on the characteristics of the pressure attenuation response, it analyzes the reference acoustic echo signal. Dynamic compensation is performed to generate a compensated reference acoustic echo signal that adapts to the physical properties of the newly replaced material. The signal processing and decision module is configured to calculate the pressure half-life by taking the time it takes for the pressure decay response to decay from its peak to half its peak value, and to determine a relative viscosity coefficient related to the melt viscosity of the new material based on the pressure half-life. Furthermore, based on the relative viscosity coefficient, the signal processing and decision module applies a stored nonlinear mapping function to the reference acoustic echo signal. The time-domain axis is adjusted.

2. The predictive maintenance system for automotive interior panel injection molds according to claim 1, characterized in that, The signal processing and decision module is also configured to perform closed-loop self-purification of the cooling water channel. When the signal processing and decision module determines that there are fluctuation characteristics in the acoustic echo signal that conform to the noise interference model, it controls the acoustic pulse injection device to send a swept frequency acoustic signal. Furthermore, the signal processing and decision module analyzes the signal spectrum received by the acoustic echo receiving device to identify the resonant absorption peak caused by bubbles in the cooling medium.

3. The predictive maintenance system for automotive interior panel injection molds according to claim 2, characterized in that, Upon detecting a resonant absorption peak, the signal processing and decision-making module controls the acoustic pulse injection device to send sound waves corresponding to the frequency of the resonant absorption peak to remove air bubbles adhering to the inner wall of the cooling water circuit. After removal, it switches back to the original operating mode and re-determines the degradation index. .

4. The predictive maintenance system for automotive interior panel injection molds according to claim 1, characterized in that, The signal processing and decision module is also configured to perform accompanying sensing of material mixing uniformity. This signal processing and decision module processes the injection pressure signal to separate high-frequency pressure fluctuation residual signals. Furthermore, based on the energy characteristics of the pressure fluctuation residual signals, the signal processing and decision module determines the mixing uniformity index of the plastic material in the current injection cycle. When the mixing uniformity index is worse than the process threshold, it generates control instructions for adjusting the plasticizing parameters of the injection molding machine.

5. The predictive maintenance system for automotive interior panel injection molds according to claim 1, characterized in that, Both the acoustic pulse injection device and the acoustic echo receiving device are piezoelectric ceramic oscillators.

6. The predictive maintenance system for automotive interior panel injection molds according to claim 1, characterized in that, Association rules include: when the degradation index When the degradation threshold is exceeded and the injection pressure shows an upward trend, the potential failure type is judged as a flow anomaly related to overflow or microcracks.

7. The predictive maintenance system for automotive interior panel injection molds according to claim 1, characterized in that, Association rules include: when the degradation index When the degradation threshold is exceeded and the rate of temperature drop in the mold slows down after product ejection, the potential failure type is identified as a heat dissipation anomaly related to carbon buildup or reduced cooling efficiency.

Images