Injection mold cavity state monitoring method, system, equipment and medium

Abstract

The present application relates to a method, system, equipment and medium for monitoring the cavity state of an injection mold, and belongs to the technical field of mold state monitoring. The monitoring method includes: receiving the cavity state data of the injection mold to be detected and performing clock synchronization and preprocessing; dynamically segmenting the time series data according to the injection molding process cycle, generating a sliding window for each process stage of the injection molding process cycle, and marking key points in each sliding window; extracting statistical features and cavity difference features, calculating the trend features of adjacent data key points in each sliding window, combining to obtain a high-dimensional feature vector and inputting it into a normal process benchmark model for comparison, and calculating the feature difference of each feature; performing anomaly detection according to the feature difference, locating the abnormal time period and abnormal spatial area, and obtaining anomaly detection results; generating maintenance prompt information according to the anomaly detection results and sending it to the management terminal. The present application can achieve timely early warning of abnormal mold states.

Description

Technical Field

[0001] The present application relates to the technical field of mold status monitoring, and in particular to a method, system, equipment and medium for monitoring the status inside an injection mold cavity. Background Art

[0002] The injection molding process is one of the most commonly used molding methods in the production of plastic products. Its principle is to inject heated molten plastic into the cooled mold cavity, complete the filling, pressure holding, cooling and other steps in the cavity, and finally solidify and shape it. The injection molding process has the characteristics of high molding efficiency, high product precision, and a wide range of applications. It is widely used in automobile manufacturing, electronic appliances, medical equipment and other fields. With the development of refined industrial manufacturing, the requirements for dimensional accuracy, structural complexity and material properties of injection molded products are constantly increasing. The stability of the process and product consistency have become key factors affecting the quality of the finished product.

[0003] During the injection molding process, the temperature and pressure in the mold cavity are important process parameters that directly affect product quality. Fluctuations in temperature and pressure may cause defects in the product, such as insufficient filling, flash, shrinkage, warping, surface defects or internal stress concentration. Therefore, real-time monitoring of the mold cavity status has become a key link in improving the quality of injection molded products and production efficiency.

[0004] Currently, abnormal conditions that occur during the injection molding process, such as insufficient filling, uneven cooling, or abnormal material flow, can usually only be discovered through subsequent quality inspections. This delayed abnormality detection method may lead to the production of a large number of defective products, increasing resource waste and production costs. In addition, due to the long detection time, it is also difficult to locate the specific source of the problem, especially when the abnormality only occurs in a specific cavity or process stage, and effective adjustment measures cannot be taken in time. Summary of the invention

[0005] In order to achieve timely early warning of abnormal mold conditions, the present application provides a method, system, equipment and medium for monitoring the condition inside the cavity of an injection mold.

[0006] In a first aspect, the present application provides a method for monitoring the state of an injection mold cavity, which adopts the following technical solution:

[0007] A method for monitoring the state of an injection mold cavity, the monitoring method comprising:

[0008] Receive the cavity state data of the injection mold to be tested; the cavity state data includes temperature signal data and pressure signal data collected in real time by multi-point temperature sensors and pressure sensors in the cavity;

[0009] Performing clock synchronization and data preprocessing on the intracavity state data to obtain standardized time series data;

[0010] Dynamically segmenting the time series data according to the injection molding process cycle, generating a sliding window for each process stage of the injection molding process cycle, and marking key points in each sliding window to obtain segmented multi-window time series data and data key points;

[0011] Extracting statistical features and inter-cavity difference features in the multi-window time series data, calculating trend features of adjacent data key points in each sliding window, and combining them to obtain a high-dimensional feature vector;

[0012] Inputting the high-dimensional feature vector into a pre-trained normal process benchmark model for comparison, and calculating the feature difference of each feature in the high-dimensional feature vector;

[0013] Performing anomaly detection according to the characteristic difference, locating the abnormal time period and abnormal spatial area, and obtaining an abnormal detection result of the injection mold to be detected;

[0014] A maintenance prompt message is generated according to the abnormality detection result and sent to the management terminal.

[0015] By adopting the above technical solution, accurate monitoring and abnormal location of the injection mold cavity status are achieved. Based on the comprehensive analysis of dynamic sliding windows and feature vectors, combined with process benchmark models and difference evaluation, process monitoring is upgraded from qualitative observation to quantitative analysis, which can provide timely warnings for abnormal mold conditions and significantly improve the stability of the injection molding process and the efficiency of mold life management.

[0016] Optionally, the steps of dynamically segmenting the time series data according to the injection molding process cycle, generating a sliding window for each process stage of the injection molding process cycle, and marking key points in each sliding window include:

[0017] Receive the input injection molding process information and extract the injection molding process cycle signal, and divide the time axis into multiple continuous process stages;

[0018] Slicing the time series data according to each process stage to generate a data subset for each stage;

[0019] Determine the corresponding sliding window length according to different process stages and generate the sliding window;

[0020] Sliding the sliding window on the data subset of each stage to obtain continuous multi-window time series data;

[0021] According to the preset key point recognition rules of each process stage, the key points in each sliding window are marked to obtain a key point marking set.

[0022] By adopting the above technical solution, the dynamic segmentation of time series data, sliding window generation and key point marking are closely combined. This technical solution realizes the refined analysis of data at different stages of the injection molding process. The adjustment of the dynamic window length adapts to the characteristics of the process stage, and the rule-based key point marking makes the capture of important states more accurate.

[0023] Optionally, the step of extracting the difference features between cavities in the multi-window time series data includes:

[0024] Partitioning the multi-window time series data according to the cavity number of the injection mold to be inspected to obtain partition data;

[0025] Dividing the partition data according to the process stage to obtain a time series data set of each cavity in each process stage;

[0026] Extract the basic feature vector of each cavity at each process stage;

[0027] According to the basic feature vector, mean difference calculation, variance ratio calculation and dynamic difference analysis are performed on each cavity respectively to obtain mean difference characteristics, variance ratio characteristics and dynamic difference characteristics of each cavity;

[0028] Each cavity is grouped into regions, the mean temperature and mean pressure of each group of cavities are calculated, and the inter-group difference analysis is performed to obtain the regional mean difference characteristics;

[0029] The inter-cavity difference feature is obtained by combining the mean difference feature, the variance ratio feature, the dynamic difference feature and the regional mean difference feature.

[0030] By adopting the above technical solutions, a systematic quantitative description of the difference characteristics between cavities is achieved. The difference characteristics between cavities can accurately identify the inconsistency of each cavity in a multi-cavity mold, providing a scientific basis for mold design optimization, process control adjustment and quality improvement.

[0031] Optionally, the step of performing anomaly detection according to the feature difference, locating the abnormal time period and abnormal spatial area, and obtaining the abnormal detection result of the injection mold to be detected includes:

[0032] Comparing the feature difference of each feature in the high-dimensional feature vector with a preset difference threshold value to obtain an abnormal feature set;

[0033] Marking the timestamp and location information corresponding to each abnormal feature in the abnormal feature set;

[0034] According to the timestamp of each abnormal feature, group them based on the time clustering algorithm to obtain the corresponding abnormal time period set;

[0035] Perform regional distribution analysis based on the location information of each abnormal feature to obtain a set of abnormal spatial regions;

[0036] The abnormal time period set and the abnormal space region set are combined to obtain an abnormal detection result of the injection mold to be detected.

[0037] By adopting the above technical solutions, the time period and spatial area positioning of the anomaly can be accurately achieved. By introducing the time clustering algorithm and spatial distribution analysis method, the temporal and spatial distribution of the anomaly is described in two dimensions, making the detection results more detailed and providing strong data support for anomaly positioning and process optimization.

[0038] Optionally, performing regional distribution analysis according to the location information of each abnormal feature to obtain an abnormal spatial region set includes:

[0039] Determine a corresponding cavity number set according to the location information of each abnormal feature;

[0040] Counting the frequency of occurrence of abnormal features of each cavity according to the cavity number set;

[0041] According to the geometric distribution of each cavity, the cavity is divided into regional groups, and the frequency of abnormal features in each region is calculated;

[0042] The frequency of occurrence of abnormal features in each area is compared with the preset frequency threshold, the main abnormal area is identified, and the abnormal spatial area set is obtained.

[0043] By adopting the above technical solution, spatial distribution analysis based on abnormal feature position information is realized, and finally a set of abnormal spatial regions is obtained. Through cavity-level quantification of abnormal frequencies, regional division of geometric distribution and precise comparison of frequency thresholds, the accuracy and efficiency of abnormal analysis are ensured, regional-level positioning of abnormal problems is provided, and reliable data support is provided for mold optimization and process improvement.

[0044] Optionally, the step of generating maintenance prompt information according to the abnormality detection result includes:

[0045] Extracting the abnormal time period set and the abnormal spatial area set according to the abnormal detection result;

[0046] Acquire the time range of each process stage based on the injection molding process cycle;

[0047] Compare the abnormal time period set with the time range of each process stage to determine the process stage corresponding to each abnormal time period;

[0048] Based on the preset strategy database, maintenance strategy matching is performed according to the process phase corresponding to each abnormal time period and the corresponding abnormal spatial area;

[0049] Generate maintenance reminder information according to the matched maintenance strategy.

[0050] By adopting the above technical solutions, the process stage where the abnormality occurs can be accurately located. Through the strategy matching logic in both time and space dimensions and the optimized generation of multi-level abnormality information, efficient and accurate guidance is provided for the maintenance and optimization of injection molds.

[0051] In a second aspect, the present application provides an injection mold cavity state monitoring system, which adopts the following technical solution:

[0052] A system for monitoring the state of an injection mold cavity, the monitoring system comprising:

[0053] A state data receiving module is used to receive the intra-cavity state data of the injection mold to be tested; the intra-cavity state data includes temperature signal data and pressure signal data collected in real time by multi-point temperature sensors and pressure sensors in the cavity;

[0054] A data processing module, used for performing clock synchronization and data preprocessing on the intracavity state data to obtain standardized time series data;

[0055] A segmentation marking module is used to dynamically segment the time series data according to the injection molding process cycle, generate a sliding window for each process stage of the injection molding process cycle, and mark key points in each sliding window to obtain segmented multi-window time series data and data key points;

[0056] A feature extraction module, used to extract statistical features and inter-cavity difference features in the multi-window time series data, calculate trend features of adjacent data key points in each sliding window, and combine them to obtain a high-dimensional feature vector;

[0057] A feature difference calculation module, used for inputting the high-dimensional feature vector into a pre-trained normal process benchmark model for comparison, and calculating the feature difference of each feature in the high-dimensional feature vector;

[0058] An anomaly detection module, used to perform anomaly detection according to the feature difference, locate the abnormal time period and abnormal spatial area, and obtain the anomaly detection result of the injection mold to be detected;

[0059] The maintenance prompt module is used to generate maintenance prompt information according to the abnormality detection result and send it to the management terminal.

[0060] Optionally, the segment marking module includes:

[0061] A process stage division unit, used for receiving the input injection molding process information and extracting the injection molding process cycle signal, and dividing the time axis into a plurality of continuous process stages;

[0062] A data slicing unit, used for slicing the time series data according to each process stage to generate a data subset for each stage;

[0063] A sliding window generating unit, used for determining corresponding sliding window lengths according to different process stages and generating sliding windows;

[0064] A data window sliding unit, which slides the sliding window on the data subset of each stage to obtain continuous multi-window time series data;

[0065] The key point marking unit is used to mark the key points in each sliding window according to the preset key point recognition rules of each process stage to obtain a key point marking set.

[0066] In a third aspect, the present application provides a computer device, which adopts the following technical solution:

[0067] A computer device comprises a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method as described in the first aspect.

[0068] In a fourth aspect, the present application provides a computer-readable storage medium, which adopts the following technical solution:

[0069] A computer-readable storage medium stores a computer program that can be loaded by a processor and execute any one of the methods in the first aspect.

[0070] In summary, the present application includes at least one of the following beneficial technical effects: it realizes accurate monitoring of key states of the injection molding process and rapid positioning of abnormal states, significantly improves production efficiency and consistency of product quality through dual-dimensional positioning of abnormal time periods and spatial areas, accurate calculation of differential characteristics between cavities, and intelligent generation of targeted maintenance strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 It is a first flow chart of a method for monitoring the state in an injection mold cavity according to one of the embodiments of the present application.

[0072] Figure 2 It is a second flow chart of the method for monitoring the state in the injection mold cavity according to one of the embodiments of the present application.

[0073] Figure 3 It is a third flow chart of the method for monitoring the state in the injection mold cavity according to one of the embodiments of the present application.

[0074] Figure 4It is a fourth flow chart of the method for monitoring the state in the injection mold cavity according to one of the embodiments of the present application.

[0075] Figure 5 It is a fifth flow chart of the method for monitoring the state in the injection mold cavity according to one of the embodiments of the present application.

[0076] Figure 6 It is a sixth flow chart of the method for monitoring the state in the injection mold cavity according to one of the embodiments of the present application. DETAILED DESCRIPTION

[0077] In order to make the purpose, technical solutions and advantages of this application more clear, the following Figure 1-6 It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0078] The embodiment of the present application discloses a method for monitoring the state in an injection mold cavity.

[0079] Reference Figure 1 , a method for monitoring the state of an injection mold cavity, the monitoring method comprising:

[0080] Step S101, receiving cavity state data of an injection mold to be inspected;

[0081] The cavity state data includes temperature signal data and pressure signal data collected in real time by multi-point temperature sensors and pressure sensors in the cavity;

[0082] Specifically, the temperature signal data and pressure signal data in the injection mold cavity are collected in real time through the multi-point temperature sensors and pressure sensors arranged in the cavity. In order to ensure the real-time and integrity of the data, the sensor sends the data to the central control system through the industrial bus or wireless communication protocol (such as LoRa or BLE).

[0083] Exemplarily, for temperature signal data and pressure signal data, for example, the pressure sensor records the pressure change in real time during the injection stage, rising rapidly from 0 MPa to 60 MPa, and remains stable during the pressure holding stage, while the temperature is recorded in real time during the cooling stage, gradually dropping from 200°C to 80°C. This real-time, high-precision intracavity state data acquisition provides basic data input for subsequent processing and analysis.

[0084] Step S102, performing clock synchronization and data preprocessing on the intracavity state data to obtain standardized time series data;

[0085] Among them, a timestamp mechanism is used to time align multi-sensor data to avoid data deviation due to clock drift, and the network time protocol (NTP) or hardware timestamp is used synchronously to ensure that each data matches the real time point.

[0086] In some embodiments, the data preprocessing step includes denoising (such as low-pass filtering to remove high-frequency noise), data missing completion (linear interpolation or nearest neighbor method), unit conversion (such as conversion of Celsius temperature to Kelvin temperature) and calibration (correcting sensor errors based on a calibration model).

[0087] Step S103, dynamically segmenting the time series data according to the injection molding process cycle, generating a sliding window for each process stage of the injection molding process cycle, and marking key points in each sliding window to obtain segmented multi-window time series data and data key points;

[0088] The data is sliced ​​using process cycle signals (such as time stamps of injection, pressure holding and cooling stages), and the data of each process stage has independent characteristics (such as rapid pressure rise in the injection stage, stable pressure in the pressure holding stage, and gradual temperature drop in the cooling stage). By generating a sliding window with adjustable length in each process stage, the local characteristics within the window are extracted. At the same time, key points within the window are identified, such as temperature peak, pressure peak, inflection point of rising or falling rate, etc.

[0089] In some embodiments, the window length is dynamically adjusted according to the characteristics of the process stage, such as a short window (20ms) in the injection stage and a long window (200ms) in the cooling stage. The moment when the pressure reaches 60MPa in the injection stage is marked as a critical point for evaluating the adequacy of the filling process.

[0090] Step S104, extracting statistical features and inter-cavity difference features in the multi-window time series data, calculating trend features of adjacent data key points in each sliding window, and combining them to obtain a high-dimensional feature vector;

[0091] Among them, basic statistical features such as mean, variance, peak, range, slope, etc. are calculated from the sliding window. For example, the temperature drop rate (temperature change slope) in the cooling stage is used to evaluate the cooling uniformity; cavity difference features such as mean difference features, variance ratio features, dynamic difference features, etc. are used to comprehensively quantify the differences between multiple cavities when facing multi-cavity injection molds; trend features, such as calculating the change rate or trend direction between adjacent key points, such as the time span from peak pressure to steady state. By combining statistical features, cavity difference features and trend features, a multi-dimensional feature representation for subsequent analysis is generated.

[0092] It can be understood that through multi-window feature extraction and trend analysis, the dynamic changes and local features in the process are captured, providing rich feature information for anomaly detection.

[0093] Step S105, inputting the high-dimensional feature vector into a pre-trained normal process benchmark model for comparison, and calculating the feature difference of each feature in the high-dimensional feature vector;

[0094] Among them, based on a large amount of historical data, including temperature and pressure characteristic data sets collected under normal process conditions of injection molds, a statistical distribution model (such as Gaussian distribution) or a boundary condition model (such as threshold range) is used to define the parameter range of the normal process.

[0095] In some embodiments, the deviation between the real-time feature and the benchmark feature is calculated item by item by comparison, and the feature difference is calculated using a similarity measurement method based on the standardized deviation vector. The degree of difference can be evaluated using Euclidean distance, Mahalanobis distance or other similarity measurement methods.

[0096] Step S106, performing anomaly detection according to the feature difference, locating the abnormal time period and abnormal spatial area, and obtaining an abnormal detection result of the injection mold to be detected;

[0097] Among them, by setting the threshold range of the difference, the anomaly is divided into mild, medium and severe levels. For example, if the real-time pressure peak deviation exceeds 5%, it is judged as an abnormality and the abnormality level is recorded. The specific stage of the abnormality is determined according to the timestamp of the characteristic difference, such as the abnormally slow temperature drop during the cooling stage. At the same time, combined with multi-sensor data, the cavity source or abnormal area is located by cavity number, realizing the dual-dimensional positioning of the abnormality in time and space, making the problem location more accurate.

[0098] Step S107: Generate maintenance prompt information according to the abnormality detection result and send it to the management terminal.

[0099] Among them, maintenance prompt information is generated based on the abnormality detection results, including the abnormality type, occurrence time, cavity number and recommended maintenance measures (such as adjusting process parameters or checking mold parts).

[0100] For example, when the pressure peak of cavity #3 is continuously lower than the normal range, the system sends a prompt "Cavity #3 may be underfilled, please check the injection pressure."

[0101] In the above implementation, accurate monitoring and abnormal location of the injection mold cavity status are achieved. Based on the comprehensive analysis of the dynamic sliding window and the feature vector, combined with the process benchmark model and difference evaluation, the process monitoring is upgraded from qualitative observation to quantitative analysis, and timely warning of abnormal mold status can be given, which significantly improves the stability of the injection molding process and the efficiency of mold life management.

[0102] Reference Figure 2 As an implementation of step S103, the steps of dynamically segmenting the time series data according to the injection molding process cycle, generating a sliding window for each process stage of the injection molding process cycle, and marking key points in each sliding window include:

[0103] Step S201, receiving the input injection molding process information and extracting the injection molding process cycle signal, dividing the time axis into a plurality of continuous process stages;

[0104] The injection molding process cycle usually consists of injection, holding pressure and cooling stages, each of which has independent time characteristics and parameter variation rules. The purpose of extracting process cycle signals is to divide the time axis using these stage markers so that data analysis can focus on the process characteristics of a specific stage. Specifically, process cycle signals include process signals provided by the injection molding machine (such as timestamps of injection, holding pressure and cooling stages) or time stamps recorded by an external monitoring system.

[0105] Step S202, slicing the time series data according to each process stage to generate a data subset for each stage;

[0106] The data performance of different process stages is significantly different. For example, the injection stage is dominated by a rapid increase in pressure, while the cooling stage is dominated by a gradual decrease in temperature. Therefore, slicing the time series data into multiple subsets helps to extract and analyze targeted features in each stage.

[0107] Step S203, determining the corresponding sliding window length according to different process stages and generating a sliding window;

[0108] The dynamic characteristics of each process stage determine the size of the sliding window. For example, the pressure changes rapidly in the injection stage, which is suitable for a shorter window, while the temperature changes slowly in the cooling stage, which is suitable for a longer window. By dynamically adjusting the sliding window length, the stage characteristics can be captured more efficiently.

[0109] In some embodiments, the window length can be set according to the process stage. For example, the sliding window length of the injection stage is set to 20ms to capture the details of the rapid rise in pressure; the sliding window length of the holding stage is set to 50ms to smooth pressure fluctuations and extract stable characteristics; the sliding window length of the cooling stage is set to 200ms to track the overall downward trend of the temperature.

[0110] Specifically, on the data subset of each stage, continuous time windows are generated according to the set window length. The window generation method can slide with a fixed step size (such as half of the window length) to ensure that the overlapping part between windows retains more data characteristics.

[0111] Step S204, sliding the sliding window on the data subset in each stage to obtain continuous multi-window time series data;

[0112] Among them, the role of the sliding window is to divide the time series data into several local subsets. Each subset retains the data characteristics within the window, thereby realizing localized analysis of the time series data. The continuity of multi-window data ensures that the analysis results cover the entire stage without omissions or redundancy.

[0113] In some embodiments, sliding can be performed on a data subset of each stage to divide the original time series data into multiple continuous windows. For example, for the injection stage, if the window length is 20ms, a new window is generated starting from the 0th ms and sliding 10ms each time until the entire stage is covered.

[0114] Step S205 , marking the key points in each sliding window according to the preset key point recognition rules of each process stage, and obtaining a key point marking set.

[0115] Among them, the key point recognition rules of different process stages are set according to their characteristics. Through these rules, important data characteristics in the sliding window can be marked. Through the rule setting based on the characteristics of the process stage and the scanning operation in the sliding window, the key point recognition is accurate and efficient, providing core data support for anomaly detection and process evaluation.

[0116] In some embodiments, key point identification rules can be pre-set according to the characteristics of the process stage. For example, for the injection stage, the time point when the pressure reaches a peak value (such as ≥60MPa) is marked as a key point for evaluating the adequacy of the filling process; for the holding stage, the time point when the pressure change rate is close to zero (such as ΔP<0.01) is marked as a stable point for judging whether the holding pressure is sufficient; for the cooling stage, the turning point where the temperature change rate drops significantly is marked as a key point for analyzing the cooling rate.

[0117] In the above implementation, the dynamic segmentation of time series data, sliding window generation and key point marking are closely combined. This technical solution realizes the refined analysis of data at different stages of the injection molding process. The adjustment of the dynamic window length adapts to the characteristics of the process stage, and the rule-based key point marking makes the capture of important states more accurate.

[0118] Reference Figure 3 As an implementation of step S104, the step of extracting the difference features between cavities in the multi-window time series data includes:

[0119] Step S301, partitioning the multi-window time series data according to the cavity number of the injection mold to be inspected to obtain partition data;

[0120] Among them, in a multi-cavity injection mold, the data of each cavity comes from an independent sensor. Therefore, it is necessary to first partition the multi-window time series data according to the cavity number so that the data of each cavity is independently assigned to the corresponding partition. The partitioning process can be filtered or grouped by data labels (such as cavity numbers) to cluster the time series in the multi-channel data into cavity 1, cavity 2, ..., cavity n according to the cavity number.

[0121] Specifically, data partitioning can be achieved through database query or array indexing. For example, the grouping function groupby in Python is used to extract time series data by number to generate an independent time series set for each cavity.

[0122] Step S302, dividing the partition data according to the process stage to obtain a time series data set of each cavity in each process stage;

[0123] Among them, the injection molding process is usually divided into three main stages: injection, holding pressure and cooling. The parameter change rules of each stage are different. Therefore, it is necessary to further refine the cavity data and divide it according to the process stage. The division is based on timestamps or process marks (such as when the injection pressure reaches the peak point, it is marked as the end point of the injection stage). The time series of each cavity is divided into multiple process stage subsets.

[0124] For example, the division rules are set according to the timestamp range or the specific value of the process signal (such as the pressure stabilization point in the holding stage) to generate a multi-stage data set for each cavity, such as the injection stage data of cavity 1 and the holding stage data of cavity 1.

[0125] Step S303, extracting the basic feature vector of each cavity at each process stage;

[0126] Among them, the time series data of each cavity contains the dynamic change information of temperature and pressure, and the basic feature vector extraction is the process of numerically describing these changes and capturing the core characteristics of the internal state of the cavity.

[0127] In some embodiments, the basic feature vectors include, but are not limited to, temperature mean, temperature variance, temperature peak, temperature change rate, pressure mean, pressure variance, pressure peak, and pressure change rate;

[0128] Step S304, performing mean difference calculation, variance ratio calculation and dynamic difference analysis on each cavity according to the basic feature vector, to obtain mean difference feature, variance ratio feature and dynamic difference feature of each cavity;

[0129] Among them, for the mean difference feature, the mean difference of the basic features of each cavity is calculated to quantify the overall distribution difference between the cavities in terms of temperature and pressure. The difference can be calculated by comparing the maximum and minimum values ​​of all cavity features. Take the temperature mean difference as an example:

[0130] ΔT mean =max(T i,mean )−min(T i,mean );

[0131] In the above formula, ΔT mean is the temperature difference between the cavities, indicating the difference between the maximum and minimum values ​​of the temperature mean of all cavities; T i,mean is the average temperature of the i-th chamber at a certain process stage.

[0132] For the variance ratio feature, by calculating the ratio of the cavity variance to the average variance, we analyze whether the fluctuation degree of each cavity data is consistent. This is used to evaluate the significance of the fluctuation within the cavity and help identify cavities with abnormal fluctuations.

[0133] For dynamic difference characteristics, by comparing the temperature and pressure change rates, the synchronization of the cavity in the dynamic process is analyzed. For example, the standard deviation of the cavity change rate is calculated to dynamically quantify the difference between cavities. Take the pressure change rate as an example:

[0134] σ(ΔP)=std(ΔP1,ΔP2,...,ΔP n );

[0135] In the above formula, σ(ΔP) is the standard deviation of the pressure change rate between cavities, which is used to quantify the consistency of dynamic changes between cavities; ΔP n is the pressure change rate of the nth cavity, indicating the speed at which the cavity pressure changes with time.

[0136] It can be understood that the dynamic difference characteristics reflect the synchronization of the cavity during dynamic changes, which is of great significance for judging whether the filling and cooling processes are uniform.

[0137] Step S305, grouping each cavity into regions, calculating the mean temperature and mean pressure of each group of cavities, and performing inter-group difference analysis to obtain regional mean difference characteristics;

[0138] Among them, the cavities are grouped according to their geometric distribution in the mold (such as left and right areas or upper and lower areas), the mean temperature and pressure of each group of cavities are calculated, and the difference between groups is analyzed. Take the temperature mean difference of the left and right areas as an example:

[0139] ΔT 区域 =mean(T 左 )−mean(T 右 );

[0140] In the above formula, ΔT 区域 is the mean temperature difference between the left area and the right area, mean(T 左 ) is the average value of the mean temperature of all cavities in the left area; mean(T 右 ) is the average of all cavity temperature means in the right region.

[0141] It can be understood that the inter-group difference analysis combined with the geometric layout reflects the asymmetry of the mold design or process configuration, which helps to optimize the cooling channel and injection runner design.

[0142] Step S306, combining the mean difference feature, the variance ratio feature, the dynamic difference feature and the regional mean difference feature to obtain the cavity difference feature.

[0143] Among them, the difference characteristics between cavities are used to comprehensively quantify the differences between cavities, providing a basis for abnormality positioning and process optimization.

[0144] In the above implementation, a quantitative description of the difference characteristics between cavities is systematically achieved. The difference characteristics between cavities can accurately identify the inconsistency of each cavity in a multi-cavity mold, providing a scientific basis for mold design optimization, process control adjustment and quality improvement.

[0145] Reference Figure 4 As an implementation method of step S106, the steps of performing anomaly detection according to feature difference, locating anomaly time period and abnormal spatial area, and obtaining anomaly detection results of the injection mold to be detected include:

[0146] Step S401, comparing the feature difference of each feature in the high-dimensional feature vector with a preset difference threshold to obtain an abnormal feature set;

[0147] Each feature in the high-dimensional feature vector represents the degree of difference in a specific state (such as temperature mean, pressure peak) in the injection molding process. By comparing the difference of these features with the preset threshold, it can be determined whether it exceeds the normal process range. The preset difference threshold can be set through statistical analysis in the benchmark model, usually based on the upper and lower limits or standard deviation range of historical normal data.

[0148] Step S402, marking the timestamp and location information corresponding to each abnormal feature in the abnormal feature set;

[0149] Among them, abnormal features may appear in different time periods. By associating the abnormal features with their timestamps, the time and location of the abnormality can be clearly identified. At the same time, each abnormal feature is accompanied by location information (including specific location or cavity number) through sensor data. Combined with the geometric layout of the cavity, the spatial distribution of the abnormality can be analyzed.

[0150] Step S403, grouping based on the time stamp of each abnormal feature based on a time clustering algorithm to obtain a corresponding abnormal time period set;

[0151] Among them, abnormal features may occur continuously at adjacent time points. These abnormal features need to be grouped through time clustering to clarify the specific time period of the abnormality. Time clustering can effectively reduce the interference of scattered data and make abnormal analysis more focused.

[0152] Specifically, a time series is constructed based on the timestamps of abnormal features, and the timestamps are grouped using a time clustering algorithm (such as DBSCAN). The time interval is used as the clustering radius ϵ, and the time points with an interval less than ϵ are grouped into the same cluster. The set of abnormal time periods is output, and each time period corresponds to a continuous abnormal area.

[0153] Step S404, performing regional distribution analysis based on the location information of each abnormal feature to obtain an abnormal spatial region set;

[0154] Among them, the abnormal spatial area collection can help locate the specific abnormal cavity or mold area.

[0155] Step S405 , combining the abnormal time period set and the abnormal space region set to obtain an abnormality detection result of the injection mold to be detected.

[0156] Among them, the abnormal time period and spatial area information are combined to generate a two-dimensional anomaly detection result, which can clearly describe the temporal and spatial location of the anomaly and the abnormal characteristics.

[0157] In the above implementation, the time period and spatial area positioning of the anomaly are accurately achieved. By introducing the time clustering algorithm and the spatial distribution analysis method, the temporal and spatial distribution of the anomaly is described in two dimensions, making the detection results more detailed and providing strong data support for anomaly positioning and process optimization.

[0158] Reference Figure 5 As an implementation of step S404, performing regional distribution analysis according to the location information of each abnormal feature to obtain an abnormal spatial region set includes:

[0159] Step S501, determining a corresponding cavity number set according to the location information of each abnormal feature;

[0160] Among them, each abnormal feature is bound to the sensor data. The sensor has a unique cavity number in the mold. Therefore, the corresponding cavity number set can be determined according to the source position of the feature. This process is a key step in establishing a connection between the abnormal feature and the mold space layout to ensure the spatial accuracy of the abnormal analysis.

[0161] Step S502, counting the frequency of occurrence of abnormal features of each cavity according to the cavity number set;

[0162] Among them, the frequency statistics of abnormal features are an important indicator to measure the degree of abnormality of each cavity. The higher the frequency, the more serious the abnormal problem of the cavity. Frequency statistics can not only reflect the overall trend of cavity abnormalities, but also provide accurate quantitative data for subsequent regional division and abnormality positioning.

[0163] Step S503, dividing the cavity into area groups according to the geometric distribution of each cavity, and calculating the frequency of occurrence of abnormal features in each area;

[0164] Among them, the cavities in the mold are usually arranged in a geometrically symmetrical manner, such as left and right areas, upper and lower areas, etc. By grouping the cavities, an overall assessment of regional level anomalies can be achieved. The frequency of occurrence of regional level abnormal features reflects the local unevenness or systemic problems of the mold, such as uneven cooling or unreasonable flow channel distribution.

[0165] Step S504 , comparing the frequency of occurrence of abnormal features in each region with a preset frequency threshold, identifying the main abnormal regions, and obtaining an abnormal spatial region set.

[0166] Among them, by comparing the regional abnormal frequency with the preset frequency threshold, it is possible to quickly identify which areas have a higher degree of abnormality, which is a key step in locating abnormal areas. The preset frequency threshold is set based on historical normal data or expert experience to reflect the allowable abnormal frequency range under normal process conditions.

[0167] In the above implementation, spatial distribution analysis based on abnormal feature position information is implemented, and finally a set of abnormal spatial regions is obtained. Through cavity-level quantification of abnormal frequencies, regional division of geometric distribution and precise comparison of frequency thresholds, the accuracy and efficiency of abnormal analysis are ensured, regional-level positioning of abnormal problems is provided, and reliable data support is provided for mold optimization and process improvement.

[0168] Reference Figure 6 As an implementation of step S107, the step of generating maintenance prompt information according to the abnormality detection result includes:

[0169] Step S601, extracting an abnormal time period set and an abnormal spatial region set according to an abnormality detection result;

[0170] Among them, the abnormal time period set includes all abnormal time periods, and the abnormal spatial area set includes all abnormal areas and corresponding cavity numbers, which respectively indicate the time range and spatial location of the abnormality.

[0171] Step S602, obtaining the time range of each process stage based on the injection molding process cycle;

[0172] The injection molding process cycle is divided into injection, holding pressure and cooling stages. Each stage has a specific time range. These time ranges provide process context information, so that abnormal time periods can be accurately located to specific process stages.

[0173] Step S603, comparing the abnormal time period set with the time range of each process stage, and determining the process stage corresponding to each abnormal time period;

[0174] Among them, the matching of abnormal time periods and process stages can reveal the abnormal process background. For example, temperature abnormalities in the cooling stage may be related to the cooling system, while pressure abnormalities in the injection stage may be related to the parameter settings of the injection molding machine.

[0175] It can be understood that by matching the abnormal time period with the process stage, the stage-specific background information of the abnormal problem is provided, which provides a contextual basis for the generation of maintenance strategies.

[0176] Step S604, based on the preset strategy database, matching maintenance strategies according to the process phase corresponding to each abnormal time period and the corresponding abnormal space area;

[0177] Among them, anomalies in different process stages and problems in different spatial areas may correspond to different maintenance strategies. By inputting the time period and spatial area information into the preset strategy database, a suitable maintenance plan can be quickly matched.

[0178] Specifically, the maintenance strategy database contains solutions for each process stage and area problem, for example, for cooling stage anomalies, check the cooling channel flow; for injection stage anomalies, adjust the injection pressure or speed. According to the process stage and abnormal spatial area corresponding to the abnormal time period, first select the strategy subset corresponding to the abnormal process stage, then further filter the specific strategies related to the abnormal area, and then output the maintenance suggestion set matching the abnormal time period and spatial area.

[0179] Step S605: Generate maintenance prompt information according to the matched maintenance strategy.

[0180] Among them, maintenance prompt information can help operators quickly understand the abnormal background and maintenance focus, and improve the efficiency of problem solving.

[0181] In the above implementation, the process stage where the abnormality occurs is accurately located, and efficient and accurate guidance is provided for the maintenance and optimization of the injection mold through strategy matching logic in both time and space dimensions and optimized generation of multi-level abnormality information.

[0182] The embodiment of the present application also discloses a system for monitoring the status of an injection mold cavity.

[0183] A system for monitoring the state of an injection mold cavity, the monitoring system comprising:

[0184] A state data receiving module is used to receive the in-cavity state data of the injection mold to be tested; the in-cavity state data includes temperature signal data and pressure signal data collected in real time by multi-point temperature sensors and pressure sensors in the cavity;

[0185] A data processing module is used to perform clock synchronization and data preprocessing on the intracavity state data to obtain standardized time series data;

[0186] The segmentation marking module is used to dynamically segment the time series data according to the injection molding process cycle, generate a sliding window for each process stage of the injection molding process cycle, and mark key points in each sliding window to obtain segmented multi-window time series data and data key points;

[0187] Feature extraction module, used to extract statistical features and inter-cavity difference features in multi-window time series data, calculate trend features of adjacent data key points in each sliding window, and combine them to obtain high-dimensional feature vectors;

[0188] A feature difference calculation module is used to input the high-dimensional feature vector into a pre-trained normal process benchmark model for comparison, and calculate the feature difference of each feature in the high-dimensional feature vector;

[0189] Anomaly detection module, used to perform anomaly detection based on feature differences, locate abnormal time periods and abnormal spatial regions, and obtain anomaly detection results for the injection mold to be detected;

[0190] The maintenance reminder module is used to generate maintenance reminder information according to the abnormality detection results and send it to the management terminal.

[0191] In the above implementation, it is possible to achieve full-process automated management from data collection, processing, feature extraction to anomaly detection and maintenance prompts. By collecting cavity temperature and pressure data through high-precision sensors, and combining dynamic segmentation, sliding windows and high-dimensional feature extraction, the system can accurately capture key state changes in the injection molding process. Through feature difference calculation and anomaly detection, the system realizes anomaly positioning in both time and space dimensions, providing data support for problem diagnosis. The maintenance prompt module further generates specific maintenance suggestions to help quickly locate the cause of the anomaly and guide operators to take targeted measures, significantly improving the efficiency and product quality of injection molding production, while reducing equipment failures and maintenance costs, which is of great significance to the intelligent development of the injection molding industry.

[0192] As an implementation of the segmentation marking module, the segmentation marking module includes:

[0193] A process stage division unit, used for receiving the input injection molding process information and extracting the injection molding process cycle signal, and dividing the time axis into a plurality of continuous process stages;

[0194] A data slicing unit is used to slice the time series data according to each process stage and generate a data subset for each stage;

[0195] A sliding window generating unit, used for determining corresponding sliding window lengths according to different process stages and generating sliding windows;

[0196] The data window sliding unit slides the sliding window on the data subset in each stage to obtain continuous multi-window time series data;

[0197] The key point marking unit is used to mark the key points in each sliding window according to the preset key point recognition rules of each process stage to obtain a key point marking set.

[0198] An injection mold cavity status monitoring system according to an embodiment of the present application can implement any of the above-mentioned monitoring methods, and the specific working process of each module in the monitoring system can refer to the corresponding process in the above-mentioned method embodiment.

[0199] In the several embodiments provided in this application, it should be understood that the provided methods and systems can be implemented in other ways. For example, the system embodiments described above are only illustrative; for example, the division of a certain module is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed.

[0200] The embodiment of the present application also discloses a computer device.

[0201] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, a method for monitoring the state in the cavity of an injection mold as described above is implemented.

[0202] The embodiment of the present application also discloses a computer-readable storage medium.

[0203] A computer-readable storage medium stores a computer program that can be loaded by a processor and execute any one of the above-mentioned methods for monitoring the in-cavity state of an injection mold.

[0204] Among them, computer-readable storage media can be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, apparatus or device; the program code contained on the computer-readable medium can be transmitted using any appropriate medium, including but not limited to wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0205] It should be noted that in the above embodiments, the description of each embodiment has different emphases, and for the parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0206] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any feature disclosed in this specification (including the abstract and drawings), unless otherwise stated, can be replaced by other equivalent or alternative features with similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

Claims

1. A method for monitoring the state of an injection mold cavity, characterized in that: The monitoring method comprises: Receive the cavity state data of the injection mold to be tested; the cavity state data includes temperature signal data and pressure signal data collected in real time by multi-point temperature sensors and pressure sensors in the cavity; Performing clock synchronization and data preprocessing on the intracavity state data to obtain standardized time series data; Dynamically segmenting the time series data according to the injection molding process cycle, generating a sliding window for each process stage of the injection molding process cycle, and marking key points in each sliding window to obtain segmented multi-window time series data and data key points; Extracting statistical features and inter-cavity difference features in the multi-window time series data, calculating trend features of adjacent data key points in each sliding window, and combining them to obtain a high-dimensional feature vector; Inputting the high-dimensional feature vector into a pre-trained normal process benchmark model for comparison, and calculating the feature difference of each feature in the high-dimensional feature vector; Performing anomaly detection according to the characteristic difference, locating the abnormal time period and abnormal spatial area, and obtaining an abnormal detection result of the injection mold to be detected; Generate maintenance prompt information according to the abnormality detection result and send it to the management terminal; The steps of performing anomaly detection according to the characteristic difference, locating the abnormal time period and abnormal spatial area, and obtaining the abnormal detection result of the injection mold to be detected include: Comparing the feature difference of each feature in the high-dimensional feature vector with a preset difference threshold value to obtain an abnormal feature set; Marking the timestamp and location information corresponding to each abnormal feature in the abnormal feature set; According to the timestamp of each abnormal feature, group them based on the time clustering algorithm to obtain the corresponding abnormal time period set; Perform regional distribution analysis based on the location information of each abnormal feature to obtain a set of abnormal spatial regions; The abnormal time period set and the abnormal space region set are combined to obtain an abnormal detection result of the injection mold to be detected.

2. The method for monitoring the state of an injection mold cavity according to claim 1, characterized in that: The steps of dynamically segmenting the time series data according to the injection molding process cycle, generating a sliding window for each process stage of the injection molding process cycle, and marking key points in each sliding window include: Receive the input injection molding process information and extract the injection molding process cycle signal, and divide the time axis into multiple continuous process stages; Slicing the time series data according to each process stage to generate a data subset for each stage; Determine the corresponding sliding window length according to different process stages and generate the sliding window; Sliding the sliding window on the data subset of each stage to obtain continuous multi-window time series data; According to the preset key point recognition rules of each process stage, the key points in each sliding window are marked to obtain a key point marking set.

3. The method for monitoring the state of an injection mold cavity according to claim 1, characterized in that: The step of extracting the difference features between cavities in the multi-window time series data includes: Partitioning the multi-window time series data according to the cavity number of the injection mold to be inspected to obtain partition data; Dividing the partition data according to the process stage to obtain a time series data set of each cavity in each process stage; Extract the basic feature vector of each cavity at each process stage; According to the basic feature vector, mean difference calculation, variance ratio calculation and dynamic difference analysis are performed on each cavity respectively to obtain mean difference characteristics, variance ratio characteristics and dynamic difference characteristics of each cavity; Each cavity is grouped into regions, the mean temperature and mean pressure of each group of cavities are calculated, and the inter-group difference analysis is performed to obtain the regional mean difference characteristics; The inter-cavity difference feature is obtained by combining the mean difference feature, the variance ratio feature, the dynamic difference feature and the regional mean difference feature.

4. The method for monitoring the state of an injection mold cavity according to claim 1, characterized in that: The steps of performing regional distribution analysis based on the location information of each abnormal feature and obtaining an abnormal spatial region set include: According to the location information of each abnormal feature, a corresponding cavity number set is determined; Counting the frequency of occurrence of abnormal features of each cavity according to the cavity number set; According to the geometric distribution of each cavity, the cavity is divided into regional groups, and the frequency of abnormal features in each region is calculated; The frequency of occurrence of abnormal features in each area is compared with the preset frequency threshold, the main abnormal area is identified, and the abnormal spatial area set is obtained.

5. The method for monitoring the state of an injection mold cavity according to claim 4, characterized in that: The step of generating maintenance prompt information according to the abnormality detection result includes: Extracting the abnormal time period set and the abnormal spatial area set according to the abnormal detection result; Acquire the time range of each process stage based on the injection molding process cycle; Compare the abnormal time period set with the time range of each process stage to determine the process stage corresponding to each abnormal time period; Based on the preset strategy database, maintenance strategy matching is performed according to the process phase corresponding to each abnormal time period and the corresponding abnormal spatial area; Generate maintenance reminder information according to the matched maintenance strategy.

6. An injection mold cavity status monitoring system, characterized in that: The monitoring system comprises: A state data receiving module is used to receive the in-cavity state data of the injection mold to be tested; the in-cavity state data includes temperature signal data and pressure signal data collected in real time by multi-point temperature sensors and pressure sensors in the cavity; A data processing module, used for performing clock synchronization and data preprocessing on the intracavity state data to obtain standardized time series data; A segmentation marking module is used to dynamically segment the time series data according to the injection molding process cycle, generate a sliding window for each process stage of the injection molding process cycle, and mark key points in each sliding window to obtain segmented multi-window time series data and data key points; A feature extraction module, used to extract statistical features and inter-cavity difference features in the multi-window time series data, calculate trend features of adjacent data key points in each sliding window, and combine them to obtain a high-dimensional feature vector; A feature difference calculation module, used for inputting the high-dimensional feature vector into a pre-trained normal process benchmark model for comparison, and calculating the feature difference of each feature in the high-dimensional feature vector; An anomaly detection module, used to perform anomaly detection according to the feature difference, locate the abnormal time period and abnormal spatial area, and obtain the anomaly detection result of the injection mold to be detected; A maintenance prompt module, used to generate maintenance prompt information according to the abnormality detection result and send it to the management terminal; The anomaly detection module is configured to: Comparing the feature difference of each feature in the high-dimensional feature vector with a preset difference threshold value to obtain an abnormal feature set; Marking the timestamp and location information corresponding to each abnormal feature in the abnormal feature set; According to the timestamp of each abnormal feature, group them based on the time clustering algorithm to obtain the corresponding abnormal time period set; Perform regional distribution analysis based on the location information of each abnormal feature to obtain a set of abnormal spatial regions; The abnormal time period set and the abnormal space region set are combined to obtain an abnormal detection result of the injection mold to be detected.

7. The injection mold cavity status monitoring system according to claim 6, characterized in that: The segment marking module comprises: A process stage division unit, used for receiving the input injection molding process information and extracting the injection molding process cycle signal, and dividing the time axis into a plurality of continuous process stages; A data slicing unit, used for slicing the time series data according to each process stage to generate a data subset for each stage; A sliding window generating unit, used for determining corresponding sliding window lengths according to different process stages and generating sliding windows; A data window sliding unit, which slides the sliding window on the data subset of each stage to obtain continuous multi-window time series data; The key point marking unit is used to mark the key points in each sliding window according to the preset key point recognition rules of each process stage to obtain a key point marking set.

8. A computer device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 5 when executing the program.

9. A computer-readable storage medium, characterized in that: A computer program is stored which can be loaded by a processor and execute the method according to any one of claims 1 to 5.

Images