An intelligent detection system for mold clamping status

Through pressure data acquisition and cluster analysis, abnormalities in the mold clamping state are identified, which solves the problem of degradation of abnormal identification of multiple clamping positions in the prior art, improves the accuracy of abnormal identification of abnormal mold clamping, and ensures the mold Safety and stability.

CN118606730BActive Publication Date: 2024-11-08NANTONG OPEN UNIV (NANTONG ARCHITECTURE VOCATIONAL & TECH SCHOOL NANTONG COMMUNITY EDUCATION SERVICE GUIDANCE CENT)
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411068450.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-11-08
Estimated Expiration
2044-08-06

AI Technical Summary

Technical Problem

When monitoring the clamping state of the mold, it is difficult to accurately identify changes in the overall data distribution caused by abnormalities in multiple clamping positions, resulting in a decrease in the accuracy of abnormal identification of single-positions.

Method used

The pressure data acquisition module, data clustering module, cluster analysis module and mold clamping abnormality judgment module are used to obtain the fluctuation degree, local outlier factor and abnormality degree of clustering through clustering, judge the mold clamping abnormality, and improve abnormality recognition. accuracy.

Benefits of technology

Through clustering and abnormality analysis, the accuracy of identification of mold clamping abnormalities is improved, and the abnormality in the mold clamping state can be more accurately judged, ensuring the safety and stability of mold use.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118606730B_ABST
    Figure CN118606730B_ABST
Patent Text Reader

Abstract

This invention relates to the field of data processing technology, specifically to an intelligent detection system for mold clamping status, comprising: a pressure data acquisition module for acquiring pressure data; a data clustering module for clustering the pressure data to obtain clusters; a cluster analysis module for determining the degree of abnormality of a cluster based on its fluctuation, local outlier factors of data points within the cluster, and normal and outlier data points; and a mold clamping anomaly judgment module for obtaining anchor point clusters and non-anchor point clusters, and determining whether mold clamping is abnormal based on the degree of abnormality of non-anchor point clusters, the degree of abnormality of anchor point clusters, and the distance between the center points of non-anchor point clusters and anchor point clusters. This invention reduces the impact of anomalies at multiple clamping positions on the identification of anomalies at a single position when detecting abnormal mold clamping, thus improving the accuracy of mold clamping anomaly identification.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of data processing, and in particular to an intelligent detection system for a mold clamping state. Background Art

[0002] Mold clamping is to clamp the mold to be used on the base through a clamping tool to prevent the mold from displacement or abnormal vibration that affects the quality of mold production. Since the clamping pressure of the mold will change during the actual production process, the mold clamping detection system can monitor the mold clamping status in real time, detect abnormal situations in time, and ensure the safety and stability of mold use.

[0003] The existing technology calculates the local outlier factor representing the abnormal clamping force data point through the CBLOF algorithm (clustering-based local anomaly factor detection algorithm), so as to detect abnormal clamping of the mold. However, in the actual monitoring process, the clamping force anomaly of a single clamping position of the mold can be accurately identified. However, when anomalies occur in multiple clamping positions, resulting in changes in the overall data distribution, the accuracy of anomaly identification for a single position decreases. Summary of the invention

[0004] In order to solve the above problems, the present invention provides an intelligent detection system for mold clamping status.

[0005] The intelligent detection system for mold clamping state of the present invention adopts the following technical solution:

[0006] An embodiment of the present invention provides a mold clamping state intelligent detection system, which includes the following modules:

[0007] A pressure data acquisition module, used to acquire pressure data, wherein the pressure data includes a plurality of data points, each data point corresponding to a pressure value;

[0008] A data clustering module, used to cluster the pressure data to obtain several clusters of pressure data;

[0009] A cluster analysis module is used to obtain the fluctuation degree of each cluster according to the pressure value corresponding to the data point in each cluster, obtain the local outlier factor of each data point in each cluster, obtain a number of normal data points and a number of outlier data points of each cluster according to the local outlier factor of the data point, and obtain the abnormality degree of each cluster according to the fluctuation degree of each cluster, the local outlier factor of each data point in each cluster, and a number of normal data points and a number of outlier data points of each cluster;

[0010] The mold clamping abnormality judgment module is used to obtain several anchor point clusters and several non-anchor point clusters according to the abnormality degree of each cluster, obtain the final abnormality degree of each non-anchor point cluster according to the abnormality degree of the non-anchor point cluster, the abnormality degree of the anchor point cluster and the distance between the center points of the non-anchor point cluster and the anchor point cluster, and judge the mold clamping abnormality according to the final abnormality degree of the non-anchor point cluster.

[0011] Furthermore, the step of obtaining the fluctuation degree of each cluster according to the pressure value corresponding to the data point in each cluster includes the following specific steps:

[0012] Record any cluster as the target cluster;

[0013]

[0014] In the formula, is the maximum value of the pressure value corresponding to all data points in the target cluster, is the preset standard pressure value, To take the absolute value, is the number of all data points in the target cluster, is the pressure value corresponding to the jth data point in the target cluster, is the average pressure value corresponding to all data points in the target cluster, is the standard deviation of the pressure values ​​corresponding to all data points in the target cluster, is the fluctuation degree of the target cluster.

[0015] Furthermore, the step of obtaining a number of normal data points and a number of outlier data points of each cluster according to the local outlier factors of the data points includes the following specific steps:

[0016] A first threshold is preset, denoted as th1. If the local outlier factor of any data point in any cluster is less than or equal to th1, then the data point is a normal data point of the cluster; if the local outlier factor of any data point in any cluster is greater than th1, then the data point is an outlier data point of the cluster.

[0017] Furthermore, the abnormality degree of each cluster is obtained according to the fluctuation degree of each cluster, the local outlier factor of each data point in each cluster, and a number of normal data points and a number of outlier data points of each cluster, and the specific steps include the following:

[0018]

[0019] In the formula, is the fluctuation degree of the target cluster, is the minimum value of the fluctuation degree among all clusters, is the maximum value of the fluctuation degree among all clusters, is the number of outlier data points in the target cluster, is the average distance between the gth outlier data point in the target cluster and all normal data points in the target cluster, is the local outlier factor of the g-th outlier data point in the target cluster, is the mean of the local outlier factors of all normal data points in the target cluster, is the sinc function, To take the absolute value, is the abnormality degree of the target cluster.

[0020] Furthermore, the final abnormality degree of each non-anchor cluster is obtained according to the abnormality degree of the non-anchor cluster, the abnormality degree of the anchor cluster, and the distance between the center points of the non-anchor cluster and the anchor cluster, and the specific steps include the following:

[0021] Record any non-anchor cluster as the target non-anchor cluster;

[0022]

[0023] In the formula, is the number of anchor clusters in all clusters, is the abnormality degree of the target non-anchor cluster, is the abnormality degree of the kth anchor cluster, is the distance between the target non-anchor cluster and the center point of the kth anchor cluster, is the sigmoid function, is the final abnormality degree of the target non-anchor cluster.

[0024] Furthermore, the method of judging the mold clamping abnormality according to the final abnormality degree of the non-anchor point cluster includes the following specific steps:

[0025] A second threshold is preset, denoted as th2, if , take the target non-anchor cluster as an abnormal cluster, and judge whether there is an abnormal cluster among all non-anchor clusters. If there is an abnormal cluster, the mold clamping is abnormal. If there is no abnormal cluster, the mold clamping is not abnormal.

[0026] Furthermore, the step of obtaining a plurality of anchor clusters and a plurality of non-anchor clusters according to the abnormality degree of each cluster includes the following specific steps:

[0027] The time corresponding to the center point of each cluster is obtained in the pressure data, and the abnormality degrees of the corresponding clusters are arranged in the order of the time to obtain an abnormality degree sequence. The abnormality degree sequence is detected by the AMPD algorithm to obtain several minimum abnormality degrees of the abnormality degree sequence. The cluster corresponding to the minimum abnormality degree is recorded as the anchor cluster, and the cluster that is not the anchor cluster is recorded as the non-anchor cluster.

[0028] Furthermore, the specific method for obtaining the local outlier factor of each data point in each cluster is as follows:

[0029] Input any cluster of pressure data into the CBLOF algorithm, and output the local outlier factor of each data point in the cluster.

[0030] Furthermore, the clustering of the pressure data to obtain a number of clusters of the pressure data includes the following specific steps:

[0031] Iterative self-organizing clustering is performed on the pressure data, and the distance measurement adopts the Euclidean distance between data points in the pressure data to obtain several clusters of pressure data.

[0032] Furthermore, the specific method for obtaining the pressure data is as follows:

[0033] The mold is placed on the fixture for clamping and fixing, TH1 pressure sensors and the fixture are connected through supporting components, and pressure values ​​of multiple clamping positions are respectively collected through the pressure sensors. The pressure values ​​of multiple clamping positions are output every TH2 seconds, and all pressure values ​​collected within the last TH3 hours are arranged in chronological order to obtain a pressure data, TH1 is a preset first value, TH2 is a preset second value, and TH3 is a preset third value.

[0034] The beneficial effect of the technical solution of the present invention is that after collecting the pressure data, the present invention obtains several clusters of pressure data by clustering. Since the significant fluctuation of the data in the cluster represents the sensitivity of the numerical change of the normal pressure value clustering position relative to the outlier pressure value, the fluctuation degree of each cluster is obtained by analyzing the pressure value corresponding to the data point in each cluster, and then the local outlier factor of the data point is obtained, and it is judged whether the data point in each cluster is outlier, and the normal data point and outlier data point of each cluster are obtained. In order to judge whether there is an abnormality in the mold clamping, the fluctuation degree of the cluster, the cluster The local outlier factor of each data point in the cluster, as well as the normal data points and outlier data points, is used to obtain the abnormality degree of the cluster, which reflects the possibility of abnormal mold clamping within the time range of the cluster. In order to judge the abnormal clamping state of the positional relationship distribution of clusters with different abnormality degrees, the final abnormality degree of the non-anchor cluster is obtained through the abnormality degree of the non-anchor cluster, the abnormality degree of the anchor cluster, and the distance between the center points of the non-anchor cluster and the anchor cluster. Then, whether there is abnormality in the mold clamping is judged according to the final abnormality degree, which improves the accuracy of identifying mold clamping abnormalities. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 A system framework diagram of a mold clamping state intelligent detection system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation, structure, features and effects of a mold clamping state intelligent detection system proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0038] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0039] The specific scheme of the intelligent detection system for mold clamping state provided by the present invention is described in detail below with reference to the accompanying drawings.

[0040] See also Figure 1 , which shows a system framework diagram of a mold clamping state intelligent detection system provided by an embodiment of the present invention, the system includes the following modules:

[0041] Pressure data acquisition module:

[0042] It should be noted that this embodiment is an intelligent detection system for mold clamping status, and data needs to be collected first before starting analysis and processing.

[0043] Specifically, the mold is placed on a fixture for clamping and fixing, TH1 pressure sensors and the fixture are connected through supporting members to ensure the authenticity of the collected data, and the pressure values ​​of multiple clamping positions are respectively collected by the pressure sensors, and the pressure values ​​of multiple clamping positions are output every TH2 seconds. All pressure values ​​collected within the last TH3 hours are arranged in chronological order to obtain a pressure data, TH1 is a preset first value, TH2 is a preset second value, and TH3 is a preset third value. This embodiment is described as TH1=5, TH2=1, and TH3=1.

[0044] It should be noted that the pressure data is two-dimensional data, and the pressure data includes several data points, each data point corresponds to a pressure value and a time, and the same time corresponds to multiple pressure values, that is, the pressure values ​​of multiple clamping positions at any time.

[0045] At this point, the pressure data is obtained.

[0046] Data clustering module:

[0047] It should be noted that the above module obtains pressure data, and through the differences between data fluctuations in each category range during the clustering process, the differences in outlier factors of each point for CBLOF between comprehensive data intervals are identified to identify abnormal clamping states.

[0048] Specifically, the pressure data is clustered to obtain several clusters of pressure data, as follows:

[0049] Iterative self-organizing clustering (ISODATA clustering) is performed on the pressure data. The distance measurement uses the Euclidean distance between data points in the pressure data to obtain several clusters of pressure data; each cluster contains several data points, and each data point corresponds to a time and a pressure value.

[0050] It should be noted that performing iterative self-organizing clustering on pressure data to obtain a number of clusters of pressure data is an existing method of ISODATA clustering algorithm, which will not be described in detail in this embodiment.

[0051] Cluster analysis module:

[0052] It should be noted that, for the analysis of the multiple clusters obtained, the significant data fluctuations of a single cluster represent the sensitivity of the normal pressure value clustering position to the numerical changes of the outlier pressure values. When the normal pressure values ​​are clustered more closely, it means that the clamping force of the clamp is relatively stable, and when the clamping pressure is abnormal, the abnormal pressure value can be accurately identified; but when the data fluctuations increase, it means that the clamping force of the clamp is abnormal, and small changes in clamping pressure are difficult to highlight from the discrete data background. Therefore, it is necessary to combine the significant data fluctuations of the cluster with the difference in the actual outlier characteristics of the outlier data points to analyze the clamping abnormality.

[0053] Specifically, the fluctuation degree of each cluster is obtained according to the pressure value corresponding to the data point in each cluster, as follows:

[0054] Record any cluster as the target cluster.

[0055]

[0056] In the formula, is the maximum value of the pressure value corresponding to all data points in the target cluster, is the preset standard pressure value. To narrate, To take the absolute value, is the number of all data points in the target cluster, is the pressure value corresponding to the jth data point in the target cluster, is the average pressure value corresponding to all data points in the target cluster, is the standard deviation of the pressure values ​​corresponding to all data points in the target cluster, is the fluctuation degree of the target cluster.

[0057] It should be noted that Indicates the difference between the maximum pressure value in the target cluster and the preset standard pressure value, indicating the deviation between the maximum pressure value in the target cluster and the standard pressure value, and judging the abnormally large pressure value, which is conducive to increasing the accuracy of the target cluster change; It indicates the change of the pressure value relative to the mean in the target cluster. When the pressure value deviates from the mean, the value of the fraction is also larger, which means that there are more pressure values ​​with larger deviations from the mean in the target cluster.

[0058] Furthermore, the local outlier factor of each data point in each cluster is obtained, and a number of normal data points and a number of outlier data points in each cluster are obtained according to the local outlier factor of the data point, as follows:

[0059] Any cluster of pressure data is input into the CBLOF algorithm, and the local outlier factor of each data point in the cluster is output. A first threshold is preset, denoted as th1. This embodiment is described as th1=1. If the local outlier factor of any data point in any cluster is less than or equal to th1, then the data point is a normal data point of the cluster; if the local outlier factor of any data point in any cluster is greater than th1, then the data point is an outlier data point of the cluster.

[0060] It should be noted that using the CBLOF algorithm to obtain the local outlier factor of each data point in each cluster is an existing method of the CBLOF algorithm and will not be described in detail in this embodiment.

[0061] Furthermore, according to the fluctuation degree of each cluster, the local outlier factor of each data point in each cluster, and a number of normal data points and a number of outlier data points in each cluster, the abnormality degree of each cluster is obtained, as follows:

[0062]

[0063] In the formula, is the fluctuation degree of the target cluster, is the minimum value of the fluctuation degree among all clusters, is the maximum value of the fluctuation degree among all clusters, is the number of outlier data points in the target cluster, is the average distance between the gth outlier data point in the target cluster and all normal data points in the target cluster, is the local outlier factor of the g-th outlier data point in the target cluster, is the average local outlier factor of all normal data points in the target cluster, is a sinc function. This embodiment uses the sinc function to perform normalization processing. To take the absolute value, is the abnormality degree of the target cluster.

[0064] It should be noted that It indicates the significant fluctuation characteristics of the fluctuation degree of the target cluster after clustering in the fluctuation degree of all clusters. The larger the value is, the more obvious the fluctuation of the clamping pressure value of the fixture is in the time range where the target cluster is located. It means that there is interference in the overall clamping situation of the fixture and the working environment of the mold has changed. Therefore, the mold clamping system is less sensitive to smaller fluctuations in pressure data in the time range where the target cluster is located, and will not misjudge the abnormal clamping state due to smaller pressure value fluctuations.

[0065] It should be further explained that It is the average distance between the gth outlier data point in the target cluster and the normal data points in the target cluster, and the ratio of the difference mapping value of the local outlier factor between the outlier data point and other normal data points. The numerator average distance indicates whether the outlier data point is out of the tight clustering position in the cluster, that is, the clamping pressure value of a certain fixture has a more obvious data change relative to other clamping pressure values ​​at the time when the pressure value is collected. The difference in the local outlier factor value in the denominator indicates that the outlier data point is screened out due to the large local outlier factor value. However, in fact, the pressure values ​​collected by sensors at various positions within the time series range of the cluster all show similar data changes. Therefore, the target cluster is relatively close to the cluster with denser data point distribution, and the data point distribution of the target cluster is relatively discrete. At the same time, it is mapped through the sinc function. Closer, When it is closer to 0, the function mapping value is made closer to the maximum value 1.

[0066] It should be noted that, therefore, it is possible to infer whether a greater degree of discrete distribution between data points appears in the time series range where the target cluster is located through a larger average distance value and a smaller local outlier factor difference. Therefore, the target cluster is more likely to represent a mold clamping anomaly within a time series.

[0067] At this point, the abnormality degree of each cluster is obtained.

[0068] Mold clamping abnormality judgment module:

[0069] It should be noted that after obtaining the abnormal degree of the cluster, it is necessary to judge the abnormal clamping state according to the positional relationship distribution of clusters with different abnormal degrees. Since the time of the abnormal clamping state is uncertain, the abnormal clamping state may not have occurred or may have occurred at the last moment of the pressure data. The abnormal clamping state will produce cumulative changes in the abnormal degree in time series. In order to avoid the small difference in the abnormal degree of adjacent clusters, the abnormality can be analyzed by accumulating abnormalities.

[0070] Specifically, several anchor clusters and several non-anchor clusters are obtained according to the abnormality degree of each cluster, as follows:

[0071] The time corresponding to the center point of each cluster is obtained in the pressure data, and the abnormality degrees of the corresponding clusters are arranged in the order of the time to obtain an abnormality degree sequence. The abnormality degree sequence is detected by the AMPD algorithm to obtain several minimum abnormality degrees of the abnormality degree sequence. The cluster corresponding to the minimum abnormality degree is recorded as the anchor cluster, and the cluster that is not the anchor cluster is recorded as the non-anchor cluster.

[0072] Furthermore, according to the abnormality degree of the non-anchor cluster, the abnormality degree of the anchor cluster, and the distance between the center points of the non-anchor cluster and the anchor cluster, the final abnormality degree of each non-anchor cluster is obtained, as follows:

[0073] Any non-anchor cluster is recorded as the target non-anchor cluster.

[0074]

[0075] In the formula, is the number of anchor clusters in all clusters, is the abnormality degree of the target non-anchor cluster, is the abnormality degree of the kth anchor cluster, is the distance between the target non-anchor cluster and the center point of the kth anchor cluster, is the sigmoid function, which is used for normalization. is the final abnormality degree of the target non-anchor cluster.

[0076] Furthermore, the final abnormality degree of the non-anchor cluster is used to determine whether the mold clamping is abnormal, as follows:

[0077] A second threshold is preset, denoted as th2. This embodiment takes th2=0.75 as an example. , take the target non-anchor cluster as an abnormal cluster, and judge whether there is an abnormal cluster among all non-anchor clusters. If there is an abnormal cluster, the mold clamping is abnormal. If there is no abnormal cluster, the mold clamping is not abnormal.

[0078] It should be noted that when there is an abnormality in the mold clamping, the use of the mold needs to be stopped immediately, and the clamp needs to be adjusted before use.

[0079] At this point, the abnormality detection of mold clamping is completed by performing abnormal analysis on the collected pressure data.

[0080] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent detection system for mold clamping status, It is characterized in that The system includes the following modules: A pressure data acquisition module, used to acquire pressure data, wherein the pressure data includes a plurality of data points, each data point corresponding to a pressure value; A data clustering module, used to cluster the pressure data to obtain several clusters of pressure data; A cluster analysis module is used to obtain the fluctuation degree of each cluster according to the pressure value corresponding to the data point in each cluster, obtain the local outlier factor of each data point in each cluster, obtain a number of normal data points and a number of outlier data points of each cluster according to the local outlier factor of the data point, and obtain the abnormality degree of each cluster according to the fluctuation degree of each cluster, the local outlier factor of each data point in each cluster, and a number of normal data points and a number of outlier data points of each cluster; The mold clamping abnormality judgment module is used to obtain a number of anchor point clusters and a number of non-anchor point clusters according to the abnormality degree of each cluster, obtain the final abnormality degree of each non-anchor point cluster according to the abnormality degree of the non-anchor point cluster, the abnormality degree of the anchor point cluster and the distance between the center point of the non-anchor point cluster and the anchor point cluster, and judge the mold clamping abnormality according to the final abnormality degree of the non-anchor point cluster; The specific steps of obtaining the fluctuation degree of each cluster according to the pressure value corresponding to the data point in each cluster are as follows: Record any cluster as the target cluster; In the formula, is the maximum value of the pressure value corresponding to all data points in the target cluster, is the preset standard pressure value, To take the absolute value, is the number of all data points in the target cluster, is the pressure value corresponding to the jth data point in the target cluster, is the average pressure value corresponding to all data points in the target cluster, is the standard deviation of the pressure values ​​corresponding to all data points in the target cluster, is the fluctuation degree of the target cluster; The method of obtaining the abnormality degree of each cluster according to the fluctuation degree of each cluster, the local outlier factor of each data point in each cluster, and a number of normal data points and a number of outlier data points in each cluster includes the following specific steps: In the formula, is the fluctuation degree of the target cluster, is the minimum value of the fluctuation degree among all clusters, is the maximum value of the fluctuation degree among all clusters, is the number of outlier data points in the target cluster, is the average distance between the gth outlier data point in the target cluster and all normal data points in the target cluster, is the local outlier factor of the g-th outlier data point in the target cluster, is the mean of the local outlier factors of all normal data points in the target cluster, is the sinc function, To take the absolute value, is the abnormality degree of the target cluster.

2. According to the mold clamping state intelligent detection system of claim 1, It is characterized in that The specific steps of obtaining a number of normal data points and a number of outlier data points of each cluster according to the local outlier factors of the data points are as follows: A first threshold is preset, denoted as th1. If the local outlier factor of any data point in any cluster is less than or equal to th1, then the data point is a normal data point of the cluster; if the local outlier factor of any data point in any cluster is greater than th1, then the data point is an outlier data point of the cluster.

3. According to the mold clamping state intelligent detection system of claim 1, It is characterized in that The final abnormality degree of each non-anchor cluster is obtained according to the abnormality degree of the non-anchor cluster, the abnormality degree of the anchor cluster, and the distance between the center points of the non-anchor cluster and the anchor cluster. The specific steps include the following: Record any non-anchor cluster as the target non-anchor cluster; In the formula, is the number of anchor clusters in all clusters, is the abnormality degree of the target non-anchor cluster, is the abnormality degree of the kth anchor cluster, is the distance between the target non-anchor cluster and the center point of the kth anchor cluster, is the sigmoid function, is the final abnormality degree of the target non-anchor cluster.

4. According to claim 3, a mold clamping state intelligent detection system, It is characterized in that The specific steps of judging the mold clamping abnormality according to the final abnormality degree of the non-anchor point cluster are as follows: A second threshold is preset, denoted as th2, if , take the target non-anchor cluster as an abnormal cluster, and judge whether there is an abnormal cluster among all non-anchor clusters. If there is an abnormal cluster, the mold clamping is abnormal. If there is no abnormal cluster, the mold clamping is not abnormal.

5. According to the mold clamping state intelligent detection system of claim 1, It is characterized in that The specific steps of obtaining a plurality of anchor clusters and a plurality of non-anchor clusters according to the abnormality degree of each cluster are as follows: The time corresponding to the center point of each cluster is obtained in the pressure data, and the abnormality degrees of the corresponding clusters are arranged in the order of the time to obtain an abnormality degree sequence. The abnormality degree sequence is detected by the AMPD algorithm to obtain several minimum abnormality degrees of the abnormality degree sequence. The cluster corresponding to the minimum abnormality degree is recorded as the anchor cluster, and the cluster that is not the anchor cluster is recorded as the non-anchor cluster.

6. According to claim 1, a mold clamping state intelligent detection system, It is characterized in that The specific method for obtaining the local outlier factor of each data point in each cluster is as follows: Input any cluster of pressure data into the CBLOF algorithm, and output the local outlier factor of each data point in the cluster.

7. According to claim 1, a mold clamping state intelligent detection system, It is characterized in that The clustering of the pressure data to obtain a number of clusters of the pressure data includes the following specific steps: Iterative self-organizing clustering is performed on the pressure data, and the distance measurement adopts the Euclidean distance between data points in the pressure data to obtain several clusters of pressure data.

8. According to claim 1, a mold clamping state intelligent detection system, It is characterized in that The specific method for obtaining the pressure data is as follows: The mold is placed on the fixture for clamping and fixing, TH1 pressure sensors and the fixture are connected through supporting components, and pressure values ​​of multiple clamping positions are respectively collected through the pressure sensors. The pressure values ​​of multiple clamping positions are output every TH2 seconds, and all pressure values ​​collected within the last TH3 hours are arranged in chronological order to obtain a pressure data, TH1 is a preset first value, TH2 is a preset second value, and TH3 is a preset third value.