A method and system for monitoring pressure data of a compressor sealing system

Abstract

The present invention relates to the technical field of data processing, and particularly to a method and system for monitoring pressure data of a compressor sealing system. The method includes the steps of: collecting pressure data, clustering the pressure data using the DIANA algorithm, taking any cluster in any round of clustering as the current cluster, obtaining the overall distance within the current cluster and the local distance within the current cluster; based on the overall distance and the local distance, obtaining the degree of abnormality of the current cluster; obtaining the fluctuation amplitude of the current cluster, and according to the fluctuation amplitude of the cluster and the degree of abnormality of the cluster, obtaining the splitting threshold of the cluster; clustering the pressure data according to the method for obtaining the splitting threshold of the cluster to obtain the final clustering result, and performing anomaly monitoring on the pressure data of the compressor sealing system according to the final clustering result; in the clustering process of the present invention, multiple clusters are split simultaneously, thereby improving the sensitivity and accuracy of anomaly detection.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular, to a method and system for monitoring pressure data of a compressor sealing system. Background Art

[0002] As a key device in many industrial processes, the normal operation of a compressor is crucial for ensuring production efficiency and safety. The integrity of the sealing system inside the compressor directly affects the operating state and safety of the compressor. Once the sealing system fails, it may lead to gas leakage, reduced efficiency, and even safety accidents. By monitoring the pressure data of the sealing system, problems existing in the sealing system, such as leakage and blockage, can be discovered and solved, thereby improving the operating efficiency of the compressor.

[0003] Xu Kesheng, Journal of Dalian Jiaotong University, December 2007, Volume 28, Issue 4. In this journal, a clustering algorithm for automatically obtaining the k value was proposed. The DIANA algorithm in this journal is a hierarchical clustering algorithm that adopts a top-down strategy. First, all objects are placed in one class, and then they are gradually divided into smaller and smaller classes until each object forms a class by itself, and the algorithm stops when the termination condition is reached.

[0004] Since the pressure data of a compressor is often affected by factors such as environmental changes, operating conditions, and load changes, this means that the pressure value may change significantly over time, resulting in different characteristics of the pressure data distribution in different time periods. As a hierarchical clustering algorithm, the DIANA algorithm forms a hierarchical tree structure by repeatedly splitting clusters. This splitting method enables the algorithm to carefully analyze the changes in pressure data, whether due to normal random fluctuations in the system or abnormal fluctuations caused by external disturbances. However, the DIANA algorithm only selects one cluster for splitting each time, so a single split may cause abnormal data to be submerged, thereby delaying the detection and response time of system failures and increasing the risk of equipment damage. Summary of the Invention

[0005] To solve the technical problem that the DIANA algorithm only selects one cluster for splitting each time, and a single split may cause abnormal data to be submerged, thereby delaying the detection and response time of system failures and increasing the risk of equipment damage, the present invention provides a method and system for monitoring pressure data of a compressor sealing system.

[0006] In the first aspect, the present invention provides a method for monitoring pressure data of a compressor sealing system, adopting the following technical solution:

[0007] A method for monitoring pressure data of a compressor sealing system includes the steps:

[0008] Collect pressure data; use the DIANA algorithm to cluster the pressure data, and denote any cluster obtained from any round of clustering as the current cluster; obtain the abnormality degree of the current cluster, where the abnormality degree is the ratio of the overall distance within the current cluster to the local distance within the current cluster; obtain the fluctuation amplitude of the current cluster;

[0009] Obtain the splitting threshold of the current cluster , represents the fluctuation amplitude of the current cluster; represents the abnormality degree of the current cluster; represents the first hyperparameter; represents the second hyperparameter; exp() represents the exponential function with the natural constant as the base; according to the method for obtaining the splitting threshold, perform cluster splitting on the pressure data to obtain the final clustering result; perform anomaly monitoring on the pressure data of the compressor sealing system according to the final clustering result.

[0010] The innovation of the present invention lies in improving the DIANA algorithm. During each round of clustering, it is judged whether to classify the cluster according to the splitting threshold of the cluster. By setting different splitting thresholds for different clusters, the splitting strategy can be flexibly adjusted according to the characteristics of each cluster, thereby enhancing the detection ability for abnormal situations. It can effectively reduce false alarms and missed alarms caused by over-splitting or under-splitting, thereby improving the reliability of the monitoring system and enabling operators to more accurately judge the state of the system.

[0011] Preferably, the steps for obtaining the overall distance within the current cluster are as follows:

[0012] ;

[0013] In the formula, Q represents the overall distance within the current cluster; represents the number of data in the current cluster; represents the standard deviation of the data in the current cluster; represents the Euclidean distance between the i-th data in the current cluster and the central data of the current cluster; represents the interquartile range of the data in the current cluster; represents the maximum data value in the current cluster; represents the minimum data value in the current cluster.

[0014] The larger the overall distance within the cluster, the more dispersed the data in the cluster, that is, the more abnormal the cluster. Therefore, the abnormality degree of the cluster can be obtained according to the overall distance within the cluster.

[0015] Preferably, the steps for obtaining the local distance within the current cluster are as follows:

[0016] ;

[0017] In the formula, represents the local distance within the current cluster; represents the number of data in the current cluster; represents the average of the Euclidean distances between the i-th data in the current cluster and all its neighboring points.

[0018] It is convenient to subsequently obtain the abnormality degree of the cluster based on the local distance within the cluster.

[0019] Preferably, the obtaining of the fluctuation amplitude of the current cluster includes:

[0020] Obtaining the fluctuation characteristics of each data in the current cluster;

[0021] ;

[0022] In the formula, represents the fluctuation amplitude of the current cluster; represents the number of data in the current cluster; represents the fluctuation characteristic of the i-th data in the current cluster; represents the maximum value of the fluctuation characteristics of all data in the current cluster; represents the minimum value of the fluctuation characteristics of all data in the current cluster.

[0023] The larger the fluctuation amplitude of the cluster, the more it indicates that the cluster needs to be split, which is convenient to subsequently obtain the splitting threshold of the cluster based on the fluctuation amplitude of the cluster.

[0024] Preferably, the obtaining of the fluctuation characteristics of each data in the current cluster includes:

[0025] Obtaining the Euclidean distance between the central data of the current cluster and the central data of each other cluster, and denoting the cluster corresponding to the minimum Euclidean distance as the nearest neighbor cluster of the current cluster;

[0026] Taking the distance between the i-th data in the current cluster and the central data of the current cluster and the distance between the i-th data in the current cluster and the central data of its nearest neighbor cluster as the fluctuation characteristic vector of the i-th data in the current cluster, and taking the modulus of the fluctuation characteristic vector of the i-th data in the current cluster as the fluctuation characteristic of the i-th data in the current cluster.

[0027] It is convenient to obtain the fluctuation amplitude of the cluster based on the fluctuation characteristics of each data in the cluster.

[0028] Preferably, the clustering and splitting of the pressure data according to the obtaining method of the splitting threshold to obtain the final clustering result includes:

[0029] Preset the clustering round T. According to the clustering round, use the DIANA algorithm to cluster the pressure data. In any round of clustering, obtain the variance of each cluster and the splitting threshold of each cluster. If the variance of any cluster in this round of clustering is greater than or equal to the splitting threshold of this cluster in this round of clustering, select the data farthest from the cluster center in this cluster of this round of clustering as the splitting point of this cluster of this round of clustering, split this cluster of this round of clustering into two sub-clusters, obtain the clustering result of this round of clustering, and record the clustering result of the last round as the final clustering result.

[0030] Preferably, the abnormal monitoring of the pressure data of the compressor sealing system according to the final clustering result includes:

[0031] Preset the number of data , if the number of data in any cluster in the final clustering result is less than the preset number of data At this time, this cluster in the final clustering result is an abnormal cluster. At this time, trigger the warning mechanism and remind relevant personnel to handle it, so as to realize the monitoring of the pressure data of the compressor sealing system.

[0032] In a second aspect, the present invention provides a pressure data monitoring system for a compressor sealing system, adopting the following technical solution:

[0033] A pressure data monitoring system for a compressor sealing system includes: a processor and a memory. The memory stores computer program instructions. When the computer program instructions are executed by the processor, the above-mentioned pressure data monitoring method for a compressor sealing system is implemented.

[0034] By adopting the above technical solution, the above-mentioned pressure data monitoring method for a compressor sealing system is generated into a computer program and stored in the memory to be loaded and executed by the processor, so as to manufacture a terminal device according to the memory and the processor, which is convenient to use.

[0035] The present invention has the following technical effects: The purpose of the present invention is to improve the DIANA algorithm. In each round of its clustering process, it judges whether to classify a cluster according to the splitting threshold of the cluster. By setting different splitting thresholds for different clusters, the splitting strategy can be flexibly adjusted according to the characteristics of each cluster, thereby enhancing the detection ability for abnormal situations. It can effectively reduce false alarms and missed alarms caused by excessive or insufficient splitting, thereby improving the reliability of the monitoring system, enabling operators to more accurately judge the state of the system. Further, the acquisition of the splitting threshold of the cluster takes into account the overall distance within the cluster and the local distance within the cluster, making the obtained splitting threshold of the cluster more accurate. Description of the Drawings

[0036] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features, and advantages of the exemplary embodiments of the present invention will become readily understood. In the drawings, several embodiments of the present invention are shown in an exemplary but not restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts.

[0037] Figure 1 It is a flowchart of a method for monitoring pressure data in a compressor sealing system according to an embodiment of the present invention. Specific embodiments

[0038] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0039] It should be understood that when terms such as "first" and "second" are used in the claims, specifications, and drawings of the present invention, they are only used to distinguish different objects and not to describe a specific order. The terms "comprising" and "including" used in the specifications and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0040] An embodiment of the present invention discloses a method for monitoring pressure data of a compressor sealing system. Referring to Figure 1 , it includes steps S1 - S3:

[0041] S1: Collect pressure data.

[0042] In the embodiment of the present invention, a pressure sensor is installed inside the compressor sealing system. Each sampling moment is two minutes, and pressure data is collected for a total of two hours to obtain all pressure data.

[0043] S2: Use the DIANA algorithm to cluster the pressure data. Denote any cluster in any round of clustering as the current cluster. Obtain the overall distance within the current cluster and the local distance within the current cluster. According to the overall distance within the current cluster and the local distance within the current cluster, obtain the degree of abnormality of the current cluster; obtain the fluctuation amplitude of the current cluster.

[0044] It should be noted that since the pressure data of the compressor is often affected by factors such as environmental changes, operating conditions, and load changes, this means that the pressure value may change significantly over time, resulting in different characteristics of the pressure data distribution in different time periods. As a hierarchical clustering algorithm, the DIANA algorithm forms a hierarchical tree structure by repeatedly splitting clusters. This splitting method enables the algorithm to carefully analyze the changes in pressure data, whether due to normal random fluctuations in the system or abnormal fluctuations caused by external disturbances. However, the DIANA algorithm only selects one cluster for splitting each time. Therefore, a single split may cause abnormal data to be submerged, thus delaying the detection and response time of system failures and increasing the risk of equipment damage. Therefore, to improve the DIANA algorithm, during each round of clustering process, multiple clusters need to be selected for splitting to ensure the accuracy of the clustering results, thereby enhancing the sensitivity and accuracy of anomaly detection. When selecting clusters, it is necessary to judge whether to split the clusters according to the splitting threshold of the clusters to ensure the accuracy of the clustering results, thereby improving the sensitivity and accuracy of anomaly monitoring.

[0045] Step S2 includes steps S20 - S21, which are specifically as follows:

[0046] S20: Use the DIANA algorithm to cluster the pressure data. Denote any cluster in any round of clustering as the current cluster, obtain the overall distance within the current cluster and the local distance within the current cluster, and obtain the degree of abnormality of the current cluster based on the overall distance within the current cluster and the local distance within the current cluster.

[0047] It should be noted that when a cluster has a high degree of abnormality, it often means that there are noises or outliers within the cluster, which will reduce the overall quality of the cluster. It is necessary to perform corresponding splitting to ensure that the data within each sub - cluster is more consistent, so as to better isolate the abnormal data. At the same time, the fluctuation amplitude of the data in the cluster is a key indicator to measure the data fluctuation in the cluster. A larger fluctuation amplitude means that the data change amplitude in the cluster is larger, which may indicate the existence of potential anomalies or changes. Monitoring the fluctuation amplitude of the cluster helps to identify potential anomalies in the compressor sealing system. Therefore, subsequently, based on the degree of abnormality and the fluctuation amplitude of the cluster, comprehensively analyze the necessity of splitting the cluster, and obtain the splitting threshold of the cluster to judge whether the cluster needs to be split, so as to ensure that each cluster can contain more similar samples, thereby improving the effect and accuracy of clustering, and facilitating the identification and monitoring of abnormal data.

[0048] It should be noted that when obtaining the degree of abnormality within a cluster, it is often measured based on the average Euclidean distance from all the data in the cluster to the central data of the cluster. This distance provides global information about the cluster, indicating the overall dispersion degree of the cluster. The larger its value, the more dispersed the data in the cluster, and it may contain outliers. Therefore, first, the overall distance within the cluster is obtained based on this feature. However, in the actual data distribution of the cluster, there may be some data that seems normal as a whole but shows abnormalities locally. At this time, it is necessary to combine the average of the Euclidean distances between each data and all its neighboring points in the cluster to comprehensively evaluate the local distance within the cluster to improve the accuracy of anomaly detection. Finally, based on the overall distance and the local distance within the cluster, the degree of abnormality of the cluster is obtained.

[0049] In the embodiment of the present invention, the DIANA algorithm is used to cluster the pressure data. Any cluster obtained from any round of clustering is denoted as the current cluster, and the overall distance within the current cluster is obtained:

[0050] ;

[0051] In the formula, Q represents the overall distance within the current cluster; represents the number of data in the current cluster; represents the standard deviation of the data in the current cluster; represents the Euclidean distance between the i-th data in the current cluster and the central data of the current cluster; represents the interquartile range of the data in the current cluster; represents the maximum data value in the current cluster; represents the minimum data value in the current cluster; represents the average Euclidean distance from all the data in the current cluster to the central data of the current cluster. Here, serves to standardize the Euclidean distance from all the data in the current cluster to the central data of the current cluster. The larger the average Euclidean distance from all the data in the current cluster to the central data of the current cluster, the larger the overall distance within the current cluster; The larger the value of, the more concentrated the distribution of the data in the current cluster, indicating that the overall distance within the current cluster is smaller; is the range of the data in the current cluster, and the larger its value, the larger the overall distance within the current cluster.

[0052] In the embodiment of the present invention, the local distance within the current cluster is obtained:

[0053] ;

[0054] In the formula, represents the local distance within the current cluster; represents the number of data in the current cluster; represents the mean of the Euclidean distances between the \(i\)-th data in the current cluster and all its neighboring points; the greater the mean of the Euclidean distances between each data in the current cluster and all its neighboring points, the greater the local distance within the current cluster.

[0055] Take the ratio of the overall distance within the current cluster to the local distance within the current cluster as the degree of abnormality of the current cluster. It should be noted that when the overall distance within the current cluster is relatively large compared to the local distance within the current cluster, it indicates that there is a relatively large abnormal dispersion phenomenon within the current cluster, and the degree of abnormality of the current cluster is higher.

[0056] S21: Obtain the fluctuation amplitude of the current cluster.

[0057] It should be noted that when the distance from any data in the cluster to the center data of the cluster and the distance from this data in the cluster to the center data of its nearest neighboring cluster are larger, it indicates the fluctuation characteristics of this data in the cluster. If the mean of the fluctuation characteristics of all data in the cluster is larger, it indicates that the fluctuation amplitude of the cluster is larger.

[0058] In the invention embodiment, obtain the Euclidean distance between the center data of the current cluster and the center data of each other cluster, and denote the cluster corresponding to the minimum Euclidean distance as the nearest neighboring cluster of the current cluster.

[0059] Take the distance from the \(i\)-th data in the current cluster to the center data of the current cluster and the distance from the \(i\)-th data in the current cluster to the center data of its nearest neighboring cluster as the fluctuation feature vector of the \(i\)-th data in the current cluster, and take the modulus of the fluctuation feature vector of the \(i\)-th data in the current cluster as the fluctuation feature of the \(i\)-th data in the current cluster; similarly, obtain the fluctuation feature of each data in the current cluster.

[0060] Obtain the fluctuation amplitude of the current cluster:

[0061] ;

[0062] In the formula, represents the fluctuation amplitude of the current cluster; represents the number of data in the current cluster; represents the fluctuation feature of the \(i\)-th data in the current cluster; represents the maximum value of the fluctuation features of all data in the current cluster; represents the minimum value of the fluctuation features of all data in the current cluster; the greater the mean of the fluctuation features of all data in the current cluster, the greater the fluctuation amplitude of the current cluster.

[0063] S3: Obtain the splitting threshold of the current cluster according to the fluctuation amplitude and the abnormality degree of the current cluster; perform clustering splitting on the pressure data according to the splitting threshold of the current cluster to obtain the final clustering result, and perform abnormality monitoring on the pressure data of the compressor sealing system according to the final clustering result.

[0064] Step S3 includes steps S30 - S31, which are specifically as follows:

[0065] S30: Obtain the splitting threshold of the current cluster according to the fluctuation amplitude and the abnormality degree of the current cluster.

[0066] It should be noted that the greater the abnormality degree of the known cluster, the smaller the splitting threshold needs to be set for this cluster to ensure its splitting and better highlight the abnormal points; and the greater the fluctuation amplitude of the cluster, it means that there may be potential abnormalities in the cluster and may also contain different data change trends, which requires more detailed splitting to identify and manage these changes, and a smaller splitting threshold also needs to be set. Therefore, according to the abnormality degree and the fluctuation amplitude of the cluster, the splitting threshold of the cluster is obtained.

[0067] In the embodiment of the present invention, the splitting threshold of the current cluster is obtained:

[0068] ;

[0069] wherein, represents the splitting threshold of the current cluster; represents the fluctuation amplitude of the current cluster; represents the abnormality degree of the current cluster; represents the first hyperparameter; represents the second hyperparameter; represents the abnormality degree weight of the current cluster; represents the fluctuation amplitude weight of the current cluster; since the abnormality degree of the current cluster can directly reflect the situation of abnormal data in the current cluster, and the fluctuation amplitude of the current cluster mainly reflects whether there are potential abnormalities in the current cluster, therefore, the acquisition of the splitting threshold of the current cluster mainly depends on the adjustment by the abnormality degree of the current cluster, and the fluctuation amplitude of the current cluster is only to avoid some potential abnormalities being ignored. It is necessary to comprehensively analyze and obtain the splitting threshold of the current cluster in combination with the fluctuation amplitude of the current cluster to ensure the accuracy of the final result. Therefore, in the embodiment of the present invention, it is preset , ; exp() represents the exponential function with the natural constant as the base; if the degree of abnormality of the current cluster is greater, a smaller splitting threshold needs to be set for the current cluster to ensure that the cluster is split and the abnormal data is better highlighted; if the fluctuation range of the current cluster is greater, it means that there may be potential abnormalities or different data change trends in the current cluster, and a smaller splitting threshold needs to be set for the current cluster to identify and manage these changes.

[0070] S31: According to the splitting threshold of the current cluster, perform clustering splitting on the pressure data to obtain the final clustering result, and perform anomaly monitoring on the pressure data of the compressor sealing system according to the final clustering result.

[0071] It should be noted that the known traditional DIANA algorithm usually selects the cluster with the largest inter-class variance for splitting, and this single splitting may cause abnormal data to be submerged, thus delaying the detection and response time of faults and increasing the risk of equipment damage. Therefore, the DIANA algorithm is improved. In each round of its clustering process, multiple clusters need to be selected for splitting to ensure the accuracy of the clustering result, thereby improving the sensitivity and accuracy of anomaly detection. When selecting a cluster, it is necessary to judge whether to classify the cluster according to the splitting threshold of the cluster to ensure the accuracy of the clustering result, and thus the sensitivity and accuracy of anomaly monitoring.

[0072] In the embodiment of the present invention, a preset clustering round T is set, and the DIANA algorithm is used to cluster the pressure data. In any round of clustering, the variance of each cluster and the splitting threshold of each cluster are obtained. If the variance of any cluster in this round of clustering is greater than or equal to the splitting threshold of this cluster in this round of clustering, the data farthest from the center of this cluster in this cluster of this round of clustering is selected as the splitting point of this cluster in this round of clustering, and this cluster in this round of clustering is split into two sub-clusters to obtain the clustering result of this round of clustering, and the clustering result of the last round is recorded as the final clustering result.

[0073] Preset the number of data , if the number of data in any cluster in the final clustering result is less than the preset number of data at this time, this cluster in the final clustering result is an abnormal cluster, that is, there is an abnormality in the pressure data of the compressor sealing system. At this time, the warning mechanism is triggered to remind relevant personnel to handle it to realize the monitoring of the pressure data of the compressor sealing system.

[0074] In the embodiment of the present invention, the preset clustering round T = 10 and the preset number of data Q = 10. In other embodiments, the implementer can preset the values of the clustering round and the number of data according to the specific implementation method.

[0075] An embodiment of the present invention also discloses a pressure data monitoring system for a compressor sealing system, which includes a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, a pressure data monitoring method for a compressor sealing system according to the present invention is implemented.

[0076] The above system also includes other components well-known to those skilled in the art, such as a communication bus and a communication interface. Their settings and functions are known in the art, so they will not be described in detail here.

[0077] In the present invention, the aforementioned memory can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, device, or component. For example, a computer-readable storage medium can be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory, a dynamic random access memory, a static random access memory, an enhanced dynamic random access memory, a high-bandwidth memory, a hybrid storage cube, etc., or any other medium that can be used to store the required information and can be accessed by an application program, a module, or both. Any such computer storage medium can be a part of the device or accessible or connectable to the device.

[0078] Although this specification has shown and described multiple embodiments of the present invention, it is obvious to those skilled in the art that such embodiments are provided only by way of example. Those skilled in the art will think of many changes, alterations, and alternative ways without departing from the spirit and idea of the present invention. It should be understood that various alternative solutions to the embodiments of the present invention described herein can be adopted in the process of practicing the present invention.

[0079] The above are all preferred embodiments of the present invention, and the protection scope of the present invention is not limited thereby. Therefore, all equivalent changes made according to the structure, shape, and principle of the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for monitoring pressure data of a compressor sealing system, characterized in that, Including the steps: Collect pressure data; cluster the pressure data using the DIANA algorithm, and denote any cluster in any round of clustering as the current cluster; obtain the abnormality degree of the current cluster, where the abnormality degree is the ratio of the overall distance within the current cluster to the local distance within the current cluster; obtain the fluctuation amplitude of the current cluster; Obtain the splitting threshold of the current cluster , represents the fluctuation amplitude of the current cluster; represents the abnormality degree of the current cluster; represents the first hyperparameter; represents the second hyperparameter; exp() represents the exponential function with the natural constant as the base; according to the method for obtaining the splitting threshold, perform clustering splitting on the pressure data to obtain the final clustering result; perform anomaly monitoring on the pressure data of the compressor sealing system according to the final clustering result; The steps for obtaining the overall distance within the current cluster are: ； Where Q represents the overall distance within the current cluster; represents the number of data in the current cluster; represents the standard deviation of the data in the current cluster; represents the Euclidean distance between the i-th data and the central data in the current cluster; represents the interquartile range of the data in the current cluster; represents the maximum data value in the current cluster; represents the minimum data value in the current cluster; The steps for obtaining the local distance within the current cluster are: ； In the formula, represents the local distance within the current cluster; represents the mean of the Euclidean distances between the $i$-th data in the current cluster and all its neighboring points.

2. The pressure data monitoring method of a compressor sealing system according to claim 1, characterized in that, The obtaining of the fluctuation amplitude of the current cluster includes: Obtain the fluctuation characteristics of each data in the current cluster; ； In the formula, represents the fluctuation amplitude of the current cluster; represents the number of data in the current cluster; represents the fluctuation feature of the i-th data in the current cluster; represents the maximum value of the fluctuation features of all data in the current cluster; represents the minimum value of the fluctuation features of all data in the current cluster.

3. The pressure data monitoring method of a compressor sealing system according to claim 2, wherein The obtaining of the fluctuation characteristics of each data in the current cluster includes: Obtain the Euclidean distance between the central data of the current cluster and the central data of each other cluster, and denote the cluster corresponding to the minimum Euclidean distance as the nearest neighbor cluster of the current cluster; Take the distance from the i-th data in the current cluster to the central data of the current cluster and the distance from the i-th data in the current cluster to the central data of its nearest neighbor cluster as the fluctuation feature vector of the i-th data in the current cluster, and take the modulus of the fluctuation feature vector of the i-th data in the current cluster as the fluctuation characteristic of the i-th data in the current cluster.

4. The pressure data monitoring method of a compressor sealing system according to claim 1, characterized in that, According to the obtaining method of the splitting threshold, perform clustering splitting on the pressure data to obtain the final clustering result, including: Preset the clustering round T. According to the clustering round, use the DIANA algorithm to cluster the pressure data. In any round of clustering, obtain the variance of each cluster and the splitting threshold of each cluster. If the variance of any cluster in this round of clustering is greater than or equal to the splitting threshold of this cluster in this round of clustering, select the data farthest from the center of this cluster in this cluster as the splitting point of this cluster in this round of clustering, split this cluster in this round of clustering into two sub-clusters to obtain the clustering result of this round of clustering, and denote the clustering result of the last round as the final clustering result.

5. A method for monitoring pressure data of a compressor sealing system according to claim 1, characterized in that, The abnormal monitoring of the pressure data of the compressor sealing system according to the final clustering result includes: Predetermined number of data , if the number of data in any cluster in the final clustering result is less than the predetermined number of data , in this case, the cluster in the final clustering result is an abnormal cluster. At this time, the warning mechanism is triggered to remind relevant personnel to handle it, so as to realize the pressure data monitoring of the compressor sealing system.

6. A pressure data monitoring system for a compressor sealing system, characterized in that, Including: A processor and a memory, where the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a method for monitoring the pressure data of a compressor sealing system according to any one of claims 1-5 is implemented.

Images