Method and system for dividing running state of industrial equipment based on data stream clustering

Abstract

This invention proposes a method and system for classifying the operating status of industrial equipment based on data stream clustering, relating to the field of industrial equipment operating status classification. The method includes: acquiring power consumption data of industrial equipment; extracting features from the power consumption data and preprocessing the extracted features; assigning dynamic weights to the preprocessed data and merging the preprocessed data into a micro-cluster structure, updating the micro-cluster structure; clustering the data in the micro-cluster structure using a density peak clustering method based on natural neighbors; and matching the operating status of the industrial equipment based on the clustering results. This invention employs a dynamic weighting method to more accurately describe and update the summary information of the data, and uses a density peak clustering algorithm based on natural neighbors to complete the final clustering task, thus improving the accuracy of determining the operating status of industrial equipment.

Description

Technical Field

[0001] This invention belongs to the field of industrial equipment operation status classification, and particularly relates to a method and system for industrial equipment operation status classification based on data flow clustering. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In recent years, the country has accelerated the optimization of its industrial structure and vigorously promoted the deep integration of industrialization and informatization. Currently, intelligentization and informatization have been integrated into various processes of industrial production, such as production, pollution control, and monitoring of wastewater discharge equipment. To promptly assess the operating status of industrial equipment, companies typically deploy sensors on the equipment to collect operating parameters, thereby determining its operational status and saving significant time and labor costs.

[0004] Currently, most companies use a fixed threshold method with a single parameter to determine the operating status of equipment. For example, a threshold range for the power parameter under normal operating conditions is predetermined. If the actual power is below the threshold, it indicates that the equipment is not operating; if it is above the threshold, it indicates that the equipment is malfunctioning. This method is simple in principle and can quickly identify the operating status of industrial equipment. However, in industrial data flow environments, static threshold methods may not be able to adapt to complex changes in operating conditions.

[0005] The inventors also discovered that using a single parameter cannot fully reflect the operating status of equipment, and may even lead to misjudgments of the operating status of industrial equipment. While some companies have implemented their own inference algorithms to determine the operating status of industrial equipment, most companies have not yet reached the level of using artificial intelligence technology for judgment.

[0006] The inventors also discovered that there are often correlations between different parameters collected by sensors, and analyzing multiple collected parameters can often improve the accuracy of equipment status assessment. With the continuous innovation of computer hardware and the development of sensor technology, the amount of industrial data is exploding, and the flow speed is extremely fast; the academic community refers to this type of data as data streams. When these industrial data streams are needed to determine equipment status, it is inevitable to classify, store, and analyze the data. However, this data cannot all be stored on storage media, and in most cases, its tagging information is unknown. Furthermore, the low value inherent in data streams makes tagging them a very costly and low-return task.

[0007] Therefore, how to store and analyze data streams, and at the same time, accurately classify the operating status of devices, is an urgent problem to be solved. Summary of the Invention

[0008] To overcome the shortcomings of the prior art, this invention provides a method and system for classifying the operating status of industrial equipment based on data flow clustering. It employs a two-stage online / offline data flow clustering framework for data processing. In the online stage, continuously arriving industrial electricity data is stored in a micro-cluster structure, and a dynamic weighting method is used to initially assign weights to the micro-cluster information and dynamically update it to adapt to the constantly changing industrial environment. In the offline stage, the micro-cluster information maintained in the online stage is clustered to obtain clustering results, thereby matching the operating status of the industrial equipment.

[0009] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0010] The first aspect of this invention provides a method for classifying the operating status of industrial equipment based on data flow clustering.

[0011] A method for classifying the operating status of industrial equipment based on data stream clustering includes:

[0012] Obtain power consumption data for industrial equipment;

[0013] Features are extracted from electricity consumption data and the extracted features are preprocessed.

[0014] Assign dynamic weights to the preprocessed data, merge the preprocessed data into the micro-cluster structure, and update the micro-cluster structure;

[0015] The density peak clustering method based on natural neighbors is used to cluster data in micro-cluster structures, and the operating status of industrial equipment is matched according to the clustering results.

[0016] The second aspect of the present invention provides an industrial equipment operation status classification system based on data flow clustering.

[0017] An industrial equipment operation status classification system based on data stream clustering includes:

[0018] The electricity consumption data acquisition module is configured to acquire electricity consumption data of industrial equipment.

[0019] The feature extraction and preprocessing module is configured to: extract features from electricity consumption data and preprocess the extracted features;

[0020] The online microcluster module is configured to: assign dynamic weights to the preprocessed data, merge the preprocessed data into the microcluster structure, and update the microcluster structure;

[0021] The offline clustering module is configured to cluster data in the micro-cluster structure using a density peak clustering method based on natural neighbors, and match the operating status of industrial equipment based on the clustering results.

[0022] A third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in the industrial equipment operating state partitioning method based on data flow clustering as described in the first aspect of the present invention.

[0023] The fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the industrial equipment operating state partitioning method based on data flow clustering as described in the first aspect of the present invention.

[0024] The above one or more technical solutions have the following beneficial effects:

[0025] This invention employs a dataflow clustering algorithm to perform unsupervised incremental processing of electrical parameter data from industrial equipment, thereby enabling the determination of the equipment's operating status. This invention improves upon the traditional two-stage dataflow clustering framework: in the online stage, a dynamic weighting method is proposed to more accurately describe and update the summary information of the data; in the offline stage, a density peak clustering algorithm based on natural neighbors is proposed to complete the final clustering task, improving the accuracy of determining the operating status of industrial equipment.

[0026] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0027] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0028] Figure 1 This is a flowchart of the method in the first embodiment.

[0029] Figure 2 This is a system structure diagram of the second embodiment.

[0030] Figures 3(a)-(c) are schematic diagrams of natural neighbors.

[0031] Figure 4 (a) is a schematic diagram of micro-clusters formed in the online phase based on the data flow clustering method.

[0032] Figure 4 (b) Clustering results generated during the offline phase. Detailed Implementation

[0033] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, 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 pertains.

[0034] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0035] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0036] The overall approach of this invention is as follows: Facing complex and ever-changing electrical parameter data from industrial equipment, this invention proposes a data flow clustering-based method for classifying the operating states of industrial equipment. The newly proposed method improves upon the traditional two-stage framework of data flow clustering. In the online stage, a dynamic weighting method is proposed. First, when assigning initial weights to data, the positional information between newly arrived data and existing data is considered, and the summary information of the data is continuously updated using the dynamic weighting method. Second, the decay of data weights over time is taken into account. In the offline stage, to handle clusters of arbitrary shapes, a density peak clustering algorithm based on natural neighbors is proposed, achieving adaptive and efficient clustering.

[0037] Example 1

[0038] This embodiment discloses a method for classifying the operating status of industrial equipment based on data flow clustering.

[0039] like Figure 1 As shown, the method for classifying the operating status of industrial equipment based on data flow clustering includes:

[0040] Obtain power consumption data for industrial equipment;

[0041] Features are extracted from electricity consumption data and the extracted features are preprocessed.

[0042] Assign dynamic weights to the preprocessed data, merge the preprocessed data into the micro-cluster structure, and update the micro-cluster structure;

[0043] The density peak clustering method based on natural neighbors is used to cluster data in micro-cluster structures, and the operating status of industrial equipment is matched according to the clustering results.

[0044] The following is a detailed description of the implementation steps of the present invention:

[0045] 1. Characteristic selection and preprocessing stage of industrial equipment power consumption data

[0046] Because the collected raw data has certain non-standardizations, such as duplication and redundancy, inconsistent attribute units, etc., and in addition to numerical data reflecting equipment power consumption, there are also some identifying data, such as equipment number and administrator number, which are unrelated to equipment operating status, it is necessary to adopt reasonable data preprocessing methods to standardize the raw data and select features from the raw data to extract those data features related to equipment operating status for subsequent research.

[0047] 1.1 Feature Extraction of Electricity Consumption Data from Industrial Equipment

[0048] First, feature selection is performed on the raw power consumption data of industrial equipment. Identifying attributes such as company number, equipment number, and operator number are removed from the raw data. Then, the following data features are selected: A-phase current (ia), B-phase current (ib), C-phase current (ic), total active power (pw), A-phase voltage (ua), B-phase voltage (ub), and C-phase voltage (uc) as data features for subsequent data mining.

[0049] 1.2 Preprocessing of Power Consumption Data for Industrial Equipment

[0050] First, duplicate and redundant data were removed from the original dataset. Simultaneously, to eliminate the influence of the numerical magnitude of features on the analysis results, the Z-score method was used to standardize the data after feature extraction.

[0051]

[0052] in Let s represent the mean of the data, s represent the standard deviation of the data, and x represent the mean of the data. i Let y represent the i-th data point, n be the number of data points, and y be the number of data points. i This represents the i-th standardized data.

[0053] 2. Dataflow clustering algorithm based on dynamic weights and density peaks

[0054] This stage uses preprocessed industrial equipment power consumption data to construct a data flow clustering model based on dynamic weights and density peaks to classify the operating states of industrial equipment. The algorithm of this invention comprises two parts: an online stage and an offline stage. The algorithm flow is described in detail below based on these two stages.

[0055] 2.1 Online Phase

[0056] 2.1.1 Attenuation Function and Microcluster Structure

[0057] In industrial systems, the most recent data objects often reflect the current operating status of equipment more accurately than historical objects. However, completely ignoring historical data during research is also unreasonable. To reduce the impact of historical data on current clustering analysis, it is necessary to attenuate the historical data. This invention uses an exponential decay function to simulate the attenuation of data streams, specifically:

[0058]

[0059] Where t0 represents the arrival time of the data sample, t1 represents the current time, and λ is the decay factor. The larger the value of λ, the faster the data decays.

[0060] Since data streams are theoretically infinite, their storage needs to be considered. This invention uses a micro-cluster data structure to store summary information of continuously arriving industrial electricity consumption data. The micro-cluster structure enables incremental data storage; data only needs to pass through once to store summary information in a micro-cluster. A micro-cluster contains information from multiple data points, which means the number of micro-clusters is much smaller than the number of data points, saving storage space and improving data processing speed.

[0061] The micro-cluster structure is further divided into potential core micro-clusters (PMC) and outlier micro-clusters (OMC). The data maintained by the potential core micro-clusters is used for clustering in the offline phase. A potential core micro-cluster is defined as follows:

[0062]

[0063] in, Let T represent the linear sum of all points in the microcluster; where t represents the current time, T i p represents the time when the data arrives. i This indicates information about newly arrived data points. Let represent the sum of squares of all points in the microcluster; the weight of the microcluster is represented as . Here, β is the outlier threshold, and μ is the integer weight of a given microcluster; both parameters are user-defined. It is important to note that the weight ω of a potential core microcluster must be greater than or equal to βμ.

[0064] The center c of the potential core microcluster is represented as:

[0065]

[0066] The radius r of the potential core microcluster is expressed as:

[0067]

[0068] Here, parameter ∈ is a radius threshold given by the user, and the radius of all microclusters must be less than or equal to ∈.

[0069] Outlier microclusters serve to temporarily store outliers; they are defined as follows:

[0070]

[0071] The definitions of all its parameters (CF1,CF2,ω,c,r) are the same as those of the potential core microcluster. It should be noted that the weight ω of the outlier microcluster must be lower than βμ.

[0072] 2.1.2 Merge Operation

[0073] like Figure 1 As shown, the process of the merge operation will be explained in detail below. When a new industrial equipment power parameter data object p arrives, the following steps are executed sequentially.

[0074] 1) First, find the PMC closest to p. The nearest PMC is defined by the distance from data point p to the center point of each PMC. This nearest PMC is denoted as PMCp.

[0075] 2) If the latest radius of the microcluster is less than or equal to the specified radius threshold after p is inserted into PMCp, then p is merged into the PMC, and the weight, radius and center of the microcluster are updated.

[0076] This invention proposes a dynamic weighting method that calculates the initial weight of the data based on the distance between p and the center of PMCp. This calculation method takes into account the location information of the data, describes the micro-cluster related information more accurately, and is beneficial for subsequent clustering processing of the data.

[0077]

[0078] Where dist(p,center) represents the Euclidean distance between point p and the center point of the micro-cluster to be merged.

[0079] Because microclusters can perform incremental data processing, their weights, radius, and center are updated synchronously. The weight update is defined by the following formula:

[0080] w PMCp =w PMCp +weight p (3)

[0081] Among them, w PMCp This represents the weight of PMCp.

[0082] 3) If the insertion fails, find the nearest OMC to p and denote it as OMCp.

[0083] 4) If the latest radius of the microcluster is less than or equal to the specified radius threshold after p is inserted into OMCp, then p will be merged into the OMCp, and its weight, radius and center will be updated using the same operation as when PMC is used.

[0084] 5) After successful insertion, the algorithm further checks the new weights of OMCp. If its weight is greater than βμ, it means that OMCp has been transformed into a PMC. Therefore, OMCp is removed from the outlier buffer, forming a new PMC.

[0085] 6) Otherwise, create a new OMC for p.

[0086] 2.1.3 Pruning and Renewal Operations

[0087] like Figure 1 As shown, pruning and updating operations ensure that information such as the weights of microclusters is updated in a timely manner. For potential core microclusters, if no new data arrives for a long time, their weights will continuously decay, and after a period of time, they may no longer meet the conditions for potential core microclusters. If outlier microclusters do not receive new data for a period of time, they are proven to be noise points. Therefore, pruning operations are needed for microclusters.

[0088] Meanwhile, some outlier microclusters may evolve into potential core microclusters as a large number of data points are merged. Therefore, this method stipulates that the weight of each microcluster be checked periodically, and then pruning and updating operations are performed on the microclusters.

[0089] This method calculates the shortest time for role changes to occur between microclusters, defined as T. p

[0090]

[0091] Every T p Over time, the weights of all microclusters are checked. For potential core microclusters, if their weight is below βμ, the potential core microcluster is downgraded to an outlier microcluster.

[0092] A weight threshold was defined for each outlier microcluster:

[0093]

[0094] Wherein, parameter ∈ is the radius threshold given by the user; t0 represents the time of microcluster construction, λ is the decay factor, and t represents the current time.

[0095] The formula above shows that the longer an outlier microcluster exists, the higher its minimum weight threshold becomes. If it falls below this threshold, the outlier microcluster will be deleted.

[0096] like Figure 4 (a) shows a schematic diagram of micro-clusters formed in the online phase based on the data flow clustering method.

[0097] 2.2 Offline Phase

[0098] In this stage, the present invention proposes a two-stage density peak clustering algorithm based on natural neighbors. The PMCs maintained online are treated as weighted pseudo-points with the micro-cluster centers as coordinates. These pseudo-points are processed to obtain the final clustering result, which matches the operating status of the equipment. The process of this stage is as follows:

[0099] 1) First, find the natural neighbors (nn) of each point. Natural neighbors are defined as the mutual neighbor relationships obtained by continuously expanding the neighborhood range. The neighborhood expansion stops when all points in the sample space except outliers have at least one natural neighbor. The size of the natural neighbor does not need to be specified in advance, and the neighbor relationships can be obtained adaptively.

[0100] Figure 3 is a schematic diagram of natural neighbors. This figure illustrates the process of obtaining natural neighbors. First, each point finds its nearest neighbor and then verifies the reverse neighbor relationship. If two data points are each other's nearest neighbors, a neighbor relationship is established. AC, DG, and EF establish neighbor relationships with each other in Figure 3(a). Then, the neighbor range is expanded, and the neighbor relationships shown in Figure 3(b) are obtained. It can be seen that all points except H have at least one neighbor. After expanding the neighborhood range again, point H still has no neighbors, therefore H is noise, and the natural neighbor search is complete.

[0101] 2) Calculate the local density of data points. This method stipulates that the local density of each point is composed of the weight of that point and the weights of its natural neighbors. The weight of the natural neighbors is determined by the distance from the neighbor to the point; the closer the distance, the greater the weight, and vice versa.

[0102] The formula for calculating local density is:

[0103]

[0104] Where, ω i Let ω be the weight of the point, d(i,j) be the Euclidean distance between points i and j, and ω be the weight of the point. j The weight of the natural neighbors.

[0105] 3) Calculate the distance to the parent point and the γ value of each data point. For each point, its parent point is the nearest point with the higher local density; the distance between them is the parent point distance. For the point with the highest local density in the space, its farthest distance to all other points in the sample space is defined as its parent point distance. Then, the local density of the data point is multiplied by the parent point distance to obtain the γ value of the data. Points with higher γ values ​​are more likely to be cluster centers. This method sorts the data points in descending order of γ values, and then selects c points from highest to lowest as cluster centers, assigning them different labels.

[0106]

[0107] Where, ρ i Let ρ represent the local density at point i. j Let ρ represent the local density at point j. max represents the maximum local density in the sample space, and dij represents the Euclidean distance between points i and j.

[0108] γ i =ρ i ×δ i (8)

[0109] 4) After assigning labels to the center point, assign labels to the non-center points.

[0110] This method proposes a two-stage assignment approach. First, a diffusion method is used to assign the label information of cluster centers to dense points among their natural neighbors. Dense points are defined as those with a large number of natural neighbors, forming the initial shape of each cluster. For the remaining points, the second step assigns them to the cluster of their parent point. If a parent point has not yet been assigned to a cluster, the search continues. Since the initial cluster shape has already been formed, the parent point can be found without multiple iterations. After obtaining the final clustering result, the operating status of industrial equipment can be matched. Figure 4 (b) shows the clustering results generated during the offline phase.

[0111] Example 2

[0112] This embodiment discloses an industrial equipment operation status classification system based on data flow clustering.

[0113] like Figure 2 As shown, the industrial equipment operation status classification system based on data flow clustering includes:

[0114] The electricity consumption data acquisition module is configured to acquire electricity consumption data of industrial equipment.

[0115] The feature extraction and preprocessing module is configured to: extract features from electricity consumption data and preprocess the extracted features;

[0116] The online microcluster module is configured to: assign dynamic weights to the preprocessed data, merge the preprocessed data into the microcluster structure, and update the microcluster structure;

[0117] The offline clustering module is configured to cluster data in the micro-cluster structure using a density peak clustering method based on natural neighbors, and match the operating status of industrial equipment based on the clustering results.

[0118] Example 3

[0119] The purpose of this embodiment is to provide a computer-readable storage medium.

[0120] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the industrial equipment operating state classification method based on data flow clustering as described in Embodiment 1 of this disclosure.

[0121] Example 4

[0122] The purpose of this embodiment is to provide an electronic device.

[0123] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the industrial equipment operating state classification method based on data flow clustering as described in Embodiment 1 of this disclosure.

[0124] The steps and methods involved in the apparatuses of Embodiments 2, 3, and 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0125] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0126] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for classifying the operating status of industrial equipment based on data flow clustering, characterized in that, include: Obtain power consumption data for industrial equipment; Features are extracted from electricity consumption data and the extracted features are preprocessed. Assign dynamic weights to the preprocessed data, merge the preprocessed data into the micro-cluster structure, and update the micro-cluster structure; The density peak clustering method based on natural neighbors is used to cluster data in micro-cluster structures, and the operating status of industrial equipment is matched according to the clustering results. The dynamic weights are specifically as follows: in, express p The Euclidean distance between the point and the center point of the microcluster structure to be merged, where p is the preprocessed data.

2. The method for classifying the operating status of industrial equipment based on data flow clustering as described in claim 1, characterized in that, The features extracted from the electricity consumption data are as follows: A-phase current, B-phase current, C-phase current, total active power, A-phase voltage, B-phase voltage, and C-phase voltage are extracted as data features.

3. The method for classifying the operating status of industrial equipment based on data flow clustering as described in claim 1, characterized in that, The extracted features are preprocessed, specifically by removing duplicate data and standardizing the extracted data using the Z-score method. in, Let s represent the mean of the data, s represent the standard deviation of the data, and x represent the mean of the data. i This represents the i-th data.

4. The method for classifying the operating status of industrial equipment based on data flow clustering as described in claim 1, characterized in that, The microcluster structure includes potential core microclusters and outlier microclusters.

5. The method for classifying the operating status of industrial equipment based on data flow clustering as described in claim 4, characterized in that, Assign dynamic weights to the preprocessed data, merge the preprocessed data into the micro-cluster structure, and update the micro-cluster structure, specifically as follows: 1) Find the closest p The potential core microcluster, denoted as PMCp ; 2) p Insert into PMCp ,judge PMCp If the latest radius is less than or equal to the specified radius threshold, then p is merged into the potential core micro-cluster, and the weight, radius, and center of the micro-cluster are updated. The weight update formula is as follows: Among them, w PMCp express PMCp The weights; If not, proceed to step 3); 3) Find the distance p The most recent outlier microcluster, denoted as OMCp ; 4) p Insert into OMCp ,judge OMCp Is the latest radius less than or equal to the specified radius threshold? If so, then... p Merge into OMCp In progress, updates OMCp The weights, radius, and center are determined, and then step 5 is executed. If not, proceed to step 6); 5) Inspection OMCp If the new weight is greater than the set value, then remove it. OMCp This forms a new potential core microcluster; 6) For p Create a new OMC.

6. The method for classifying the operating status of industrial equipment based on data flow clustering as described in claim 3, characterized in that, The updated micro-cluster structure is clustered using a density peak clustering method based on natural neighbors, specifically as follows: 1) Treat each potential core micro-cluster as a data point and find the natural neighbors of each data point; 2) Calculate the local density of the data points. The local density is composed of the weight of the point and the weights of its natural neighbors: Where, ω i Let ω be the weight of the point, d(i,j) be the Euclidean distance between points i and j, and ω be the weight of the point. j The weight of natural neighbors; 3) Calculate the distance to the parent point of each data point and the γ value. Sort the data points in descending order of their γ values, select a set number of points from highest to lowest as center points, and assign different labels to them. in, Represents the local density at point i. Represents the local density at point j. The maximum local density is represented by , and dij represents the Euclidean distance between points i and j. 4) Assign non-center points to obtain the final clustering results.

7. An industrial equipment operation status classification system based on data flow clustering, characterized in that: include: The electricity consumption data acquisition module is configured to acquire electricity consumption data of industrial equipment. The feature extraction and preprocessing module is configured to: extract features from electricity consumption data and preprocess the extracted features; The online microcluster module is configured to: assign dynamic weights to the preprocessed data, merge the preprocessed data into the microcluster structure, and update the microcluster structure; The offline clustering module is configured to: cluster data in the micro-cluster structure using the density peak clustering method based on natural neighbors, and match the operating status of industrial equipment based on the clustering results; The dynamic weights are specifically as follows: in, express p The Euclidean distance between the point and the center point of the microcluster structure to be merged, where p is the preprocessed data.

8. A computer-readable storage medium having a program stored thereon, characterized in that, When executed by the processor, the program implements the steps in the industrial equipment operating state classification method based on data flow clustering as described in any one of claims 1-6.

9. An electronic device, comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the industrial equipment operating state classification method based on data flow clustering as described in any one of claims 1-6.

Images