Distributed storage method for heterogeneous data of power CPS

Abstract

The present application relates to the technical field of computer data storage, and specifically to a distributed storage method serving heterogeneous data of electric power CPS, the method comprising: using a clustering algorithm to determine the relatively high and low relationship clusters of each front-end device; based on the difference between the maximum values ​​in the high relationship values ​​of all front-end devices and the discreteness of the heterogeneous data, determining the difference threshold of the electric power CPS, and combining the average distribution of all elements in the relatively low relationship cluster of each front-end device and the high relationship value, obtaining the electric power high relationship cluster of each front-end device; based on the total number of elements in the electric power high relationship cluster, determining the dynamic virtual node number of the electric power CPS, and storing the electric power data of all front-end devices of the electric power CPS. The present application improves the utilization rate of storage space and the uniformity of data distribution by dynamically allocating the number of virtual nodes.

Description

Technical Field

[0001] The present application relates to the technical field of computer data storage, and in particular to a distributed storage method for serving heterogeneous data of power CPS. Background Art

[0002] Power CPS, or Cyber-Physical Systems for Power, is a complex system that deeply integrates physical power grids and information networks. By integrating computing, communication, and control technologies, it enables real-time perception, dynamic control, and information services for the power system, thereby improving its efficiency, reliability, and safety.

[0003] Heterogeneous data refers to data with different structures and characteristics, coming from various sources, such as power generation, transmission, transformation, distribution, consumption, and dispatch. This data may include structured data (such as tabular data in a database), semi-structured data (such as data in XML or JSON format), and unstructured data (such as text files, images, and videos).

[0004] Traditional technologies for distributing heterogeneous data in power CPS systems typically use consistent hashing algorithms for distributed storage. To ensure uniform distribution of heterogeneous data, consistent hashing algorithms introduce virtual nodes. However, due to the significant similarities in heterogeneous power CPS data, such as operational variations in power lines, hash values ​​for heterogeneous data may be concentrated in the same area when calculating hash values, resulting in a more concentrated data distribution. Excessive or insufficient virtual nodes in the algorithm reduces storage space utilization and data distribution uniformity. Summary of the Invention

[0005] In order to solve the above technical problems, this application provides a distributed storage method for serving power CPS heterogeneous data to solve existing problems.

[0006] The distributed storage method for heterogeneous data of power CPS in this application adopts the following technical solutions:

[0007] One embodiment of the present application provides a distributed storage method for serving heterogeneous data of power CPS, the method comprising the following steps:

[0008] In the power CPS, the power data of each front-end device at all collection moments within a preset time period is collected, and the binary data of all power data are combined into a power vector;

[0009] Determine the power similarity between each front-end device and any other front-end device based on the difference in power vector length between each front-end device and any other front-end device, and the similarity of the power vectors;

[0010] Based on the power similarity between each front-end device and all other front-end devices, a clustering algorithm is used to determine the relatively high and low relationship clusters of each front-end device; based on the distribution of all elements in the relatively high relationship cluster of each front-end device, the high relationship value of each front-end device is determined; based on the dispersion degree of the high relationship values ​​of all front-end devices, the dispersion degree of the heterogeneous data of the power CPS is determined;

[0011] Based on the difference between the maximum values ​​in the high relationship values ​​of all front-end devices and the dispersion of the heterogeneous data, a difference threshold of the power CPS is determined; based on the average distribution of all elements in the relatively low relationship cluster of each front-end device and the high relationship value, the difference of each front-end device is determined, and the relatively high and low relationship clusters are merged in combination with the difference threshold to obtain a high power relationship cluster for each front-end device;

[0012] Based on the total number of elements in the power high-relationship cluster, the number of dynamic virtual nodes of the power CPS is determined, and the power data of all front-end devices of the power CPS is stored.

[0013] Preferably, the method for determining the power similarity between each front-end device and any other front-end device is:

[0014] In the power CPS, the power vector of each front-end device is compared with the power vector of any other front-end device. In the power vector with the longest length, each sub-power vector consisting of elements of the same length and continuous with the power vector with the shortest length is obtained. The similarity between the power vector with the shortest length and all sub-power vectors is calculated, and the maximum similarity is calculated.

[0015] The ratio of the maximum similarity to the length difference of the power vector is used as the power similarity between each front-end device and any other front-end device.

[0016] Preferably, the method for determining the relatively high and low relationship clusters of each front-end device is:

[0017] A clustering algorithm is used to cluster the power similarity between each front-end device and all other front-end devices, and two clusters are obtained. The cluster with the cluster center with the largest power similarity is recorded as a relatively high-relationship cluster, and the other cluster is recorded as a relatively low-relationship cluster.

[0018] Preferably, the high relationship value of each front-end device is an average value of the corresponding power similarities of all front-end devices in the relatively high relationship cluster of each front-end device.

[0019] Preferably, the expression for the discreteness of the heterogeneous data of the power CPS is: k=1-exp(-V); wherein k represents the discreteness of the heterogeneous data of the power CPS; V represents the degree of discreteness of the high-relationship mean of all front-end devices in the power CPS; and exp() represents an exponential function with a natural constant as the base.

[0020] Preferably, the expression of the difference threshold of the power CPS is: T=E×k; T represents the difference threshold of the power CPS; E represents the extreme difference of the high relationship values ​​of all front-end devices in the power CPS.

[0021] Preferably, the difference degree of each front-end device is the absolute value of the difference between the mean of all elements in the relatively low-relationship cluster of each front-end device and the high-relationship value.

[0022] Preferably, the method for determining the power high-relationship cluster of each front-end device is:

[0023] If the difference is less than the difference threshold, the cluster formed by merging the relatively low-relationship cluster and the relatively high-relationship cluster is used as the high-power-relationship cluster of each front-end device; conversely, if the difference is greater than or equal to the difference threshold, the relatively high-relationship cluster is used as the high-power-relationship cluster of each front-end device.

[0024] Preferably, the expression for the number of dynamic virtual nodes of the power CPS is: Where Num represents the number of dynamic virtual nodes of the power CPS; N represents the number of preset virtual nodes; n represents the number of preset computers; m represents the number of all front-end devices in the power CPS; LS i represents the total number of elements in the power high-relationship cluster of front-end device i; ceil() represents the floor function; norm() represents the normalization function.

[0025] Preferably, the storing of power data of all front-end devices of the power CPS includes:

[0026] The power vectors of all front-end devices in the power CPS are used as the input of the consistent hash algorithm, where the number of virtual nodes of the consistent hash algorithm is set to the dynamic virtual node number of the power CPS, and all hash values ​​are output and stored in the computer.

[0027] This application has at least the following beneficial effects:

[0028] This application determines the power similarity between each front-end device and any other front-end device by constructing a power vector and the difference in the power vector length between each front-end device and any other front-end device, as well as the similarity of the power vector. Its beneficial effect is that it can solve the problem of uneven load caused by many similar data clustering together in the hash space to form data hotspots, and improve the uniformity of data distribution; this application adopts a clustering algorithm to determine the relatively high and low relationship clusters of each front-end device; based on the distribution of all elements in the relatively high relationship cluster of each front-end device, the high relationship value is determined; based on the degree of dispersion of the high relationship values ​​of all front-end devices, the heterogeneous data dispersion of the power CPS is determined; based on the difference between the maximum values ​​in the high relationship values ​​of all front-end devices, and the The discreteness of the heterogeneous data is combined with the average distribution of all elements in the relatively low-relationship cluster of each front-end device, and the high-relationship value, to determine the difference of each front-end device, and the relatively high- and low-relationship clusters are merged in combination with the difference threshold to obtain the power high-relationship cluster of each front-end device. The beneficial effect is that the power data can be dispersed to more physical nodes, thereby avoiding the overload of a single node; the present application determines the number of dynamic virtual nodes of the power CPS based on the total number of elements in the power high-relationship cluster and the total number of elements in the relatively high-relationship cluster, and stores the power data of all front-end devices of the power CPS. The beneficial effect is that it can reduce the complexity of power CPS management, improve the flexibility of power CPS expansion and the uniformity of data distribution. The present application improves the consistent hashing algorithm by dynamically allocating the number of virtual nodes, thereby improving the utilization of storage space and the uniformity of data distribution. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0030] Figure 1 A flowchart of the steps of a distributed storage method for heterogeneous data of power CPS provided in one embodiment of the present application;

[0031] Figure 2 A schematic diagram of a power similarity extraction process provided in one embodiment of the present application;

[0032] Figure 3 A schematic diagram of a process for obtaining power high-relationship clusters provided in one embodiment of the present application. DETAILED DESCRIPTION

[0033] To further illustrate the technical means and effectiveness of this application to achieve the intended invention objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effectiveness of the distributed storage method for heterogeneous data in power CPSs proposed in this application. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0034] 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 application belongs.

[0035] The following describes in detail the specific scheme of the distributed storage method for heterogeneous data of power CPS provided by this application with reference to the accompanying drawings.

[0036] An embodiment of the present application provides a distributed storage method for serving electric power CPS heterogeneous data. Specifically, the following distributed storage method for serving electric power CPS heterogeneous data is provided. Figure 1 , the method comprises the following steps:

[0037] Step S1: In the power CPS, the power data of each front-end device at all collection moments within a preset time period is collected, and the binary data of all power data are combined into a power vector.

[0038] Heterogeneous data refers to data with different data types, structures and formats. The data may come from different sources and exist in different formats. In this embodiment, the power data from different front-end devices in the power CPS is heterogeneous data. For the power CPS (such as various sensors, video equipment, etc.), the power data of each front-end device at all collection moments within a preset time period is collected. For example, the power data collected by the current sensor is current data, and the power data collected by the voltage sensor is voltage data. In addition, it also includes repair log information generated when the computer control system maintains, updates or repairs the front-end equipment, as well as operating information (i.e., power fault information) when monitoring the equipment.

[0039] Furthermore, all power data from each front-end device is transmitted to the central server of the power CPS via wireless or wired network transmission. The central server converts the power data into binary data through a built-in program. All binary data from each front-end device constitutes the power vector of each front-end device.

[0040] Step S2: determining the power similarity between each front-end device and any other front-end device based on the difference in power vector length between each front-end device and any other front-end device, and the similarity of the power vectors.

[0041] Regarding the distributed storage of heterogeneous power CPS data, in this embodiment, power data includes current, voltage, service information, repair log information, and power fault information. Each type of power data, generated during normal operation, exhibits significant duplication. For example, power fault information, under normal circumstances, indicates normal operation. When no abnormalities are present, the power fault information is largely identical, except for the operating time.

[0042] Before distributed storage, power data was typically stored in centralized databases or data warehouses. This data could originate from various front-end devices, such as sensors, smart meters, or other data acquisition systems. Data collected by the same front-end device, or the same type of front-end device, can be highly similar. For example, current and voltage data typically fluctuate within an expected normal range. Therefore, since different front-end devices in the power system operate in the same mode, the collected power data can also share many similar characteristics.

[0043] Since different front-end devices may have different models and operating frequencies, the length of the power data collected at the same time may also be different in the power CPS (Note: the length of the power data refers to the length of its binary data), but there may be a large number of similar features between the power data. Therefore, based on the difference in power vector length between each front-end device and any other front-end device, as well as the similarity of the power vector, the power similarity between each front-end device and any other front-end device is determined. Specifically:

[0044] (1) In the power CPS, the power vector of each front-end device is compared with the power vector of any other front-end device. In the power vector with the longest length, each sub-power vector consisting of elements of the same length and continuous with the power vector with the shortest length is obtained. The similarity between the power vector with the shortest length and all sub-power vectors is calculated, and the maximum similarity is calculated.

[0045] (2) Further, the difference in the length of the power vector between each front-end device and any other front-end device is analyzed, and the ratio of the maximum similarity to the length difference of the power vector is used as the power similarity between each front-end device and any other front-end device.

[0046] It should be noted that there are many methods for measuring the similarity between vectors. In this embodiment, the similarity between any two power vectors is measured by calculating the cosine similarity between the power vector with the minimum length and each sub-power vector. The implementer may also use other methods for measuring the similarity between vectors, such as the inverse of the Euclidean distance. This embodiment does not impose any special restrictions on the selection of the method for measuring the similarity between vectors.

[0047] The calculation process of cosine similarity is a well-known technology, and its specific calculation process will not be repeated here.

[0048] Furthermore, based on the power similarity between each front-end device and any other front-end device, it can be understood that in the power CPS, the greater the similarity between the power data collected by the front-end devices, the higher the consistency of these power data in characteristics and patterns, the greater the correlation between the minimum length power vector and each sub-power vector, and the more similar the power data are, the smaller the difference in the length of the power vector is, and the greater the power similarity between the front-end devices; conversely, in the power CPS, the smaller the similarity between the power data collected by the front-end devices, the greater the difference in characteristics and patterns of these power data, the smaller the correlation between the minimum length power vector and each sub-power vector, and the greater the difference between the power data, the greater the difference in the length of the power vector is, and the smaller the power similarity between the front-end devices.

[0049] Preferably, the power similarity extraction process diagram provided in this embodiment is as follows: Figure 2 shown.

[0050] Step S3: Based on the power similarity between each front-end device and all other front-end devices, a clustering algorithm is used to determine the relatively high and low relationship clusters of each front-end device; based on the distribution of all elements in the relatively high relationship cluster of each front-end device, the high relationship value of each front-end device is determined; based on the degree of dispersion of the high relationship values ​​of all front-end devices, the heterogeneous data dispersion of the power CPS is determined; based on the difference between the maximum values ​​in the high relationship values ​​of all front-end devices and the heterogeneous data dispersion, the difference threshold of the power CPS is determined; based on the average distribution of all elements in the relatively low relationship cluster of each front-end device and the high relationship value, the difference of each front-end device is determined, and the relatively high and low relationship clusters are fused in combination with the difference threshold to obtain the power high relationship cluster of each front-end device.

[0051] When the power data from a front-end device is highly similar to the power data from other front-end devices, the power data similarity exhibits a certain clustering characteristic, i.e., it is distributed within a relatively small numerical range. Due to the characteristics of power data similarity calculation, the power data similarity between highly similar power data is distributed towards the larger numerical range. When a power CPS front-end device fails, or abnormal power data is collected, the abnormal power data is less similar to the normal power data, resulting in lower similarity between the power data.

[0052] Therefore, in order to eliminate the interference caused by abnormal data generated by front-end device failures, and by analyzing the similarity of power data between different front-end devices, the front-end devices with greater similarity are divided into one category, specifically:

[0053] (1) Cluster the power similarity between the power vector of each front-end device and the power vectors of all other front-end devices, set the number of clusters to 2, and output two clusters. The cluster with the cluster center with the largest power similarity is recorded as a relatively high-relationship cluster, and the other cluster is recorded as a relatively low-relationship cluster.

[0054] It should be noted that there are many commonly used clustering algorithms. In this embodiment, the k-means clustering algorithm is used to cluster power similarity. The implementer can also use other clustering methods such as fuzzy C-means clustering. This embodiment does not impose any special restrictions on the selection of clustering algorithms.

[0055] (2) For power data with high similarity, all of them exist in relatively high-relationship clusters. However, there may be power data with high similarity to other types of power data. The elements in the relatively high-relationship cluster and the elements in the relatively low-relationship cluster belong to the same category. It is unreasonable for the clustering algorithm to divide them into two categories. Therefore, it is necessary to judge the relatively low-relationship cluster. The specific process is as follows:

[0056] Calculate the mean of all elements in the relatively high-relationship cluster of each front-end device, and record it as the high-relationship value of each front-end device;

[0057] The expression for the discreteness k of heterogeneous data in the power CPS is: k = 1-exp(-V); where V represents the discreteness of the high-relationship values ​​of all front-end devices in the power CPS; and exp() represents an exponential function with a natural constant as the base.

[0058] It should be noted that there are many methods to measure the degree of dispersion of a set of data. In this embodiment, the dispersion coefficient of the high relationship mean of all front-end devices is calculated to measure the dispersion degree of the high relationship mean. The implementer can also use other methods to measure the degree of dispersion of data such as variance and standard deviation. Regarding the selection of methods to measure the degree of dispersion of data, this embodiment does not impose any special restrictions.

[0059] The calculation process of the dispersion coefficient is a well-known technology, and its specific calculation steps are not repeated here.

[0060] Furthermore, based on the discreteness of the heterogeneous data of the power CPS, it can be understood that the smaller the discreteness of the high relationship values ​​of all front-end devices, the greater the similarity between the power data, that is, the smaller the discreteness of the heterogeneous data of the power CPS; conversely, the greater the discreteness of the high relationship values ​​of all front-end devices, the greater the difference between the power data, that is, the greater the discreteness of the heterogeneous data of the power CPS.

[0061] (3) Furthermore, in order to classify front-end devices with greater similarity into one category, the power CPS difference threshold is determined based on the difference between the maximum values ​​of the high relationship values ​​of all front-end devices and the dispersion of the heterogeneous data, specifically:

[0062] The expression of the difference threshold T of the power CPS is: T=E×k; E represents the extreme difference of the high relationship values ​​of all front-end devices in the power CPS.

[0063] According to the difference threshold of the power CPS, it can be understood that when the range of the high relationship values ​​of all front-end devices is larger, it means that the similarity between different front-end devices in the power CPS is lower, the correlation between the power data is lower, and the heterogeneous difference dispersion of the power CPS is larger, then the difference threshold of the power CPS is larger, indicating that the difference in power data between different front-end devices is larger; conversely, when the range of the high relationship values ​​of all front-end devices is smaller, it means that the similarity between different front-end devices in the power CPS is higher, the correlation between the power data is higher, and the heterogeneous difference dispersion of the power CPS is smaller, then the difference threshold of the power CPS is smaller, indicating that the difference in power data between different front-end devices is smaller.

[0064] (4) Further, based on the average distribution of all elements in the relatively low-relationship cluster of each front-end device and the high-relationship value, the difference of each front-end device is determined, specifically:

[0065] The difference degree of each front-end device is the absolute value of the difference between the mean of all elements in the relatively low-relationship cluster of each front-end device and the high-relationship value.

[0066] According to the degree of difference of each front-end device, it can be understood that when the similarity of the power data between the front-end devices is high, the smaller the difference between the relatively low-relationship cluster and the relatively high-relationship cluster of the front-end devices, that is, the smaller the difference between the mean of all elements in the relatively low-relationship cluster and the high-relationship value, the higher the similarity between the elements in the relatively low-relationship cluster and the elements in the relatively high-relationship cluster, and the relatively low-relationship cluster and the relatively high-relationship cluster should be merged into one cluster; conversely, when the similarity of the power data between the front-end devices is low, the greater the difference between the relatively low-relationship cluster and the relatively high-relationship cluster of the front-end devices, that is, the greater the difference between the mean of all elements in the relatively low-relationship cluster and the high-relationship value, the lower the similarity between the elements in the relatively low-relationship cluster and the elements in the relatively high-relationship cluster, and only the relatively high-relationship cluster should be retained.

[0067] (5) Further, based on the difference degree and the difference threshold, a high power relationship cluster is constructed, specifically: if the difference degree is less than the difference threshold, it means that the similarity between the elements in the relatively low relationship cluster and the elements in the relatively high relationship cluster is higher, and the cluster formed by merging the relatively low relationship cluster and the relatively high relationship cluster is used as the high power relationship cluster of each front-end device; conversely, if the difference degree is greater than or equal to the difference threshold, the relatively high relationship cluster is used as the high power relationship cluster of the corresponding front-end device.

[0068] Preferably, the schematic diagram of the process of obtaining the power high-relationship cluster provided in this embodiment is as follows: Figure 3 shown.

[0069] Step S4: Based on the total number of elements in the power high-relationship cluster of each front-end device, the number of dynamic virtual nodes of the power CPS is determined, and the power data of all front-end devices of the power CPS is stored.

[0070] For the power high-relationship cluster, the number of elements in the cluster represents the number of front-end devices similar to the corresponding front-end device. When the number of similar front-end devices is greater, the number of virtual nodes should be used in the consistent hashing algorithm to improve the distribution uniformity of the heterogeneous data of the power CPS.

[0071] Therefore, based on the total number of elements in the power high-relationship cluster and the total number of elements in the relatively high-relationship cluster, the number of dynamic virtual nodes of the power CPS is determined, specifically:

[0072] The expression of the dynamic virtual node number Num of the power CPS is: In the formula, N represents the number of preset virtual nodes; n represents the number of preset computers; m represents the number of all front-end devices in the power CPS; LS i represents the total number of elements in the power high-relationship cluster of front-end device i; ceil() represents the floor function; norm() represents the normalization function.

[0073] It should be noted that the values ​​of the preset number of virtual nodes and the preset number of computers are all set manually. In this embodiment, the value of the preset number of virtual nodes is 160, and the value of the preset number of computers is 5. The implementer can also set them according to the specific situation. This embodiment does not impose any special restrictions.

[0074] Furthermore, the power vectors of all front-end devices in the power CPS are used as the input of the consistent hashing algorithm, where the number of virtual nodes of the consistent hashing algorithm is set to the dynamic number of virtual nodes of the power CPS, and all output hash values ​​are stored in the computer according to the size of the computer's storage capacity.

[0075] Among them, the consistent hashing algorithm is a well-known technology, and the specific process of obtaining the hash value will not be described in detail. It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0076] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0077] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them. Modifications to the technical solutions described in the aforementioned embodiments, or equivalent replacements of some of the technical features therein, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A distributed storage method for heterogeneous data of power CPS, characterized by: The method comprises the following steps: In the power CPS, the power data of each front-end device at all collection moments within a preset time period is collected, and the binary data of all power data are combined into a power vector; Determine the power similarity between each front-end device and any other front-end device based on the difference in power vector length between each front-end device and any other front-end device, and the similarity of the power vectors; Based on the power similarity between each front-end device and all other front-end devices, a clustering algorithm is used to determine the relatively high and low relationship clusters of each front-end device; based on the distribution of all elements in the relatively high relationship cluster of each front-end device, the high relationship value of each front-end device is determined; based on the dispersion degree of the high relationship values ​​of all front-end devices, the dispersion degree of the heterogeneous data of the power CPS is determined; Based on the difference between the maximum values ​​in the high relationship values ​​of all front-end devices and the dispersion of the heterogeneous data, a difference threshold of the power CPS is determined; based on the average distribution of all elements in the relatively low relationship cluster of each front-end device and the high relationship value, the difference of each front-end device is determined, and the relatively high and low relationship clusters are merged in combination with the difference threshold to obtain a high power relationship cluster for each front-end device; Based on the total number of elements in the power high-relationship cluster, the number of dynamic virtual nodes of the power CPS is determined, and the power data of all front-end devices of the power CPS is stored.

2. The distributed storage method for serving heterogeneous data of power CPS according to claim 1, characterized in that: The method for determining the power similarity between each front-end device and any other front-end device is as follows: In the power CPS, the power vector of each front-end device is compared with the power vector of any other front-end device. In the power vector with the longest length, each sub-power vector consisting of elements of the same length and continuous with the power vector with the shortest length is obtained. The similarity between the power vector with the shortest length and all sub-power vectors is calculated, and the maximum similarity is calculated. The ratio of the maximum similarity to the length difference of the power vector is used as the power similarity between each front-end device and any other front-end device.

3. The distributed storage method for serving heterogeneous data of power CPS according to claim 1, characterized in that: The method for determining the relatively high and low relationship clusters of each front-end device is as follows: A clustering algorithm is used to cluster the power similarity between each front-end device and all other front-end devices, and two clusters are obtained. The cluster with the cluster center with the largest power similarity is recorded as a relatively high-relationship cluster, and the other cluster is recorded as a relatively low-relationship cluster.

4. The distributed storage method for serving heterogeneous data of power CPS according to claim 1, characterized in that: The high relationship value of each front-end device is the average value of the corresponding power similarities of all front-end devices in the relatively high relationship cluster of each front-end device.

5. The distributed storage method for serving heterogeneous data of power CPS according to claim 1, characterized in that: The expression for the discreteness of the heterogeneous data of the power CPS is: k = 1-exp(-V); where k represents the discreteness of the heterogeneous data of the power CPS; V represents the degree of discreteness of the high-relationship mean of all front-end devices in the power CPS; and exp() represents an exponential function with a natural constant as the base.

6. The distributed storage method for serving heterogeneous data of power CPS according to claim 5, characterized in that: The expression of the difference threshold of the power CPS is: T=E×k; T represents the difference threshold of the power CPS; E represents the extreme difference of the high relationship values ​​of all front-end devices in the power CPS.

7. The distributed storage method for serving heterogeneous data of power CPS according to claim 1, characterized in that: The difference degree of each front-end device is the absolute value of the difference between the mean of all elements in the relatively low-relationship cluster of each front-end device and the high-relationship value.

8. The distributed storage method for serving heterogeneous data of power CPS according to claim 1, characterized in that: The method for determining the power high-relationship cluster of each front-end device is as follows: If the difference is less than the difference threshold, the cluster formed by merging the relatively low-relationship cluster and the relatively high-relationship cluster is used as the power high-relationship cluster of each front-end device; On the contrary, if the difference is greater than or equal to the difference threshold, the relatively high-relationship cluster is used as the high-power-relationship cluster of each front-end device.

9. The distributed storage method for serving heterogeneous data of power CPS according to claim 1, characterized in that: The expression of the dynamic virtual node number of the power CPS is: Where Num represents the number of dynamic virtual nodes of the power CPS; N represents the number of preset virtual nodes; n represents the number of preset computers; m represents the number of all front-end devices in the power CPS; LS i represents the total number of elements in the power high-relationship cluster of front-end device i; ceil() represents the floor function; norm() represents the normalization function.

10. The distributed storage method for serving heterogeneous data of power CPS according to claim 1, characterized in that: The storage of power data of all front-end devices of the power CPS includes: The power vectors of all front-end devices in the power CPS are used as the input of the consistent hash algorithm, where the number of virtual nodes of the consistent hash algorithm is set to the dynamic virtual node number of the power CPS, and all hash values ​​are output and stored in the computer.

Images