Industrial park power utilization sensing method and system based on key load cluster

Abstract

The invention discloses an industrial park power utilization sensing method and system based on a key load cluster, and the method comprises the following steps: taking a key power utilization load in an industrial park as a sensing object, recognizing the core power utilization characteristics, building a multi-dimensional clustering space, and obtaining a plurality of key load cluster division schemes through a clustering algorithm; based on the distance between the loads in each key load cluster division scheme, constructing a key load cluster division scheme evaluation model used for screening the multiple key load cluster division schemes to obtain an optimal key load cluster division scheme; and constructing a cost sensing model for screening and obtaining the representative load of each key load cluster in the optimal key load cluster division scheme, deploying a sensor at the representative load of each key load cluster, and deducing the power consumption data of each key load cluster based on the power consumption data measured by the sensor.

Description

Technical Field

[0001] The present invention belongs to the technical field of load power consumption perception, and more specifically, relates to a method for fine-grained power consumption perception in industrial parks based on key load clusters. The method can be applied to fields such as key load cluster identification and fine-grained power consumption perception in industrial parks. Background Art

[0002] The observability and measurability of electricity consumption in industrial parks is of great significance to the efficient use of electricity on the load side and the generation side of the power system. However, the current acquisition of electricity consumption data of multiple loads in industrial parks is based on the measurement method of gateway meters, which lacks the perception of various power loads within the industrial park. Therefore, the refined and accurate perception technology of multiple loads in industrial parks is studied to obtain the electricity consumption data of key power loads in industrial parks, and further support the load-side demand response and capacity improvement of industrial parks.

[0003] So far, many scholars have focused their research on power consumption sensing methods on two aspects: non-invasive load decomposition model and direct measurement. Among them, non-invasive load decomposition is a method based on big data analysis to decompose the load superposition signal in the grid current data and extract the operating status of the power equipment. However, the power load in the industrial park has the characteristics of a large number of types, including industrial power loads and domestic power loads with large power differences. The non-invasive load method has the problem of low accuracy in identifying the power consumption of multiple loads in the industrial park. The direct measurement method is a method of installing a smart meter for a single load. This method can accurately perceive the power consumption of multiple loads in the industrial park, but because of its high cost, cumbersome installation process and the need to ensure that the installation location is dry and ventilated, it is mainly suitable for sensing the power consumption of high-power equipment, and it is difficult to deploy it on a large scale in industrial parks with low load power, large data volume and large differences in load types. Magnetic sensitive electrical quantity sensing technology can circumvent the above problems to a certain extent, but full deployment will result in high costs.

[0004] At present, domestic and foreign research teams have conducted research on load clustering methods for multi-loads in industrial parks, and proposed methods for building park load clusters based on algorithms such as rough sets, principal component analysis, grey correlation analysis, and cluster analysis. This type of method first identifies the multi-dimensional core power consumption characteristics of key loads based on the rough set algorithm, and then uses the clustering algorithm to group the key loads to obtain a series of key load clusters. This type of method provides a technical solution for the refined power consumption perception of industrial parks, but does not consider the selection of the optimal clustering scheme and the generation of multi-temporal and spatial scale data assets for power consumption in industrial parks.

[0005] The power load data clustering and fusion method in the existing technology widely studies the impact of social, economic, meteorological and other factors on various loads in the power industry, clusters various load curves in the power industry and evaluates the load curve characteristics, which has a low contribution to specific applications. It only decomposes the power consumption data of the substation bus into multiple samples in units of 24 hours, performs dimensionality reduction and clustering on them, and can only perceive the total power consumption of the load at the feeder level, but cannot obtain the detailed power consumption status of various loads on the demand side. Summary of the invention

[0006] In order to solve the deficiencies in the prior art, the present invention provides an industrial park electricity consumption sensing method based on key load clusters, which solves the technical difficulties that the load cluster division method in the prior art is relatively general and the existing industrial park load electricity consumption sensing method fails to comprehensively consider the sensor deployment plan from the perspective of accuracy and economy.

[0007] The present invention adopts the following technical solution.

[0008] A first aspect of the present invention provides an industrial park power consumption sensing method based on a critical load cluster, comprising the following steps:

[0009] The key power loads in the industrial park are taken as the sensing objects. After identifying the core power consumption characteristics, a multi-dimensional clustering space is established. The clustering algorithm is used to obtain a variety of key load cluster division schemes.

[0010] Based on the distance between loads in each key load clustering scheme, a key load clustering scheme evaluation model is constructed to screen and obtain the optimal key load clustering scheme from multiple key load clustering schemes;

[0011] A perception cost model is constructed to screen and obtain the representative load of each key load cluster in the optimal key load cluster division scheme, and sensors are deployed at the representative load of each key load cluster. The power consumption data of each key load cluster is deduced based on the power consumption data measured by the sensors.

[0012] Preferably, various loads in the industrial park are clustered according to a clustering algorithm, and a variety of key load cluster division schemes are obtained by setting different clustering parameters.

[0013] Preferably, based on the distance between loads in each critical load clustering partitioning scheme, constructing a critical load clustering partitioning scheme evaluation model includes:

[0014] The average Euclidean distance from each load node to the cluster center in each key load cluster division scheme is calculated and expressed as follows:

[0015]

[0016] Where:

[0017] S i，j is the average Euclidean distance from each load node to the cluster center in the jth critical load cluster of the i-th critical load cluster partitioning scheme;

[0018] N i，j is the number of load nodes in the jth critical load cluster of the i-th critical load cluster partitioning scheme;

[0019] X i，j，p is the pth load node of the jth critical load cluster, p∈[1,N i，j ];

[0020] ω i，j The cluster center of the jth critical load cluster of the i-th critical load cluster partitioning scheme.

[0021] Preferably, the constructed critical load cluster partitioning scheme evaluation model is expressed by the following formula:

[0022]

[0023] Where:

[0024] D i is the evaluation value of the i-th critical load clustering partitioning scheme, i∈[1,n], n is the number of critical load clustering partitioning schemes obtained by clustering;

[0025] M i is the number of clusters in the i-th critical load cluster partitioning scheme;

[0026] S i，j is the average Euclidean distance from each load node to the cluster center in the jth critical load cluster of the i-th critical load cluster partitioning scheme, S i，k is the average Euclidean distance from each load node to the cluster center in the kth critical load cluster of the i-th critical load cluster partitioning scheme, and k is the other critical load clusters except the jth critical load cluster in the i-th critical load cluster partitioning scheme;

[0027] ω i，j is the cluster center of the jth critical load cluster of the i-th critical load cluster partitioning scheme, ω i，k is the cluster center of the kth critical load cluster of the i-th critical load cluster partitioning scheme, ||ω i，j -ω i，k ||2 is the Euclidean distance between the centers of two clusters.

[0028] Preferably, selecting and obtaining the optimal key load cluster division scheme from a plurality of key load cluster division schemes includes:

[0029] Calculate the evaluation values ​​of n key load cluster partitioning scheme items and select the minimum value D min The corresponding critical load clustering partitioning scheme is taken as the optimal critical load clustering partitioning scheme.

[0030] Preferably, constructing the perceived cost model includes:

[0031] A perception cost model is established that takes into account the communication cost and the difficulty of deduction. Specifically, the communication cost model is constructed by taking into account the communication distance and the selected frequency, and the deduction difficulty model is constructed by taking into account the accuracy of the representative load measurement data in deducing the power consumption of each load in the key load cluster.

[0032] Preferably, the perception cost model is as shown in the following formula:

[0033]

[0034] Where:

[0035] Q i，j，q is the evaluation objective function of the critical load cluster j for the optimal critical load cluster partitioning scheme i;

[0036] N i，j The number of load nodes in critical load cluster j of the optimal critical load cluster partitioning scheme i;

[0037] C i，j，q is the communication cost between the qth load node of the critical load cluster j and the edge smart energy unit of the optimal critical load cluster partitioning scheme i, D i，j，q The difficulty of deducing the power consumption of other loads in the critical load cluster from the measured value of the qth load node of the critical load cluster j of the optimal critical load cluster partitioning scheme i;

[0038] f is the communication frequency,

[0039] x i，j，q The communication distance between the qth load node of the critical load cluster j and the edge smart energy unit of the optimal critical load cluster partitioning scheme i;

[0040] R i，j，q，p is the Euclidean distance between the qth load node and the pth load node of the critical load cluster j of the optimal critical load cluster partitioning scheme i, where q is the load node other than p in the critical load cluster, and R i，j It is the radius of the distance between the two farthest load nodes in the critical load cluster.

[0041] Preferably, each load node in each key load cluster of the optimal key load cluster division scheme is used as a calculation object of the perception cost model, and its evaluation value as a representative load is calculated, and the load node corresponding to the minimum value is taken as the representative load of the key load cluster.

[0042] Preferably, the power consumption data of the key loads in the industrial park is deduced based on the similarity between the power consumption data of the representative load measured by the sensor and the power consumption characteristics of other loads in the same key load cluster.

[0043] A second aspect of the present invention provides an industrial park power consumption sensing system based on a critical load cluster, and the operation of the industrial park power consumption sensing method based on a critical load cluster includes:

[0044] The key load clustering division scheme construction module is used to take the key power loads in the industrial park as the sensing object, identify the core power consumption characteristics, and construct the industrial park power consumption sensing multi-dimensional clustering space, and use the clustering algorithm to determine the key load clustering division scheme;

[0045] The critical load clustering scheme evaluation module builds a critical load clustering scheme evaluation model based on the distance between loads in each critical load clustering scheme, which is used to screen and obtain the optimal critical load clustering scheme from multiple critical load clustering schemes;

[0046] Cluster representative load solving module, used to establish a perception cost model and use the model to select representative loads from each load node in the cluster;

[0047] The sensor deployment plan building module forms a sensor deployment plan based on the cluster division of the optimal key loads and the location of the representative loads in the cluster.

[0048] Compared with the prior art, the beneficial effects of the present invention include at least:

[0049] The present invention provides a refined power consumption perception method for industrial parks based on key load clusters, which realizes fine-grained power consumption perception of multiple loads in the park while taking into account the perception accuracy and implementation cost. Specifically, by identifying key load clusters, the research object of power consumption perception in industrial parks is clarified, and the purposefulness of power consumption perception in the park is improved; then, the optimal sensor deployment scheme based on representative loads of the cluster is obtained by comprehensively considering the communication cost and deduction difficulty, and the minimization of the collection of power consumption perception in industrial parks is realized, which provides technical support for the generation of key power consumption data assets in industrial parks and improves the power consumption status perception capability for multiple loads in the park.

[0050] More specifically, (1) the present invention proposes for the first time a method for clustering key loads based on the Davidson-Botting Index (DBI), which characterizes the effectiveness of the load clustering results by calculating the clustering similarity index within and between each load cluster class. This overcomes the technical bottleneck that the existing clustering algorithm is difficult to accurately judge the advantages and disadvantages of load clustering schemes, achieves accurate judgment of the optimal partitioning scheme for key load clusters, improves the purposefulness of power consumption perception in industrial parks, and provides an analysis object for the refined power consumption perception of multiple loads in industrial parks.

[0051] (2) The present invention proposes for the first time a perception cost model that takes into account communication costs and deduction difficulties. Representative loads are selected based on the calculation results of the perception cost model to obtain a refined electricity consumption perception solution for industrial parks, solving the problem of optimal deployment of sensors for refined electricity consumption perception in industrial parks. The difficulty of calculating load cluster electricity consumption perception data is effectively reduced at a lower communication cost, and accurate perception of electricity consumption of multiple loads in the park is achieved. The internal load of the industrial park is refinedly analyzed, providing technical support for generating data assets for industrial park electricity consumption perception and for smooth response to peak demand. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0053] Figure 1 The technical framework and workflow diagram of the method of the present invention;

[0054] Figure 2 A corresponding relationship diagram between the target problem and the mathematical problem for the refined electricity consumption perception method;

[0055] Figure 3 The load distribution and power supply topology of the electric tool factory in the embodiment of the present invention;

[0056] Figure 4 Schematic diagram of a critical load cluster division scheme G1 in an embodiment of the present invention;

[0057] Figure 5 Schematic diagram of critical load cluster division scheme G2 in an embodiment of the present invention;

[0058] Figure 6 Schematic diagram of critical load cluster division scheme G3 in an embodiment of the present invention;

[0059] Figure 7 Schematic diagram of critical load cluster division scheme G4 in an embodiment of the present invention;

[0060] Figure 8 Schematic diagram of the optimal sensor deployment implementation plan in an embodiment of the present invention. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The described embodiments are only embodiments of a part of the present invention, not all embodiments. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of the present invention.

[0062] The method for fine-grained electricity consumption perception in industrial parks based on key load clusters proposed in the present invention takes the key electricity loads in industrial parks as the perception objects, obtains a variety of key load cluster division schemes through the K-means clustering algorithm, and uses the Davidson-Botting Index to evaluate the optimal division scheme of the key load clusters; then, a perception cost model considering the communication cost and the difficulty of deduction is constructed, and the optimal representative load is calculated using the model; finally, a magnetic sensor deployment implementation plan corresponding to the optimal solution is formed.

[0063] like Figure 1 As shown, embodiment 1 of the present invention provides a method for fine-grained power consumption perception in an industrial park based on a key load cluster, comprising the following steps:

[0064] Step 1: Establish a technical framework for the refined power consumption perception method in industrial parks. This step establishes a technical framework for the refined power consumption perception method and clarifies the work content of each step. The technical framework of the refined power consumption perception method includes three parts: determining the optimal division scheme of key load clusters, solving the representative loads in each key load cluster based on the perception cost model, and forming a sensor deployment implementation plan. The overall framework is as follows: Figure 1 shown.

[0065] Step 2: Evaluate the optimal partitioning scheme of key load clusters. First, a multi-dimensional clustering space is constructed based on the existing load clustering method in the power consumption perception of industrial parks; then, clustering algorithms including but not limited to the K-means clustering algorithm are used to cluster various loads, and various key load clustering schemes are obtained by setting different clustering parameters (such as the K value of the K-means clustering algorithm); finally, various key load clustering schemes are evaluated based on DBI to obtain the optimal partitioning scheme of key load clusters.

[0066] Preferably, the step 2 specifically includes:

[0067] Step 2.1: Construct a multidimensional clustering space for power consumption perception in industrial parks. According to the existing key load cluster identification method in power consumption perception in industrial parks, the key power loads in the industrial park are taken as the perception objects, and a multidimensional clustering space is established after identifying the core power consumption characteristics.

[0068] Step 2.2: Obtain multiple key load clustering schemes based on the clustering algorithm. Use the K-means clustering algorithm to classify the sensing objects in the multi-dimensional clustering space and obtain n key load clustering schemes G = {..., G (i) ,.....}, i∈[1,n], the number of key load clusters in each key load cluster partitioning scheme in G is 2 as the first item and increases by 1. The n key load cluster partitioning schemes correspond to n magnetic sensor deployment schemes.

[0069] It is worth noting that in step 2.2, as long as a clustering algorithm is used to obtain multiple clustering partitioning schemes, K-means is only one of the clustering algorithms. It can be that multiple clustering algorithms are used to cluster the clustering space separately, and finally n schemes are obtained, or clustering algorithm 1 obtains n1 clustering partitions, clustering algorithm 2 obtains n2 clustering partitions, n1 and n2 are different, and there are n different clustering partitions in n1 and n2, or only K-means is used to obtain n different partitions by adjusting the k value.

[0070] Step 2.3: Use DBI to determine the optimal partitioning scheme for key load clusters. Use DBI to evaluate the clustering scheme for key load clusters. It can be expressed as the following formula (1):

[0071]

[0072] Where:

[0073] D i is the DBI indicator value;

[0074] M i is the number of clusters in the i-th critical load cluster partitioning scheme;

[0075] S i，j is the average Euclidean distance from each load node to the cluster center in the jth key load cluster of the i-th key load cluster partitioning scheme, that is, the average Euclidean distance from each load node to the cluster center, S i，k is the average Euclidean distance from each load node to the cluster center in the kth critical load cluster of the i-th critical load cluster partitioning scheme, k is the other critical load clusters except the jth critical load cluster in the i-th critical load cluster partitioning scheme, and the average Euclidean distances in each critical load cluster in the scheme form an array s i ;

[0076] ω i，j is the cluster center of the jth critical load cluster of the i-th critical load cluster partitioning scheme, ω i，k is the cluster center of the kth critical load cluster of the i-th critical load cluster partitioning scheme, ||ω i，j -ω i，k ||2 is the Euclidean distance between the centers of two clusters. The Euclidean distance matrix of the centers of each key load cluster in the i-th key load cluster partitioning scheme is ω i .

[0077] Furthermore, S in formula (1) i，j The variables are expressed as follows:

[0078]

[0079] Where:

[0080] N i，j is the number of load nodes in the jth critical load cluster in the i-th critical load cluster partitioning scheme. Each critical load cluster partitioning scheme forms an array N i ,

[0081] X i，j，p It is the pth load node in the critical load cluster.

[0082] The smaller the DBI standard value, the better the classification effect.

[0083] As a prominent substantive feature of the present invention and one of the significant advances it brings to the prior art, the research object is the key load cluster in the demand-side industrial park. The power consumption status of the key load cluster is studied from a more detailed observation object and time dimension, which improves the purposefulness of power consumption perception in the industrial park and provides an analysis object for the refined power consumption perception within the industrial park.

[0084] Step 3: Solve the representative load in each key load cluster based on the perception cost model. Through the above steps, the number of key load clusters in the industrial park can be determined and the optimal division scheme can be obtained. Next, this step further determines the representative load in each key load cluster, considers the communication cost and deduction difficulty of each load node as a representative load, and establishes a perception cost model for the refined electricity consumption perception method. The model is used to select representative loads from each load node in the key load cluster, providing technical support for the subsequent steps of implementing the deployment of magnetic sensors. The relationship between step 3 and the previous and subsequent steps is as follows: Figure 2 shown.

[0085] Preferably, the step 3 comprises:

[0086] Step 3.1: Establish a perception cost model that takes into account communication cost and deduction difficulty. The perception cost model uses the weighted value of communication cost and deduction difficulty as an evaluation indicator. Each load node is taken as a representative load and is calculated in the perception cost model. The smaller the index value obtained, the lower the communication cost and the lower the deduction difficulty. The communication cost is the cost of communication between the sensor and the edge smart energy unit, which depends on the communication distance and the selected frequency. The deduction difficulty refers to the accuracy of deducing the power consumption of each load in the key load cluster from the representative load measurement data. If the accuracy is low, it indicates that the deduction difficulty is high. The perception cost model is shown in formula (3):

[0087]

[0088] Where:

[0089] Q i，j，q To evaluate the objective function;

[0090] N i，j is the number of load nodes;

[0091] C i，j，q is the communication cost between the qth load node and the edge smart energy unit, D i，j，q The difficulty of deducing the power consumption of other loads in the critical load cluster from the measured value of the qth load node;

[0092] f is the frequency (in MHz),

[0093] x i，j，q is the communication distance between the qth load node and the edge smart energy unit (in km);

[0094] R i，j，q，p is the Euclidean distance between the qth load and the pth load (in km), q is the load other than p in the critical load cluster, R i，j It is the radius (in km) between the two farthest load nodes in the critical load cluster.

[0095] It is worth noting that formula (3) selects the optimal representative node from the jth key load cluster of the i-th key load cluster division scheme, and selects the one with the smallest value calculated according to the perception cost model. Each parameter in formula (3) has the corresponding meaning of the jth key load cluster of the i-th key load cluster division scheme.

[0096] Step 3.2: Take each load in the critical load cluster as the research object and calculate its evaluation value as the representative load. The evaluation value of each load in each critical load cluster of the i-th critical load cluster division scheme is calculated by formula (3), and the load node corresponding to the minimum value is taken as the representative load. On the basis of clarifying the critical load cluster division scheme in the previous step, the representative load in each critical load cluster can be further clarified.

[0097] As a prominent substantive feature of the present invention and one of the significant advances it brings to the prior art, the electricity consumption characteristics of industrial loads in the power industry are studied, the electricity consumption status within the industrial park is analyzed, and a perception cost model is proposed to select representative loads to obtain a refined electricity consumption perception solution for the industrial park; through a refined analysis of the internal loads of the industrial park, technical support is provided for generating data assets for industrial park electricity consumption perception and for smooth response to peak summer demand.

[0098] Step 4: Form a sensor deployment implementation plan. From the above steps, we can know the key load cluster division method of the key loads in the industrial park and the representative loads within the key load clusters. Deploy sensors on the representative loads to form a sensor deployment implementation plan. The sensors transmit the measured power consumption data to the edge smart energy unit, and the smart energy unit obtains the power consumption data of the key loads in the industrial park based on digital deduction.

[0099] Preferably, the power consumption data of the key loads in the industrial park is deduced based on the similarity between the power consumption data of the representative load measured by the sensor and the power consumption characteristics of other loads in the same key load cluster.

[0100] In summary, in order to address the fuzzy problem of load electricity consumption after the gateway meter in industrial parks, a refined electricity consumption perception method for industrial parks is proposed, which can lay a technical foundation for the overall data asset deduction of industrial parks.

[0101] Embodiment 2 of the present invention provides a refined power consumption sensing system for an industrial park based on a critical load cluster, and runs a refined power consumption sensing method for an industrial park based on a critical load cluster described in Embodiment 1, including:

[0102] The key load clustering division scheme construction module is used to take the key power loads in the industrial park as the sensing object, identify the core power consumption characteristics, and construct the industrial park power consumption sensing multi-dimensional clustering space, and use the clustering algorithm to determine the key load clustering division scheme;

[0103] The critical load clustering scheme evaluation module builds a critical load clustering scheme evaluation model based on the distance between loads in each critical load clustering scheme, which is used to screen and obtain the optimal critical load clustering scheme from multiple critical load clustering schemes;

[0104] Cluster representative load solving module, used to establish a perception cost model and use the model to select representative loads from each load node in the cluster;

[0105] The sensor deployment plan building module forms a sensor deployment plan based on the cluster division of the optimal key loads and the location of the representative loads in the cluster.

[0106] Embodiment 3 of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, the method for fine-grained power consumption perception in an industrial park based on a critical load cluster as described in Embodiment 1 is implemented.

[0107] Embodiment 4 of the present invention provides a computer-readable storage medium, which stores a computer program, and is characterized in that when the computer program is executed by a processor, it implements the method for refined electricity consumption perception in industrial parks based on critical load clusters according to Embodiment 1.

[0108] In order to more clearly introduce the outstanding essential features of the present invention and the significant progress it brings to the prior art, an example is given below to illustrate the core concept of the present invention. The present invention provides a refined power consumption sensing method for industrial parks based on key load clusters, the steps of which are as follows: Figure 1 The following describes the method flow of the present invention to achieve refined power consumption perception in an industrial park in combination with the actual situation of a power tool factory.

[0109] First, according to step 1 of the method of the present invention, a technical framework of a refined power consumption sensing method is established. The technical framework includes three parts: determining the optimal division scheme of key load clusters, solving the representative loads in each key load cluster based on the sensing cost model, and forming a sensor deployment implementation plan.

[0110] Then, the optimal partitioning scheme of the park's key load clusters is determined according to step 2 of the method of the present invention. First, the multi-dimensional clustering space is determined according to the existing key load cluster identification method in the power consumption perception of industrial parks; second, the perception objects are clustered according to the K-means clustering algorithm, and different K values ​​are set to obtain multiple key load clustering schemes. Finally, the Davidson-Botting index is used to determine the optimal partitioning scheme of the key load clusters.

[0111] According to step 2.1 of the method of the present invention, a multidimensional clustering space for power consumption perception in industrial parks is constructed. According to the existing method for clustering key loads in power consumption perception in industrial parks, the key power loads in the industrial park are taken as the perception objects. Considering that the total load power of the power tool factory in the embodiment of the present invention is 6423kW, with reference to the production process and production schedule of the power tool factory, injection molding machines, air conditioners, and water heaters are selected as the perception objects. The total power of the key loads is 4421kW (accounting for 68.8%), see Figure 3 Next, after identifying the core power consumption characteristics, a multi-dimensional clustering space is established. According to the existing method for identifying the core power consumption characteristics of industrial parks, the load adjustable attribute, the attribute of the continuous working time of the load within a day, and the power supply topology attribute of the load are selected as the core power consumption characteristics of the power tool factory, and a three-dimensional clustering space is established based on this.

[0112] According to step 2.2 of the method of the present invention, multiple key load clustering schemes are obtained according to the clustering algorithm. The K-means clustering algorithm is used to classify the sensed objects in the three-dimensional clustering space, and a total of n key load clustering schemes G = {…, G i ,…}, i∈{1,n}, the number of key load clusters in each key load cluster division scheme in G is 2 as the first item and increases by 1 in sequence. In this implementation case, four key load cluster division schemes are considered, among which the key load cluster division schemes in G={G1,G2,G3,G4} are shown in Table 1, respectively. Figure 4 , Figure 5 , Figure 6 and Figure 7 shown.

[0113] Table 1 Division scheme of key load clusters

[0114]

[0115]

[0116] According to step 2.3 of the method of the present invention, the DBI index is used to determine the optimal partitioning scheme of the critical load clusters. The DBI is used to evaluate the clustering scheme of the critical load clusters, and the DBI index value corresponding to each critical load cluster partitioning scheme in G is calculated according to formula (1), and D = {D1, D2, D3, D4} is obtained.

[0117] Table 2 Calculation parameters and results of various partitioning schemes

[0118]

[0119] Through comparison, it is found that the DBI index value corresponding to the key load cluster division scheme G4 is the smallest, so it is considered most appropriate to divide the key loads of the industrial park into 5 categories.

[0120] Afterwards, according to step 3 of the method of the present invention, the representative loads in each key load cluster are solved based on the perception cost model. Through the above steps, the number of key load clusters for the key loads in the industrial park can be clarified, and the optimal division scheme for the key load clusters can be obtained. Next, this step further determines the representative loads in each key load cluster, considers the communication cost and deduction difficulty generated by each load node as a representative load, and establishes a perception cost model for the refined electricity consumption perception method. The model is used to select representative loads from each load node in the key load cluster, providing technical support for the subsequent steps of implementing the deployment of magnetic sensors.

[0121] According to step 3.1 of the method of the present invention, a perception cost model considering communication cost and deduction difficulty is established. It should be noted that since each key load cluster has independent observation characteristics and the edge node deployment location has been determined. In order to facilitate the feasibility of the present invention, it is assumed that f = 500; R 5,1 =1.10; R 5,2 =0.60; R 5,3 =0.90; R 5,4 =1.80; R 5,5 =1.30. The parameters of other key load clusters are shown in Table 3.

[0122] Table 3 Calculation parameters and results of the perception cost model in each key load cluster

[0123]

[0124] According to step 3.2 of the method of the present invention, the load node corresponding to the minimum value is taken as the representative load. By comparison, it is found that the minimum evaluation values ​​in each key load cluster are Q 5,1,2 , Q 5,2,1 , Q 5,3,2 , Q 5,4,2 , Q 5,5,1 ,Therefore, the corresponding load is selected as the representative load in each key load cluster.

[0125] According to step 4 of the method of the present invention, a sensor deployment implementation plan is formed. According to the aforementioned step 2, the key load cluster division method of the optimal key load in the industrial park can be known, and according to step 3, the representative load position within the key load cluster can be known. Afterwards, the sensor is installed at the representative load to form a sensor deployment implementation plan. The sensor transmits the measured power consumption data to the edge smart energy unit, and the smart energy unit obtains the power consumption data of the key loads in the industrial park based on digital deduction. The specific implementation plan is as follows Figure 8 shown.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

[0127] The present disclosure may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for sensing power consumption in an industrial park based on a critical load cluster, characterized in that: The following steps are involved: The key power loads in the industrial park are taken as the sensing objects. After identifying the core power consumption characteristics, a multi-dimensional clustering space is established. The clustering algorithm is used to obtain a variety of key load cluster division schemes. Based on the distance between loads in each key load clustering scheme, a key load clustering scheme evaluation model is constructed to screen and obtain the optimal key load clustering scheme from multiple key load clustering schemes; A perception cost model is constructed to screen and obtain the representative load of each key load cluster in the optimal key load cluster division scheme, and sensors are deployed at the representative load of each key load cluster. The power consumption data of each key load cluster is deduced based on the power consumption data measured by the sensors.

2. According to claim 1, a method for sensing power consumption in an industrial park based on a critical load cluster is characterized in that: Various loads in the industrial park are clustered according to the clustering algorithm, and a variety of key load cluster division schemes are obtained by setting different clustering parameters.

3. According to claim 1, a method for sensing power consumption in an industrial park based on a critical load cluster is characterized in that: Based on the distance between loads in each key load clustering scheme, the key load clustering scheme evaluation model is constructed, including: The average Euclidean distance from each load node to the cluster center in each key load cluster division scheme is calculated and expressed as follows: Where: S i,j is the average Euclidean distance from each load node to the cluster center in the jth critical load cluster of the i-th critical load cluster partitioning scheme; N i,j is the number of load nodes in the jth critical load cluster of the i-th critical load cluster partitioning scheme; X i,j,p is the pth load node of the jth critical load cluster, p∈[1,N i,j ]; ω i,j The cluster center of the jth critical load cluster of the i-th critical load cluster partitioning scheme.

4. The method for sensing power consumption in an industrial park based on a critical load cluster according to claim 1 or 3, characterized in that: The constructed evaluation model of key load cluster partitioning scheme is expressed as follows: Where: D i is the evaluation value of the i-th critical load clustering partitioning scheme, i∈[1,n], n is the number of critical load clustering partitioning schemes obtained by clustering; M i is the number of clusters in the i-th critical load cluster partitioning scheme; S i，j is the average Euclidean distance from each load node to the cluster center in the jth critical load cluster of the i-th critical load cluster partitioning scheme, S i，k is the average Euclidean distance from each load node to the cluster center in the kth critical load cluster of the i-th critical load cluster partitioning scheme, and k is the other critical load clusters except the jth critical load cluster in the i-th critical load cluster partitioning scheme; ω i，j is the cluster center of the jth critical load cluster of the i-th critical load cluster partitioning scheme, ω i，k is the cluster center of the kth critical load cluster of the i-th critical load cluster partitioning scheme, ||ω i，j -ω i，k ||2 is the Euclidean distance between the centers of two clusters.

5. The method for sensing power consumption in an industrial park based on a critical load cluster according to claim 4 is characterized in that: The optimal key load cluster partitioning scheme is selected from a variety of key load cluster partitioning schemes, including: Calculate the evaluation values ​​of n key load cluster partitioning scheme items and select the minimum value D min The corresponding critical load clustering partitioning scheme is taken as the optimal critical load clustering partitioning scheme.

6. The method for sensing power consumption in an industrial park based on a critical load cluster according to claim 1 or 5, characterized in that: Building a perceived cost model involves: A perception cost model is established that takes into account the communication cost and the difficulty of deduction. Specifically, the communication cost model is constructed by taking into account the communication distance and the selected frequency, and the deduction difficulty model is constructed by taking into account the accuracy of the representative load measurement data in deducing the power consumption of each load in the key load cluster.

7. The method for sensing power consumption in an industrial park based on a critical load cluster according to claim 6 is characterized in that: The perceived cost model is shown in the following formula: Where: Q i，j，q is the evaluation objective function of the critical load cluster j for the optimal critical load cluster partitioning scheme i; N i，j The number of load nodes in critical load cluster j of the optimal critical load cluster partitioning scheme i; C i，j，q is the communication cost between the qth load node of the critical load cluster j and the edge smart energy unit of the optimal critical load cluster partitioning scheme i, D i，j，q The difficulty of deducing the power consumption of other loads in the critical load cluster from the measured value of the qth load node of the critical load cluster j of the optimal critical load cluster partitioning scheme i; f is the communication frequency, x i，j，q The communication distance between the qth load node of the critical load cluster j and the edge smart energy unit of the optimal critical load cluster partitioning scheme i; R i，j，q，p is the Euclidean distance between the qth load node and the pth load node of the critical load cluster j of the optimal critical load cluster partitioning scheme i, where q is the load node other than p in the critical load cluster, and R i，j It is the radius of the distance between the two farthest load nodes in the critical load cluster.

8. The method for sensing power consumption in an industrial park based on a critical load cluster according to claim 1 or 7, characterized in that: Each load node in each key load cluster of the optimal key load cluster division scheme is taken as the calculation object of the perception cost model, and its evaluation value as a representative load is calculated, and the load node corresponding to the minimum value is taken as the representative load of the key load cluster.

9. The method for sensing power consumption in an industrial park based on a critical load cluster according to claim 1, characterized in that: The power consumption data of the key loads in the industrial park is deduced by comparing the power consumption data of the representative loads measured by the sensors with the power consumption characteristics of other loads in the same key load cluster.

10. An industrial park power consumption sensing system based on a critical load cluster, running an industrial park power consumption sensing method based on a critical load cluster according to claims 1-9, characterized in that: The key load clustering division scheme construction module is used to take the key power loads in the industrial park as the sensing object, identify the core power consumption characteristics, and construct the industrial park power consumption sensing multi-dimensional clustering space, and use the clustering algorithm to determine the key load clustering division scheme; The critical load clustering scheme evaluation module builds a critical load clustering scheme evaluation model based on the distance between loads in each critical load clustering scheme, which is used to screen and obtain the optimal critical load clustering scheme from multiple critical load clustering schemes; Cluster representative load solving module, used to establish a perception cost model and use the model to select representative loads from each load node in the cluster; The sensor deployment plan building module forms a sensor deployment plan based on the cluster division of the optimal key loads and the location of the representative loads in the cluster.

Images