A method for mining installed capacity of residential air conditioners based on K-medoids cluster analysis

Abstract

The present invention provides a method for mining the installed capacity of residential air conditioners based on K-medoids clustering analysis, including the following steps: Step 1: Obtain electricity consumption data; Step 2: Calculate the power mean and power variance; Step 3: Establish a corresponding fuzzy matrix; Step 4: Calculate power data; Step 5: Calculate the average distance density value of electricity consumption power data; Step 6: Classify different cluster distances C X ; Step 7: End clustering when the clustering data set is an empty set; Step 8: Update the data center; Step 9: Obtain the clustering center sequence #imgabs0#; Step 10: Extract the clustering clusters of different clustering centers; Step 11: Obtain the total electricity load including air conditioner electricity consumption; Step 12: Calculate the electricity load of other electrical appliances; Step 13: Deduce the installed capacity of air conditioners in the household of this residential user. Applying this technical solution can achieve the calculation of the installed capacity of air conditioners.

Description

Technical Field

[0001] The invention relates to the technical field of power supply and demand balance, in particular to a method for mining installed capacity of residential air conditioners based on K-medoids cluster analysis. Background Art

[0002] At present, the peak load of air conditioning load in various regions generally accounts for more than 20%. In developed regions such as Beijing, Shanghai, Jiangsu, and Zhejiang, the proportion is even more than 50%, and this proportion is still increasing year by year. Air conditioning load has become the main factor affecting the peak-to-valley difference of the power grid in summer. The large-scale grid connection of renewable energy has led to increasingly tight system balancing resources. The rapid development of smart grid two-way communication technology and advanced measurement systems has provided technical support for user-side load monitoring and control. Air conditioning load has the characteristics of strong controllability and great dispatching potential. Through the aggregation of massive air conditioning loads by load aggregators, coordinated control and participation in system regulation, it has shown great potential in peak shaving and valley filling, maintaining the stability of the power system, and providing auxiliary services. The key to fully tapping and utilizing this potential lies in tapping the installed capacity of residential air conditioning. Existing technologies have not yet proposed an effective tapping method. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide a residential air-conditioning installed capacity mining method based on K-medoids cluster analysis. Based on the nighttime electricity consumption patterns of residential users in summer, a clustering analysis is performed on the users' nighttime electricity consumption through a clustering method, and the users' electricity loads with air conditioning and without air conditioning on similar days are extracted to further calculate the air-conditioning installed capacity.

[0004] To achieve the above object, the present invention adopts the following technical solution: a method for mining installed capacity of residential air conditioners based on K-medoids cluster analysis, comprising the following steps:

[0005] Step 1: Obtain electricity consumption data of residential users through the electricity consumption information collection system;

[0006] Step 2: Calculate the power mean and power variance of the 12 power measurement points during the early morning hours;

[0007] Step 3: Establish the corresponding fuzzy matrix according to the power mean, power variance, and maximum and minimum power indicators of the user during the early morning hours;

[0008] Step 4: Calculate power data;

[0009] Step 5: Calculate the average distance density value of power consumption data;

[0010] Step 6: Distance C for different clusters X To classify;

[0011] Step 7: Repeat steps 4 to 5 according to the power data set given in step 3, and terminate clustering when the clustering data set is an empty set;

[0012] Step 8: Update the data center;

[0013] Step 9: According to the set minimum threshold φ xy , when the cluster center is within the threshold σ x When the operating range is within the range and the change is small, that is, φ xy ≤σ x When , this center is considered as the cluster center, and the cluster center sequence C'={C1,C2,...,C x};

[0014] Step 10: Extract clusters with different cluster centers, calculate the average value of the power curves between clusters, and preliminarily divide the power cluster load into: power consumption of other electrical appliances at night, power consumption of air conditioners, and different power consumption of air conditioners due to different temperatures;

[0015] Step 11: After extracting the suspected 'dates containing air conditioning electricity consumption' and clusters in step 10, the analysis of the dates suspected to contain air conditioning is extended to the whole day to obtain the total electricity load containing air conditioning electricity consumption;

[0016] Step 12: Use similar days to calculate the power load of other electrical appliances without air conditioning;

[0017] Step 13: Subtract the 'total power load including air conditioning' calculated in step 11 from the 'total power load excluding air conditioning' in step 12 to obtain the air conditioning load value for each period, extract the maximum value among all periods, take the average value of the period as the air conditioning load of the residential user, and compare it with the commonly used air conditioning power consumption characteristics on the market to estimate the installed capacity of the residential user's household air conditioning.

[0018] In a preferred embodiment, the step 1 is specifically as follows: analyzing whether there is air conditioning usage behavior from the residential user's early morning electricity usage behavior; collecting the power data of the residential user during the early morning from July to October, and after data cleaning and filling the missing values ​​of the data, obtaining the power values ​​of 12 measurement points of the user during the early morning, namely:

[0019] P={P1,P2,...,P 12} (1).

[0020] In a preferred embodiment, the step 2 is specifically as follows:

[0021] Power mean:

[0022] Power variance:

[0023] In a preferred embodiment, the step three is specifically as follows: according to the power mean, power variance, and maximum and minimum power indexes of the user during the early morning, assuming that the number of clusters to be divided is X, and the number of indexes considered for each cluster center is Y, the corresponding fuzzy matrix is ​​established as follows:

[0024]

[0025] In the formula, a1, a2, ..., a X is the number of clusters, a X is the number of clustering indicators of the i-th cluster, w x1 ,w x2 ,...,w xY for a i Corresponding to Y indicator coefficients.

[0026] In a preferred embodiment, the step 4 is specifically as follows: calculating the density ρ of all power data in the power data set X (w Xy ,D Xav ),ρ X Defined as w Xy Centered, D Xav The sample area with the radius of the largest power consumption data density value ρ Xmax As the first cluster center, the expression of the corresponding radius is:

[0027]

[0028] Where x,z∈X.

[0029] In a preferred embodiment, the step 5 is specifically as follows: calculating the average distance density value d of the power consumption data X , specifically:

[0030]

[0031] In a preferred embodiment,

[0032] The specific step six is: if w Xy For applied power data density ρ Xmax is the maximum value, then it is defined as maxd xy If w Xy The corresponding power consumption data density ρ Xmax If it is not the maximum value, it is defined as mind xy , that is, C X ∈[mind xy ,maxd xy ].

[0033] In a preferred embodiment, the step eight is specifically as follows: updating the data center to ∑mind' xy The minimum is the new cluster center. According to the new cluster center, K-medoids clustering is used to cluster the probability distribution curve. The specific steps are:

[0034] Calculate electricity consumption data to ∑mind' xy The distance D(d i ' nx ), the specific expression is as follows:

[0035]

[0036] Where: d i ' n xy is the data value in the xth cluster; ∑mind′ xy is the cluster center distance;

[0037] Select d i ' n xy The average As the new cluster center, repeat the above formula.

[0038] In a preferred embodiment, the similar day in step 12 is the same time period as the calculated date attributes but without using air conditioning; the similar day is selected from the cluster with the lowest electricity clustering load in step 10, and the similar day is also extended to the whole day. Combined with the air conditioning electricity consumption characteristics and power duration information in step 11, the air conditioning electricity consumption is excluded to obtain the total electricity load at the same time as the calculated day without air conditioning electricity consumption.

[0039] In a preferred embodiment, the average value of the highest power period in step 13 is calculated as follows:

[0040]

[0041] Where N is the number of collection points corresponding to the highest power usage period including air conditioning power consumption, P j The power value corresponding to the number of collection points.

[0042] Compared with the prior art, the present invention has the following beneficial effects: Through the present invention, a method for mining installed capacity of residential air conditioners based on cluster analysis can be realized. By extracting the electricity consumption data of users in the early morning in summer, the K-medoids clustering algorithm is used to cluster the user power data into 4 categories, extract the corresponding clusters containing air-conditioning electricity consumption and the corresponding clusters not containing air-conditioning electricity consumption, extend the corresponding power data of the two clusters to the whole day, and use the similar day algorithm to further obtain the maximum value of the user's air-conditioning load, and then calculate the installed capacity of the household air conditioner of the resident user. The patented method can be used to calculate the total installed capacity of air conditioners for residential users in a community in summer, making great contributions to peak shaving and valley filling of the power system, maintaining the stability of the power system, and providing auxiliary services. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of the implementation of the preferred embodiment of the present invention;

[0044] Figure 2 Schematic diagram of K-medoids clustering results of a preferred embodiment of the present invention, wherein (a) is the overall clustering diagram, and (b) is the sub-diagram corresponding to each cluster label;

[0045] Figure 3 1 is a user power curve diagram of a preferred embodiment of the present invention, wherein (a) is a power curve diagram including air conditioning, and (b) is a power curve diagram without air conditioning on a similar day. DETAILED DESCRIPTION

[0046] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0047] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0048] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application; as used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or their combinations.

[0049] The present invention provides a method for mining installed capacity of residential air conditioners based on K-medoids cluster analysis. Figure 1 , which mainly includes the following steps:

[0050] Step 1: Obtain the electricity consumption data of residential users through the electricity consumption information collection system. Considering that ordinary residential users do not use other high-energy-consuming electrical appliances except air conditioners at night, it is possible to analyze whether there is air conditioning use behavior based on the electricity consumption behavior of residential users in the early morning. Collect the power data of residential users in the early morning (0:00-3:00) from July to October. After data cleaning and filling the missing values ​​of the data, the power values ​​of 12 measurement points of users in the early morning are obtained, namely:

[0051] P={P1,P2,...,P 12} (1)

[0052] Step 2: Calculate the power mean and power variance of the 12 power measurement points during the early morning hours:

[0053] Power mean:

[0054] Power variance:

[0055] Step 3: Based on the power mean, power variance, maximum and minimum power values ​​of the user during the early morning, assuming that the number of clusters to be divided is X, and the number of indicators considered for each cluster center is Y, the corresponding fuzzy matrix is ​​established as follows:

[0056]

[0057] In the formula, a1, a2, ..., a X is the number of clusters, a X is the number of clustering indicators of the i-th cluster, w x1 ,w x2 ,...,w xY for a i Corresponding to Y indicator coefficients.

[0058] Step 4: Calculate the density ρ of all power data in the power data set X (w Xy ,D Xav ),ρ X Defined as w Xy Centered, D Xav The sample area with the radius of the largest power consumption data density value ρ Xmax As the first cluster center, the expression of the corresponding radius is:

[0059]

[0060] Where x,z∈X.

[0061] Step 5: Calculate the average distance density value d of power consumption data X , specifically:

[0062]

[0063] Step 6: Distance C for different clusters X For classification, if w Xy For applied power data density ρ Xmax is the maximum value, then it is defined as maxd xy If w Xy The corresponding power consumption data density ρ Xmax If it is not the maximum value, it is defined as mind xy , that is, C X ∈[mind xy ,maxd xy ].

[0064] Step 7: Repeat steps 4 to 5 according to the power data set given in step 3, and terminate clustering when the cluster data set is an empty set.

[0065] Step 8: Update the data center to ∑mind' xy The minimum is the new cluster center. According to the new cluster center, K-medoids clustering is used to cluster the probability distribution curve. The specific steps are:

[0066] Calculate electricity consumption data to ∑mind' xy The distance D(d i ' nx ), the specific expression is as follows:

[0067] D(d i ' nx )=angmin||d i ' n xy -∑mind′ xy ||2 (7)

[0068] Where: d i ' n xy is the data value in the xth cluster; ∑mind′ xy is the cluster center distance.

[0069] Select d i ' n xy The average As the new cluster center, repeat the above formula.

[0070] Step 9: According to the set minimum threshold φ xy , when the cluster center is within the threshold σ x When the operating range is within the range and the change is small, that is, φ xy≤σ x When , this center is considered as the cluster center, and the cluster center sequence C'={C1,C2,...,C x}.

[0071] Step 10: Extract clusters with different cluster centers, calculate the average value of the power curves between the clusters, and preliminarily divide the power cluster load into: power consumption of other electrical appliances at night, power consumption of air conditioners, different power consumption of air conditioners due to different temperatures, etc.

[0072] Step 11: After the suspected "dates containing air-conditioning electricity consumption" and clusters are extracted in step 10, the analysis of the dates suspected of containing air-conditioning is extended to the whole day. The purpose is to calculate the maximum power that the air-conditioning can reach when it is used all day, and combine the air-conditioning electricity consumption characteristics, power duration and other information (that is, the power used by normal users when using air-conditioning ranges from 1kw to 3kw, and the duration is more than one hour as the air-conditioning use criterion), to obtain the total power load containing air-conditioning electricity consumption.

[0073] Step 12: Use similar days to calculate the power load of other electrical appliances without air conditioning. Similar days are the same period as the calculation date attributes but without air conditioning (for example, the calculation day and the similar day are both working days or both weekends). Similar days are selected from the cluster with the lowest power clustering load in step 10. Similarly, similar days are extended to the whole day. Combined with the power characteristics, power duration and other information of air conditioning in step 11, after excluding air conditioning power, the total power load without air conditioning power at the same time as the calculation day is obtained.

[0074] Step 13: Subtract the 'total power load including air conditioning' calculated in step 11 from the 'total power load excluding air conditioning' in step 12 to get the air conditioning load value for each period, extract the maximum value among all periods, take the average value of the period as the air conditioning load of the residential user, and compare it with the commonly used air conditioning power consumption characteristics on the market to estimate the installed capacity of the residential user's household air conditioning. The average value of the highest power period is calculated as follows:

[0075]

[0076] Where N is the number of collection points corresponding to the highest power usage period including air conditioning power consumption, P j The power value corresponding to the number of collection points.

[0077] Specifically:

[0078] (1) Cluster the power curves of users in the early morning (0:00-3:00) from July to October. The clustering results are shown in the following figure. The horizontal axis in the figure is the 12 sampling points corresponding to the user's 0:00-3:00, and the vertical axis is the power (kw). Each curve is the power curve of the corresponding period of each day. Figure 2(a) is the overall clustering diagram, Figure 2 (b) is the corresponding sub-graph of each cluster label obtained. It can be found that the pink one is the basic power load of household appliances at night. The curves corresponding to the other three types of labels may be the power consumption of other appliances at night, the power consumption of air conditioners, and the different power consumption of air conditioners due to different temperatures, which need further analysis.

[0079] (2) By comparing the power curves of users on July 2 and July 9 (similar days, both Fridays), we can extract the time periods that may be used by air conditioners. By subtracting the load curves of the same time period, we can get the power consumption of the air conditioner. That is, the total load of the time period minus the basic load of other electrical appliances in the same time period can be used to get the power load of the air conditioner. For example, Figure 3 The result after subtracting the two is about 1.1kw. Compared with the common air-conditioning power parameters on the market, it is about the power consumption of a 1.5-horsepower air-conditioner.

Claims

1. A method for mining installed capacity of residential air conditioners based on K-medoids cluster analysis, characterized in that The following steps are involved: Step 1: Obtain electricity consumption data of residential users through the electricity consumption information collection system; Step 2: Calculate the power mean and power variance of the 12 power measurement points during the early morning hours; Step 3: Establish the corresponding fuzzy matrix according to the power mean, power variance, and maximum and minimum power indicators of the user during the early morning hours; Step 4: Calculate power data; Step 5: Calculate the average distance density value of power consumption data; Step 6: Distance C for different clusters X To classify; Step 7: Repeat steps 4 to 5 according to the power data set given in step 3, and terminate clustering when the clustering data set is an empty set; Step 8: Update the data center; Step 9: According to the set minimum threshold φ xy , when the cluster center is within the threshold σ x When the operating range is within the range and the change is small, that is, φ xy ≤σ x When , this center is considered as the cluster center, and the cluster center sequence C'={C1,C2,...,C x }; Step 10: Extract clusters with different cluster centers, calculate the average value of the power curves between clusters, and preliminarily divide the power cluster load into: power consumption of other electrical appliances at night, power consumption of air conditioners, and different power consumption of air conditioners due to different temperatures; Step 11: After extracting the suspected 'dates containing air conditioning electricity consumption' and clusters in step 10, the analysis of the dates suspected to contain air conditioning is extended to the whole day to obtain the total electricity load containing air conditioning electricity consumption; Step 12: Use similar days to calculate the power load of other electrical appliances without air conditioning; Step 13: Subtract the 'total power load including air conditioning' calculated in step 11 from the 'total power load excluding air conditioning' in step 12 to obtain the air conditioning load value for each period, extract the maximum value among all periods, take the average value of the period as the air conditioning load of the residential user, and compare it with the commonly used air conditioning power consumption characteristics on the market to estimate the installed capacity of the residential user's household air conditioning; The step three is specifically as follows: according to the power mean, power variance, and maximum and minimum power indicators of the user during the early morning, assuming that the number of clusters to be divided is X, and the number of indicators considered for each cluster center is Y, the corresponding fuzzy matrix is ​​established as follows: In the formula, a1, a2, ..., a X is the number of clusters, a X is the number of clustering indicators of the i-th cluster, w x1 ,w x2 ,...,w xY for a i Corresponding to Y indicator coefficients; The step 4 is specifically as follows: Calculate the density ρ of all power data in the power data set X (w xy ,D Xav ),ρ X Defined as w Xy Centered, D Xav The sample area with the radius of the largest power consumption data density value ρ Xmax As the first cluster center, the expression of the corresponding radius is: In the formula, x,z∈X, X represents the number of clusters; The step eight is specifically as follows: updating the data center with ∑mind' xy The minimum is the new cluster center. According to the new cluster center, K-medoids clustering is used to cluster the probability distribution curve. The specific steps are: Calculate electricity consumption data to ∑mind' xy The distance D(d′ inx ), the specific expression is as follows: D(d′ inx )=angmin||d′ inxy -∑mind′ xy ||2 (7) Where: d′ inxy is the data value in the xth cluster; ∑mind′ xy is the cluster center distance; Select d i ' nxy The average As the new cluster center, repeat the above formula.

2. The method for mining installed capacity of residential air conditioners based on K-medoids cluster analysis according to claim 1, characterized in that: The step 1 is specifically as follows: analyzing whether there is air conditioning usage behavior based on the residential users' early morning electricity usage behavior; collecting the power data of residential users during the early morning from July to October, and after data cleaning and filling in the missing values ​​of the data, obtaining the power values ​​of 12 measurement points of the users during the early morning, namely: P={P1,P2,...,P 12 } (1)。 3. The method for mining installed capacity of residential air conditioners based on K-medoids cluster analysis according to claim 2, characterized in that: The step 2 is specifically as follows: Power mean: Power variance:

4. The method for mining installed capacity of residential air conditioners based on K-medoids cluster analysis according to claim 3, characterized in that: The step 5 is specifically as follows: calculating the average distance density value d of the power consumption data X , specifically:

5. The method for mining installed capacity of residential air conditioners based on K-medoids cluster analysis according to claim 1, characterized in that: The specific step six is: if w Xy For applied power data density ρ Xmax is the maximum value, then it is defined as maxd xy If w Xy The corresponding power consumption data density ρ Xmax If it is not the maximum value, it is defined as mind xy , that is, C X ∈[mind xy ,maxd xy ].

6. The method for mining installed capacity of residential air conditioners based on K-medoids cluster analysis according to claim 1, characterized in that: The similar day in step 12 is the same time period as the calculated date attributes but without using the air conditioner; the similar day is selected from the cluster with the lowest electricity clustering load in step 10, and the similar day is also extended to the whole day. Combined with the air conditioning electricity consumption characteristics and power duration information in step 11, the air conditioning electricity consumption is excluded to obtain the total electricity load excluding the air conditioning electricity consumption at the same time as the calculated day.

7. The method for mining installed capacity of residential air conditioners based on K-medoids cluster analysis according to claim 1, characterized in that: The average value of the highest power period in step 13 is calculated as follows: Where N is the number of collection points corresponding to the highest power usage period including air conditioning power consumption, P j The power value corresponding to the number of collection points.

Images