Daily reactive load curve clustering method and system based on characteristic index dimension reduction

Abstract

The invention provides a daily reactive load curve clustering method and system based on characteristic index dimensionality reduction. The method comprises the following steps: taking a load rate, a highest utilization hour rate, a daily peak-valley difference rate, a peak period load rate, maximum load occurrence time, a flat period load rate, a valley period load rate and minimum load occurrence time as daily reactive load characteristic indexes; carrying out daily reactive load data dimension reduction based on the daily reactive load characteristic indexes, obtaining a daily reactive load characteristic index data dimension reduction matrix, and carrying out weight assignment on the daily reactive load characteristic indexes by adopting an expert weighting method; and carrying out clustering calculation based on a k-means clustering algorithm and outputting a daily reactive load curve clustering result. According to the method, the established daily reactive load characteristic index system is applied to daily reactive load clustering calculation, the calculation amount is effectively reduced, the problem that the clustering effect is not ideal under the high-dimensional condition of data is avoided, curve clustering can be accurately carried out on the daily reactive load, and the method has a guiding effect on making a load scheduling plan and operation control.

Description

Technical Field

[0001] The invention belongs to the technical field of data processing, and in particular relates to a daily reactive load curve clustering method and system based on feature index dimension reduction. Background Art

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

[0003] When studying the load characteristics of power systems, one of the most important studies is the acquisition of typical daily load curves. Typical daily load curves can be used to analyze the characteristics of regional load power consumption and obtain the maximum (minimum) load and average load of a typical day, which is of great significance for the formulation of load scheduling and operation control. However, the inventors have found that the current research on daily load curves is mainly focused on active loads, and there are few studies on reactive load curves, and there are the following problems:

[0004] There are three traditional methods for selecting typical daily active load curves, namely: 1) selecting the daily load curve that is closest to the monthly average load rate and has no distortion as the typical daily load curve of the month; 2) selecting the daily load curve with the largest load in the whole month as the daily load curve of the month; 3) selecting a fixed working day of the month as the typical daily load curve of the month. However, the above methods all have certain limitations and cannot fully reflect the load characteristics over a period of time.

[0005] For example, when selecting the daily load curve that is closest to the monthly average load rate and has no distortion as the typical daily load curve of the month, on the one hand, selecting only one (even the closest one) daily load curve as the typical representative of the entire month cannot fully reflect the diversity and variability of the load within the month. Because in real life, even if the monthly average load rate is similar, there may be significant differences in the load curves on different dates, such as holidays, special weather, changes in economic activities, etc., which may affect the load; on the other hand, if the selected daily load curve does not contain or rarely contains extreme load conditions (such as the highest load day, the lowest load day, etc.), then the reference value of the typical daily load curve in dealing with extreme loads will be limited.

[0006] In addition, since the k-means clustering algorithm is a powerful means to organize and manage messy data, it is effectively applied to load curve clustering calculation. However, the reactive load curve is multi-dimensional data, and the direct application of conventional clustering methods for clustering is often not ideal. Therefore, it is necessary to take appropriate measures to reduce the dimension of the reactive load curve data. Summary of the invention

[0007] In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a daily reactive load curve clustering method and system based on feature index dimensionality reduction, which can accurately perform curve clustering on the daily reactive load.

[0008] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0009] A first aspect of the present invention provides a daily reactive load curve clustering method based on feature index dimensionality reduction, comprising:

[0010] Obtain the original data of annual reactive load of substation;

[0011] At least eight indicators, including load rate, maximum utilization hour rate, daily peak-to-valley difference rate, peak load rate, maximum load occurrence time, normal load rate, valley load rate and minimum load occurrence time, are used as daily reactive load characteristic indicators;

[0012] Based on the daily reactive load characteristic index, the daily reactive load data is reduced in dimension and the daily reactive load characteristic index data dimension reduction matrix is ​​obtained, and the daily reactive load characteristic index is assigned weights by using the expert weighting method;

[0013] The k-means clustering algorithm is used to perform clustering calculation on the reduced-dimensional daily reactive load data. When the clustering end judgment condition is met, the clustering is stopped and the clustering result of the daily reactive load curve is output.

[0014] A second aspect of the present invention provides a daily reactive load curve clustering system based on feature index dimensionality reduction, comprising:

[0015] The data acquisition module is configured to obtain the original data of the annual reactive load of the substation.

[0016] The characteristic index selection module is configured to use load rate, maximum utilization hour rate, daily peak-to-valley difference rate, peak load rate, maximum load occurrence time, normal load rate, valley load rate and minimum load occurrence time as daily reactive load characteristic indicators.

[0017] The data processing module is configured to: perform dimension reduction of daily reactive load data based on daily reactive load characteristic indicators and obtain a dimension reduction matrix of daily reactive load characteristic indicator data, and assign weights to daily reactive load characteristic indicators using an expert weighting method.

[0018] The clustering module is configured to: perform clustering calculation on the reduced-dimensional daily reactive load data based on the k-means clustering algorithm, and when the judgment condition for the end of clustering is met, stop clustering and output the clustering result of the daily reactive load curve.

[0019] The third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in a daily reactive load curve clustering method based on feature index dimensionality reduction as described in the first aspect of the present invention.

[0020] The fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps in a daily reactive load curve clustering method based on feature index dimensionality reduction as described in the first aspect of the present invention are implemented.

[0021] One or more of the above technical solutions have the following beneficial effects:

[0022] (1) The present invention reduces the dimension of reactive load data based on eight indicators that best characterize the load characteristics of daily reactive load data, taking into account the multi-dimensional characteristics of the reactive load curve and avoiding the problem of unsatisfactory clustering effect caused by direct clustering; at the same time, on the basis of dimensionality reduction, the simple, high-precision and fast k-means clustering algorithm is selected for the final clustering operation, which has good clustering effect and higher efficiency, can accurately cluster the daily reactive load curve, and has a guiding role in the formulation of load scheduling plans and operation control.

[0023] (2) The present invention applies eight load characteristic indicators to the daily reactive load clustering calculation, which effectively reduces the amount of calculation and makes up for the problem that the prior art only analyzes the load characteristics of the active load curve while ignoring the analysis of the load characteristics related to the reactive load.

[0024] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0026] Figure 1 This is a flow chart of a daily reactive load curve clustering method based on feature index dimensionality reduction in Example 1.

[0027] Figure 2 This is a calculation flow chart of the k-means clustering algorithm in Example 1.

[0028] Figure 3 This is a module schematic diagram of a daily reactive load curve clustering system based on feature index dimensionality reduction in Example 2. DETAILED DESCRIPTION

[0029] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. 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 invention belongs.

[0030] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0031] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0032] The present invention provides a daily reactive load curve clustering method and system based on feature index dimension reduction, and the overall idea is:

[0033] Obtain the original data of the annual reactive load of the substation; use at least eight indicators including load rate, maximum utilization hour rate, daily peak-to-valley difference rate, peak load rate, maximum load occurrence time, normal load rate, valley load rate and minimum load occurrence time as daily reactive load characteristic indicators. Perform dimensionality reduction on the daily reactive load data based on the daily reactive load characteristic indicators and obtain the dimensionality reduction matrix of the daily reactive load characteristic indicator data, and use the expert weighting method to assign weights to the daily reactive load characteristic indicators. Perform clustering calculation on the reduced-dimensional daily reactive load data based on the k-means clustering algorithm. When the judgment condition for the end of clustering is met, stop clustering and output the clustering result of the daily reactive load curve. Through the above means, the present invention can accurately perform curve clustering on the daily reactive load.

[0034] To facilitate understanding of the technical solution of the present invention, the present invention is further described below in conjunction with embodiments.

[0035] Embodiment 1

[0036] like Figure 1 As shown, this embodiment provides a daily reactive load curve clustering method based on feature index dimension reduction, which can be implemented by the following process:

[0037] Step S1, obtaining the original data of the annual reactive load of the substation.

[0038] When actually obtaining the annual reactive load raw data of the substation, the reactive load data of the substation for one year is collected. Specifically, the sampling interval is set to 15 minutes, and 96 reactive load data are collected every day, and then the annual reactive load raw data matrix is ​​obtained, that is:

[0039]

[0040] Among them, X i represents the reactive load data set of day i; xij represents the reactive load data at the jth time point on the i-th day; a represents the total number of days in the year in which the annual reactive load data is located, and b represents the number of reactive load data collected in one day, so a=365, b=96.

[0041] Step S2: Select daily reactive load characteristic index.

[0042] The load characteristics of a power system refer to the characteristics and properties of the power load, which can be reflected through relevant characteristic indicators. At present, there is no unified standard for the load characteristic indicators of power systems internationally. Generally, 15 indicators such as load rate, peak-to-valley difference, and average load are used to analyze the characteristics and properties of various types of power loads.

[0043] The reactive load curve has certain similarities with the active load curve, both of which have characteristics such as load rate, peak-to-valley difference rate, and peak load rate. Therefore, in this embodiment, the load rate, maximum utilization hour rate, daily peak-to-valley difference rate, peak load rate, maximum load occurrence time, normal load rate, valley load rate, and minimum load occurrence time are used as daily reactive load characteristic indicators.

[0044] The daily reactive load characteristic index is a characteristic index in different time periods, as shown in Table 1:

[0045] Table 1 Daily reactive load characteristic indicators

[0046]

[0047] Among them, Q represents reactive load; t represents time; subscripts sum, av, max, and min represent total, mean, maximum, and minimum values ​​respectively; superscripts peak, val, and sh represent peak period, valley period, and flat period respectively.

[0048] Based on the obtained raw data of the substation's annual reactive load, the average reactive load value of each period of the day is calculated throughout the year; and the whole day is divided into peak period, flat period and valley period according to the size of the average value.

[0049] Specifically, in this embodiment, the two time periods of 08:00-11:00 and 18:00-21:00 in a day are taken as peak time periods, the three time periods of 06:00-08:00, 11:00-18:00 and 21:00-22:00 in a day are taken as normal time periods, and the two time periods of 22:00-24:00 and 00:00-06:00 in a day are taken as valley time periods.

[0050] As can be seen from Table 1, the load rate, the highest utilization hour rate and the daily peak-to-valley difference rate are the daily reactive load characteristic indicators during the whole day, the peak load rate and the time when the maximum load occurs are the daily reactive load characteristic indicators during the peak period, the flat load rate is the daily reactive load characteristic indicator during the flat period, and the valley load rate and the time when the minimum load occurs are the daily reactive load characteristic indicators during the valley period.

[0051] Step S3: based on the daily reactive load characteristic index, perform dimension reduction on the daily reactive load data and obtain the dimension reduction matrix of the daily reactive load characteristic index data, and use the expert weighting method to assign weights to the daily reactive load characteristic index.

[0052] Specifically, the dimension reduction matrix of the daily reactive load characteristic index data is expressed as:

[0053]

[0054] Among them, Y represents the dimension reduction matrix of daily reactive load characteristic index data, a represents the total number of days in the year where the annual reactive load data is located, and Y a represents the reactive load characteristic index dimension reduction data set of day a, c represents the number of daily reactive load characteristic indexes, i.e. c = 8, y ac It represents the reactive load characteristic index value of the cth day on the ath day.

[0055] From the dimensionality reduction matrix of daily reactive load characteristic index data, we can see that the daily reactive load data is represented by 8 daily reactive load characteristic indexes every day of the year instead of 96 sampled data every day of the year, that is, based on 365 days in a year, the dimensionality reduction matrix of daily reactive load characteristic index data reduces the original data of the substation's annual reactive load from a high-dimensional data matrix of 365*96 to a data matrix of 365*8 dimensions, greatly reducing the dimension of the original data of the substation's annual reactive load.

[0056] The expert weighting method is used to assign weights to the daily reactive load characteristic indicators, and the obtained daily reactive load characteristic indicator weight matrix is:

[0057]

[0058] Among them, W represents the daily reactive load characteristic index weight matrix, W m represents the daily reactive load characteristic index dimension reduction weight dataset of the mth expert, d represents the number of invited experts, c represents the number of daily reactive load characteristic indexes, w ml represents the weight assigned by the mth expert to the reactive load characteristic index of the lth day, satisfying

[0059] The characteristic index weight average of all experts is calculated based on the obtained daily reactive load characteristic index weight matrix. The characteristic index weight average is composed of the weight average assigned by all experts to each daily reactive load characteristic index, that is:

[0060]

[0061] in, It represents the weighted average given by all experts for the reactive load characteristic index on the lth day.

[0062] Finally, the weighted mean matrix of feature indicators of all experts is calculated, namely:

[0063]

[0064] Among them, W ave represents the weight mean matrix of characteristic indicators of all experts, and c represents the number of daily reactive load characteristic indicators.

[0065] Step S4: Perform clustering calculation on the reduced-dimensional daily reactive load data based on the k-means clustering algorithm. When the judgment condition for the end of clustering is met, stop clustering and output the clustering result of the daily reactive load curve.

[0066] Set the number of clusters k (k is an integer, Euclidean distance is used as an indicator to measure the similarity between data objects.

[0067] like Figure 2 As shown, when the k-means clustering algorithm is used for clustering calculation, the following steps are performed:

[0068] 1) Let k = 1, and use the reactive load characteristic index to reduce the dimension of the data Y and the expert characteristic index weight W ave As input data;

[0069] 2) Execute k=k+1, and then randomly generate k initial cluster centers, that is:

[0070]

[0071] Among them, Z n represents the nth cluster center; z nl Represents the lth characteristic index value of the nth cluster center.

[0072] 3) Calculate the relationship between each data object and each cluster center Z by the following formula: n The weighted Euclidean distance is:

[0073]

[0074] 4) Reclassify each data object according to the minimum distance principle;

[0075] 5) Calculate various means and update various data centers;

[0076] 6) Calculate the sum of squared errors (SSE) of the entire data set, namely:

[0077]

[0078] Where a represents the total number of days in the year where the reactive load data is located, k represents the number of clusters, and dis(Y i Z n ,) represents the dimension reduction dataset Y of the daily reactive load characteristic index of the i-th day i With the nth cluster center Z n The weighted Euclidean distance of .

[0079] 7) Determine whether SSE has changed? If yes, return to step 3; if no, proceed to the next step.

[0080] 8) Judgment ? If yes, return to step 2; if no, proceed to the next step.

[0081] 9) Generate a curve of SSE changing with k value, take the "elbow point" in the curve of the change of error square sum SSE with k value as the optimal clustering number and output the clustering result.

[0082] Embodiment 2

[0083] This embodiment discloses a daily reactive load curve clustering system based on feature index dimensionality reduction.

[0084] like Figure 3 As shown, a daily reactive load curve clustering system based on feature index dimension reduction includes:

[0085] The data acquisition module is configured to obtain the original data of the annual reactive load of the substation.

[0086] The characteristic index selection module is configured to use load rate, maximum utilization hour rate, daily peak-to-valley difference rate, peak load rate, maximum load occurrence time, normal load rate, valley load rate and minimum load occurrence time as daily reactive load characteristic indicators.

[0087] The data processing module is configured to: perform dimension reduction of daily reactive load data based on daily reactive load characteristic indicators and obtain a dimension reduction matrix of daily reactive load characteristic indicator data, and assign weights to daily reactive load characteristic indicators using an expert weighting method.

[0088] The clustering module is configured to: perform clustering calculation on the reduced-dimensional daily reactive load data based on the k-means clustering algorithm, and when the judgment condition for the end of clustering is met, stop clustering and output the clustering result of the daily reactive load curve.

[0089] Embodiment 3

[0090] The purpose of this embodiment is to provide a computer-readable storage medium.

[0091] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in a daily reactive load curve clustering method based on feature index dimensionality reduction as described in Example 1 of the present disclosure.

[0092] Embodiment 4

[0093] The purpose of this embodiment is to provide an electronic device.

[0094] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps in a daily reactive load curve clustering method based on feature index dimensionality reduction as described in Example 1 of the present disclosure are implemented.

[0095] The steps involved in the apparatuses of the above embodiments 2, 3 and 4 correspond to the method embodiment 1, and the specific implementation methods can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0096] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0097] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A daily reactive load curve clustering method based on feature index dimension reduction, characterized in that: include: Obtain the original data of annual reactive load of substation; At least eight indicators, including load rate, maximum utilization hour rate, daily peak-to-valley difference rate, peak load rate, maximum load occurrence time, normal load rate, valley load rate and minimum load occurrence time, are used as daily reactive load characteristic indicators; Based on the daily reactive load characteristic index, the daily reactive load data is reduced in dimension and the daily reactive load characteristic index data dimension reduction matrix is ​​obtained, and the daily reactive load characteristic index is assigned weights by using the expert weighting method; The k-means clustering algorithm is used to perform clustering calculation on the reduced-dimensional daily reactive load data. When the clustering end judgment condition is met, the clustering is stopped and the clustering result of the daily reactive load curve is output.

2. A daily reactive load curve clustering method based on characteristic index dimension reduction as claimed in claim 1, characterized in that: Based on the obtained raw data of the substation's annual reactive load, the average reactive load value of each period of the day is calculated throughout the year; and the whole day is divided into peak period, flat period and valley period according to the size of the average value.

3. A daily reactive load curve clustering method based on characteristic index dimension reduction as claimed in claim 2, characterized in that: The daily reactive load characteristic index is the characteristic index in different time periods, specifically: The load rate, maximum utilization hour rate and daily peak-to-valley difference rate are the daily reactive load characteristic indicators during the whole day; the peak load rate and the time when the maximum load occurs are the daily reactive load characteristic indicators during the peak period; the normal load rate is the daily reactive load characteristic indicator during the normal period; the valley load rate and the time when the minimum load occurs are the daily reactive load characteristic indicators during the valley period.

4. A daily reactive load curve clustering method based on characteristic index dimension reduction as claimed in claim 1, characterized in that: The expert weighting method is used to assign weights to the daily reactive load characteristic indicators, and the obtained daily reactive load characteristic indicator weight matrix is: Among them, W represents the daily reactive load characteristic index weight matrix, W m represents the daily reactive load characteristic index dimension reduction weight dataset of the mth expert, d represents the number of invited experts, c represents the number of daily reactive load characteristic indexes, w ml represents the reactive load characteristic index weight of the mth expert on the lth day, satisfying 5. A daily reactive load curve clustering method based on characteristic index dimension reduction as claimed in claim 4, characterized in that: The characteristic index weight average of all experts is calculated based on the obtained daily reactive load characteristic index weight matrix, and the characteristic index weight average is composed of the weight average assigned by all experts to each daily reactive load characteristic index.

6. A daily reactive load curve clustering method based on characteristic index dimension reduction as claimed in claim 1, characterized in that: The sum of square errors (SSE) of the entire daily reactive load data set is used as the criterion for the end of clustering.

7. A daily reactive load curve clustering method based on characteristic index dimension reduction as claimed in claim 1, characterized in that: When the clustering end judgment condition is met, clustering is stopped and the "elbow point" in the change curve of the error sum square SSE with the change of k value is used as the optimal cluster number and the clustering result is output.

8. A daily reactive load curve clustering system based on feature index dimension reduction, characterized in that: include: The data acquisition module is configured to: obtain the original data of annual reactive load of the substation; The characteristic index selection module is configured to use load rate, maximum utilization hour rate, daily peak-to-valley difference rate, peak load rate, maximum load occurrence time, normal load rate, valley load rate and minimum load occurrence time as daily reactive load characteristic indexes; The data processing module is configured to: perform dimension reduction of daily reactive load data based on daily reactive load characteristic indicators and obtain a dimension reduction matrix of daily reactive load characteristic indicator data, and assign weights to daily reactive load characteristic indicators using an expert weighting method; The clustering module is configured to: perform clustering calculation on the reduced-dimensional daily reactive load data based on the k-means clustering algorithm, and when the judgment condition for the end of clustering is met, stop clustering and output the clustering result of the daily reactive load curve.

9. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the steps of a daily reactive load curve clustering method based on feature index dimensionality reduction as described in any one of claims 1 to 7 are implemented.

10. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the daily reactive load curve clustering method based on feature index dimensionality reduction as described in any one of claims 1 to 7 are implemented.

Images