A method for identifying the power load status of a power grid based on a neural network

Abstract

The present invention relates to the technical field of power grid power load voltage adjustment, and specifically relates to a method for identifying the power load state of a power grid based on a neural network. This method obtains the power load time series curve of power users, screens out the target curve according to the changes, and obtains the overall change degree based on the difference between the high power load data and the low power load data on the target curve and the change of the target curve within a specified time period; obtains the reference category based on the overall change degree; obtains the peak change degree according to the distribution of peaks on the target curve within the reference category, and determines the types of power users; determines the current power grid power load state according to the power load change of each type of power user within a preset historical time period. The present invention accurately classifies power users by determining the types of power users, and then accurately analyzes the current power grid power load state, accurately adjusts it, and ensures the stability of the voltage.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid power load voltage regulation, and particularly relates to a method for identifying the power load state of a power grid based on a neural network. Background Art

[0002] In order to ensure voltage stability and guarantee the stable operation of each electrical appliance, the voltage can be stabilized by adjusting the power load in the power grid. In actual situations, the power load conditions of different power user groups are different. For example, there are obvious differences in the daily power loads of residential power users and heavy industry power users. In order to accurately adjust the power load in the power grid, different types of power users need to be accurately classified, and then, according to the power load change conditions of each type of power user, the power load in the power grid can be accurately adjusted in real time to ensure the voltage stability in the power grid in real time.

[0003] In the existing methods, clustering is performed by obtaining the power load time series curve of each power user within a preset time period, and various power users are determined according to the clustering results. However, in actual situations, the power load time series curves of different types of power users may be similar. Directly clustering according to the similarity of the power load time series curves is likely to result in inaccurate division of the power load time series curves, and thus various power users cannot be accurately distinguished, leading to inaccurate adjustment of the power load in the power grid, and further unable to ensure voltage stability and guarantee the stable operation of electrical appliances. Summary of the Invention

[0004] In order to solve the technical problem that directly clustering according to the similarity of the power load time series curves is likely to result in inaccurate division of the power load data curves, and thus various power users cannot be accurately distinguished, leading to inaccurate adjustment of the power load in the power grid and further unable to ensure voltage stability, the purpose of the present invention is to provide a method for identifying the power load state of a power grid based on a neural network, and the specific technical solution adopted is as follows:

[0005] The present invention provides a method for identifying the power load state of a power grid based on a neural network, and the method includes the following steps:

[0006] Obtain the power load time series curve of each power user within a preset time period; according to the change condition of the power load time series curve, screen out the target curve;

[0007] Determine the high-power load data and low-power load data on each target curve according to the magnitude of the power load data on each target curve; obtain the overall change degree of each target curve according to the difference between the high-power load data and the low-power load data on each target curve, as well as the change situation of each target curve within the specified time period; cluster the target curves based on the overall change degree to obtain target curve categories;

[0008] According to the overall change degree of the target curves within each target curve category, screen out the reference category; obtain the peak change degree of each target curve within the reference category according to the distribution of the peaks on each target curve within the reference category; determine the types of each power user within the reference category based on the peak change degree;

[0009] Determine the current power grid load status according to the power load change situation of each type of power user within the preset historical time period.

[0010] Furthermore, the method for obtaining the target curves is as follows:

[0011] Obtain the variance of the power load data on each power load time series curve as the fluctuation value of each power load time series curve;

[0012] When the fluctuation value is greater than the preset fluctuation value threshold, regard the corresponding power load time series curve as the target curve.

[0013] Furthermore, the method for obtaining the high-power load data and the low-power load data is as follows:

[0014] For any target curve, cluster the power load data on this target curve through the K-means clustering algorithm to obtain clustering clusters; where the value of K in the K-means clustering algorithm is set to 2;

[0015] Obtain the mean value of the power load data in each clustering cluster as the first mean value of the corresponding clustering cluster;

[0016] Regard all the power load data within the clustering cluster corresponding to the largest first mean value as the high-power load data of this target curve;

[0017] Regard all the power load data within the clustering cluster corresponding to the smallest first mean value as the low-power load data of this target curve.

[0018] Furthermore, the method for obtaining the overall change degree is as follows:

[0019] For any target curve, obtain the first change degree of this target curve according to the mean value difference and quantity difference between the high-power load data and the low-power load data on this target curve;

[0020] Obtain the second degree of change of the target curve according to the degree of fluctuation of the power load data within the first preset specified time period and the fluctuation situation of the power load data within the second preset specified time period; wherein, the degree of fluctuation is positively correlated with the second degree of change, and the fluctuation situation is negatively correlated with the second degree of change;

[0021] Obtain the overall degree of change of the target curve according to the first degree of change and the second degree of change of the target curve; wherein, both the first degree of change and the second degree of change are positively correlated with the overall degree of change.

[0022] Further, the method for obtaining the first degree of change is as follows:

[0023] For any target curve, take the mean difference between the high power load data and the low power load data on the target curve as the first difference;

[0024] Take the ratio of the quantity corresponding to the high power load data and the low power load data on the target curve as the first ratio;

[0025] Take the difference between the first ratio and the first preset constant as the second difference;

[0026] Obtain the first degree of change of the target curve according to the first difference and the second difference; wherein, the first difference is positively correlated with the first degree of change, and the second difference is negatively correlated with the first degree of change.

[0027] Further, the method for obtaining the reference category is as follows:

[0028] Obtain the mean value of the overall degree of change values of all target curves within each target curve category as the label of each target curve category;

[0029] Take the target curve category corresponding to the largest label as the reference category.

[0030] Further, the method for obtaining the peak degree of change is as follows:

[0031] For any target curve within the reference category, take the average curve corresponding to the upper and lower envelope curves of the target curve as the overall change curve of the target curve;

[0032] Obtain the peak degree of change of the target curve according to the difference between each maximum value and its previous adjacent maximum value on the overall change curve, the change rate of each maximum value, and the number of maximum values; wherein, the difference between the maximum value and its previous adjacent maximum value, and the number of maximum values are both positively correlated with the peak degree of change, and the change rate is negatively correlated with the peak degree of change.

[0033] Further, the method for obtaining the change rate is as follows:

[0034] For any maximum value in the overall change curve, obtain the average difference between the maximum value and each of its adjacent minimum values, and use it as the reference change rate of the maximum value;

[0035] Take the largest reference change rate as the change rate of the maximum value.

[0036] Further, the method for determining the type of each power user within the reference category based on the peak change degree is as follows:

[0037] When the peak change degree does not meet the preset peak change degree range, regard the power user corresponding to the target curve within the reference category as a light industry power user;

[0038] When the peak change degree meets the preset peak change degree range, regard the power user corresponding to the target curve within the reference category as a residential power user.

[0039] Further, the types of the power users include heavy industry power users, light industry power users, and residential power users.

[0040] The present invention has the following beneficial effects:

[0041] According to the change situation of the power load time series curve, the target curve is screened to improve the efficiency of classifying power users; furthermore, according to the magnitude of the power load data on each target curve, the high-power load data and low-power load data on each target curve are determined, which is beneficial to accurately analyze the change situation of each target curve in the follow-up; therefore, according to the difference between the high-power load data and the low-power load data on each target curve, as well as the change situation of each target curve within the specified time period, the overall change degree of each target curve is obtained, the change situation of each target curve is determined, the type of each power user is initially judged, and then the target curves are clustered based on the overall change degree to obtain the target curve categories, and the power users are classified to initially determine the type of each power user; in order to ensure the accuracy of power user classification, further according to the overall change degree of the target curves within each target curve category, the reference category is screened, the category where clustering errors may exist is determined, and the efficiency of accurately determining the type of power users is improved; further according to the peak distribution situation on each target curve within the reference category, the peak change degree of each target curve within the reference category is obtained, which accurately reflects the peak situation on each target curve within the reference category, and then based on the peak change degree, the type of each power user within the reference category is accurately determined, and the type of power users is accurately classified; furthermore, according to the power load change situation of each type of power user within the preset historical time period, the power load situation of the current power grid is accurately predicted, and then the power load state of the current power grid is accurately determined, and the power grid power load is adjusted in real time to ensure voltage stability and ensure the stable operation of electrical appliances. Brief Description of the Drawings

[0042] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0043] Figure 1 Flowchart of a method for identifying the power load state of a power grid based on a neural network provided by an embodiment of the present invention;

[0044] Figure 2 Flowchart of a method for obtaining the overall change degree provided by an embodiment of the present invention;

[0045] Figure 3 Flowchart of a method for obtaining the peak change degree provided by an embodiment of the present invention;

[0046] Figure 4Structural diagram of a power grid power load status identification system based on a neural network provided by an embodiment of the present invention;

[0047] Figure 5 Schematic diagram of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0048] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following combines the accompanying drawings and preferred embodiments to detail the specific implementation manners, structures, features and effects of a power grid power load status identification method based on a neural network proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0049] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0050] The following specifically describes the specific solution of a power grid power load status identification method provided by the present invention with reference to the accompanying drawings.

[0051] Embodiment 1:

[0052] The present invention proposes a power grid power load status identification method based on a neural network. Please refer to Figure 1 , which shows a flowchart of a power grid power load status identification method provided by an embodiment of the present invention. The method includes the following steps:

[0053] Step S1: Obtain the power load time series curve of each power user within a preset time period; according to the change of the power load time series curve, screen out the target curve.

[0054] Specifically, in this embodiment, taking an urban area as an example, the power load situation of the power grid in this urban area is analyzed, and then the power load in the corresponding power grid of this urban area is adjusted to ensure the voltage stability of this urban area. The implementer can set the analyzed area according to the actual situation, which is not limited here. It should be noted that the urban areas mentioned subsequently are all the urban areas selected in this embodiment. In this embodiment, power users in the urban area are divided into three major categories according to the usage of electricity, namely heavy industrial power users, light industrial power users, and residential power users. To determine the category of each power user in the urban area, in this embodiment, the power load data of each power user in the urban area at each moment within a preset time period is obtained, and the power load data of each power user at each moment within the preset time period is fitted into the power load time series curve of each power user. Among them, the method of curve fitting is a well-known technology and will not be elaborated here. In this embodiment, the preset time period is set to 1 day, that is, 24 hours, and the time interval between two adjacent moments is set to 1 hour. The implementer can set the size of the preset time period and the time interval between two adjacent moments according to the actual situation, which is not limited here. Thus, the power load time series curve of each power user is obtained.

[0055] In actual situations, heavy industries need to maintain a relatively high production level during the preset time period. Therefore, the power load time series curve of heavy industrial power users generally shows relatively stable characteristics, with a consistently high power load level. There must be fluctuations in the power load data of light industrial power users and residential power users within the preset time period. Because the power load data of light industrial power users is relatively large during working hours and relatively small during off-duty hours; generally, the power load data of residential power users is relatively large in the morning, at noon, and in the evening, and relatively small during other time periods. Therefore, heavy industrial power users are relatively easy to identify. In this embodiment, the variance of the power load data on each power load time series curve is obtained as the fluctuation value of each power load time series curve; the larger the fluctuation value, the more unstable the corresponding power load time series curve, and the more likely the corresponding power user is a light industrial power user or a residential power user. Therefore, when the fluctuation value is less than or equal to the preset fluctuation value threshold, the power user corresponding to the corresponding power load time series curve is regarded as a heavy industrial power user; when the fluctuation value is greater than the preset fluctuation value threshold, the corresponding power load time series curve is regarded as the target curve. Among them, the target curve is the power load time series curve corresponding to light industrial power users and residential power users. In this embodiment, the preset fluctuation value threshold is set to 0.2. The implementer can set the size of the preset fluctuation value threshold according to the actual situation, which is not limited here. Thus, heavy industrial power users are determined.

[0056] Interfered by seasonal factors, there are similar situations in the target curves corresponding to light industrial power users and residential power users. For example, in summer, due to the hot weather, residential power users keep using air conditioners during the day, resulting in the similarity between the target curves of residential power users and those of light industrial power users, which easily leads to the misidentification of residential power users as light industrial power users. In order to accurately classify residential power users and light industrial power users, this embodiment accurately analyzes the target curves, which is conducive to accurately classifying the types of power users, and then accurately determining the power grid load situation at the current moment, accurately adjusting the power grid load at the current moment, ensuring the stability of the urban area voltage at the current moment, and ensuring the stable operation of all working electrical appliances in real time.

[0057] Step S2: Determine the high-power load data and low-power load data on each target curve according to the magnitude of the power load data on each target curve; obtain the overall change degree of each target curve according to the mean difference and quantity difference between the high-power load data and the low-power load data on each target curve, as well as the change situation of each target curve within a specified time period; cluster the target curves based on the overall change degree to obtain target curve categories.

[0058] Specifically, in order to accurately classify residential power users and light industrial power users, this embodiment analyzes the change situation of each target curve. In actual situations, the general working hours of light industrial power users are from 9:00 to 18:00. During the working period, the power load data of light industrial power users is relatively large, and during the non-working period, the power load data of light industrial power users is relatively small and remains stable. Therefore, the difference between the power load data of light industrial power users during the working period and that during the non-working period is relatively large. For residential power users, the difference between the power load data within the preset time period is relatively small. Therefore, this embodiment first determines the high-power load data and low-power load data on each target curve according to the magnitude of the power load data on each target curve, and then preliminarily analyzes the types of power users corresponding to each target curve according to the mean difference and quantity difference between the high-power load data and the low-power load data on each target curve.

[0059] Considering the variation law of the power load data of light industrial power users, and then analyzing the variation of each target curve within the specified time period, further distinguishing residential power users and light industrial power users. Among them, the variation law of the power load data of light industrial power users is as follows: the peak period of light industrial power users is from 9:00 to 12:00, and the flat period is from 12:00 to 18:00. Due to office needs, the power load data of light industrial power users will increase significantly with time during the peak period; after entering the flat period, in order to maintain the operation of office equipment, the power load data of light industrial power users is very stable during the flat period. While the power load data of residential power users is unstable throughout the preset time period. Therefore, in this embodiment, the specified time period is respectively set as the first preset specified time period, that is, the peak period is from 9:00 to 12:00, and the second preset specified time period, that is, the flat period is from 12:00 to 18:00. When the fluctuation degree of a certain target curve is greater within the first preset specified time period and smaller within the second preset specified time period, it indicates that the power user corresponding to the target curve is more likely to be a light industrial power user.

[0060] Therefore, in this embodiment, according to the mean difference and quantity difference between the high power load data and the low power load data on each target curve, and the variation of each target curve within the specified time period, the overall variation degree of each target curve is obtained. Among them, the greater the overall variation degree, the more likely the power user corresponding to the target curve is a light industrial power user. Then, in this embodiment, the target curves are clustered based on the overall variation degree to obtain target curve categories. In this embodiment, the K-means clustering algorithm is used to cluster the target curves, where the value of K in the K-means clustering algorithm is set to 2. Therefore, two target curve categories are obtained, corresponding to light industrial power users and residential power users respectively. Among them, the K-means clustering algorithm is a well-known technology and will not be elaborated here.

[0061] Preferably, in some possible implementation manners of this embodiment, the method for obtaining the high power load data and the low power load data is as follows: for any target curve, the power load data on the target curve is clustered by the K-means clustering algorithm to obtain clustering clusters; where the value of K in the K-means clustering algorithm is set to 2; the mean value of the power load data in each clustering cluster is obtained as the first mean value of the corresponding clustering cluster; the greater the first mean value, the greater the power load data in the corresponding clustering cluster. Therefore, all the power load data within the clustering cluster corresponding to the largest first mean value are used as the high power load data of the target curve; all the power load data within the clustering cluster corresponding to the smallest first mean value are used as the low power load data of the target curve. Thus, the high power load data and the low power load data in each target curve are determined.

[0062] Preferably, in some possible implementation manners of this embodiment, for the method of obtaining the overall change degree, please refer to Figure 2 , which shows a flowchart of a method for obtaining the overall change degree provided by an embodiment of the present invention. The method includes:

[0063] Step S201: For any target curve, obtain the first change degree of the target curve according to the mean difference and quantity difference between the high power load data and the low power load data on the target curve.

[0064] Among them, the greater the first change degree, the more likely the power user corresponding to the target curve is a light industry power user.

[0065] In some possible implementation manners of this embodiment, the method for obtaining the first change degree is as follows: For any target curve, take the mean difference between the high power load data and the low power load data on the target curve as the first difference; the greater the first difference, the more likely the power user corresponding to the target curve is a light industry power user. Take the ratio of the quantity corresponding to the high power load data and the low power load data on the target curve as the first ratio; take the absolute value of the difference between the first ratio and the first preset constant as the second difference; it is known that the power consumption working hours of light industry power users generally last for 9 hours, that is, from 9:00 to 18:00. Among them, the high power load data of light industry power users must appear within the continuous power consumption period. Therefore, the first ratio of the target curve corresponding to light industry power users should tend to be , and thus in this embodiment, the first preset constant is set to , when the second difference is smaller, it indicates that the power user corresponding to the target curve is more likely to be a light industry power user. Therefore, according to the first difference and the second difference, obtain the first change degree of the target curve; among them, the first difference and the first change degree are in a positive correlation relationship, and the second difference and the first change degree are in a negative correlation relationship. The greater the first change degree, the more likely the power user corresponding to the target curve is a light industry power user. In order to more accurately analyze the relationship between the high power load data and the low power load data on each target curve, this embodiment specifically quantifies the first change degree of the target curve to obtain the first change degree value of each target curve. The greater the first change degree value, the more likely the power user corresponding to the target curve is a light industry power user. Among them, the expression of the first change degree value is:

[0066] ;

[0067] In the formula, is the first change degree value of the k-th target curve; is the quantity of the high power load data on the k-th target curve; is the h-th high power load data on the k-th target curve; is the number of low power load data on the k-th target curve; is the l-th low power load data on the k-th target curve; is the first difference; is the first ratio; is the second difference; exp is the exponential function with the natural constant as the base; is the absolute value function.

[0068] Step S202: Obtain the second degree of change of the target curve according to the fluctuation degree of the power load data of the target curve within the first preset specified time period and the fluctuation condition of the power load data within the second preset specified time period; wherein, the fluctuation degree and the second degree of change are in a positive correlation relationship, and the fluctuation condition and the second degree of change are in a negative correlation relationship.

[0069] It is known that the first preset specified time period is the peak period of light industry power users, and the second preset specified time period is the flat period of light industry power users. When the fluctuation degree of the power load data of the target curve within the first preset specified time period is greater and the fluctuation condition of the power load data within the second preset specified time period is more stable, the power user corresponding to the target curve is more likely to be a light industry power user. Therefore, obtain the second degree of change of the target curve according to the fluctuation degree of the power load data of the target curve within the first preset specified time period and the fluctuation condition of the power load data within the second preset specified time period; wherein, the fluctuation degree and the second degree of change are in a positive correlation relationship, and the fluctuation condition and the second degree of change are in a negative correlation relationship. The greater the second degree of change, the more likely the power user corresponding to the target curve is a light industry power user. In order to more accurately represent the change situation of the target curve within the specified time period, in this embodiment, the second degree of change is specifically quantified to obtain the second degree of change value of each target curve. Among them, the greater the second degree of change value, the more likely the power user corresponding to the target curve is a light industry power user. Among them, the expression of the second degree of change value is:

[0070] ;

[0071] In the formula, is the second degree of change value of the k-th target curve; is the maximum power load data of the k-th target curve within the first preset specified time period; is the minimum power load data of the k-th target curve within the first preset specified time period; is the fluctuation degree; is the k-th target curve within the second preset specified time period The total quantity of all power load data within; For the i-th power load data of the k-th target curve within the second preset specified time period within; For the k-th target curve within the second preset specified time period the mean value of all power load data within; is the fluctuation condition; is the absolute value function; exp is the exponential function with the natural constant as the base.

[0072] Step S203: Obtain the overall change degree of the target curve according to the first change degree and the second change degree of the target curve; wherein, both the first change degree and the second change degree are positively correlated with the overall change degree.

[0073] It is known that the greater the first change degree and the greater the second change degree both indicate that the power user corresponding to the target curve is more likely to be a light industry power user. Therefore, in this embodiment, the overall change degree of the target curve is obtained according to the first change degree and the second change degree of the target curve; wherein, both the first change degree and the second change degree are positively correlated with the overall change degree. The greater the overall change degree, the more likely the power user corresponding to the target curve is to be a light industry power user. In order to accurately analyze the change degree of each target curve, in this embodiment, the overall change degree of each target curve is specifically quantified to obtain the overall change degree value of each target curve. Among them, the greater the overall change degree value, the more likely the user corresponding to the target curve is to be a light industry power user. Among them, the expression of the overall change degree value is: ; In the formula, is the overall change degree value of the k-th target curve; is the first change degree value of the k-th target curve; is the second change degree value of the k-th target curve. In other embodiments, the overall change degree value can be obtained through the addition result of the first change degree value and the second change degree value. The method for obtaining the overall change degree value is not limited herein. Thus, the overall change degree value of each target curve is obtained.

[0074] In order to cluster the target curves more accurately, in this embodiment, according to the overall change degree value of each target curve, the target curves are clustered by the K-means clustering algorithm to obtain two target curve categories, and a preliminary division of light industry power users and residential power users is made.

[0075] Step S3: According to the overall change degree of the target curves within each target curve category, filter out the reference category; wherein, the reference category is the target curve category with the largest average value of the overall change degree; according to the distribution of the peak values on each target curve within the reference category, obtain the peak change degree of each target curve within the reference category; based on the peak change degree, determine the types of each power user within the reference category.

[0076] Specifically, the two obtained target curve categories respectively correspond to light industry power users and residential power users. It is known that the overall change degree of light industry power users is relatively large. Therefore, in this embodiment, the average value of the overall change degree values of all target curves within each target curve category is obtained as the label of each target curve category; the larger the label, the more likely the power users corresponding to the target curves within the corresponding target curve category are light industry power users. Furthermore, the target curve category corresponding to the largest label is used as the reference category, and the power users corresponding to the target curves within the reference category are initially determined as light industry power users; the power users corresponding to the target curves within the other target curve category that is not the reference category are determined as residential power users. In actual situations, if there are some residential power users who are highly consuming electricity during the day, then the overall change degree values of the target curves of these residential power users are similar to those of light industry power users. During the clustering process using the K-means clustering algorithm, it is easy to classify the target curves of residential power users into the reference category, and thus misidentify residential power users as light industry power users. Therefore, in this embodiment, each target curve in the reference category is analyzed to accurately determine whether there are target curves corresponding to residential power users in the reference category.

[0077] It is known that light industry power users conduct office work during the day, and the power load data during the day is relatively large, and there are only fluctuations in the power load data increasing with time within the peak period. Therefore, the change of the target curves of light industry power users is relatively single. However, due to other electricity consumption behaviors such as cooking or entertainment of residential power users, the target curves of residential power users will have peak values in the early morning, noon, evening, and other time periods. At the same time, according to the actual real - world situation analysis, the electricity consumption of residential power users generally increases from morning to evening. Therefore, the peak values on the target curves of residential power users show an increasing state. Therefore, in this embodiment, according to the distribution of the peak values on each target curve within the reference category, obtain the peak change degree of each target curve within the reference category; wherein, the greater the peak change degree, the more likely the power user corresponding to the target curve is a residential power user. Furthermore, based on the peak change degree, accurately distinguish the target curves corresponding to light industry power users and residential power users within the reference category, and thus accurately classify residential power users and light industry power users.

[0078] Preferably, in some possible implementation manners of this embodiment, for the method of obtaining the degree of peak change, please refer to Figure 3 , which shows a flowchart of a method for obtaining the degree of peak change provided by an embodiment of the present invention. The method includes:

[0079] Step S301: For any target curve within the reference category, use the average curve corresponding to the upper and lower envelope curves of the target curve as the overall change curve of the target curve.

[0080] To more clearly analyze the peak distribution of each target curve within the reference category, in this embodiment, the upper and lower envelope curves of each target curve within the reference category are constructed, and the average curve corresponding to the upper and lower envelope curves of each target curve is obtained as the overall change curve of each target curve, which can accurately reflect the change of each target curve within the reference category and more accurately reflect the peak distribution of each target curve within the reference category. Among them, the methods for obtaining the upper and lower envelope curves and the average curve corresponding to the upper and lower envelope curves are well-known technologies and will not be elaborated here.

[0081] Step S302: Obtain the degree of peak change of the target curve according to the difference between each maximum value on the overall change curve and its previous adjacent maximum value, the change rate of each maximum value, and the number of maximum values; among them, the difference between the maximum value and its previous adjacent maximum value, and the number of maximum values are both positively correlated with the degree of peak change, and the change rate is negatively correlated with the degree of peak change.

[0082] For any overall change curve, when the difference between each maximum value on the overall change curve and its previous adjacent maximum value is larger, and the number of maximum values on the overall change curve is larger, it indicates that the power user corresponding to the target curve corresponding to the overall change curve is more likely to be a residential power user. In actual situations, the growth rate of the power load data of light industrial power users is relatively large, and the growth rate of the power load data of residential power users is relatively small. Therefore, on the overall change curve, when the change rate of the maximum value is relatively large, the power user corresponding to the target curve corresponding to the overall change curve is more likely to be a light industrial power user.

[0083] In some possible implementation manners of this embodiment, the method for obtaining the change rate is as follows: for any maximum value in the overall change curve, obtain the average difference between the maximum value and each of its adjacent minimum values, and take them as the reference change rates of the maximum value; wherein, there are two adjacent minimum values on the left and right of the maximum value, so there are two reference change rates for the maximum value. The larger the reference change rate, the greater the difference between the maximum value and its adjacent minimum value, indirectly indicating that the change amplitude of the maximum value is greater. In order to more accurately describe the change amplitude of each maximum value, the largest reference change rate is further taken as the change rate of the maximum value. As an example, taking the a-th maximum value as an example, obtain the absolute value of the difference between the a-th maximum value and each of its adjacent minimum values as the third difference between the a-th maximum value and each of its adjacent minimum values; obtain the absolute value of the difference between the sequence number of the corresponding moment of the a-th maximum value and each of its adjacent minimum values as the time sequence difference between the a-th maximum value and each of its adjacent minimum values. For example, if the a-th maximum value corresponds to the 8th moment within a preset time period, and a certain adjacent minimum value of the a-th maximum value corresponds to the 5th moment within the preset time period, then the time sequence difference between the a-th maximum value and this adjacent minimum value is , where is the absolute value function. The ratio of the third difference between the a-th maximum value and each of its adjacent minimum values to the time sequence difference is the reference change rate of each of the a-th maximum value. If there is no minimum value on the left of the a-th maximum value, then by default, the reference change rate between the a-th maximum value and its left minimum value is 0. Thus, the change rate of each maximum value on each overall change curve is determined. Among them, when the change rates of all the maximum values on a certain overall change curve are all smaller, the power user corresponding to the target curve corresponding to this overall change curve is more likely to be a residential power user.

[0084] Furthermore, in this embodiment, according to the difference between each maximum value and its previous adjacent maximum value, the change rate of each maximum value, and the number of maximum values on each overall change curve, obtain the peak change degree of the target curve corresponding to each overall change curve; wherein, the difference between the maximum value and its previous adjacent maximum value, and the number of maximum values are both positively correlated with the peak change degree, and the change rate is negatively correlated with the peak change degree. In order to more accurately analyze the types of power users corresponding to each target curve within the reference category, in this embodiment, the peak change degree is specifically quantified to obtain the peak change degree value of each target curve within the reference category. Among them, the larger the peak change degree value, the more likely the power user corresponding to the target curve is a residential power user. Among them, the expression of the peak change degree value is:

[0085] ;

[0086] In the formula, is the peak change degree value of the q-th target curve within the reference category; The total number of maximum values on the overall change curve of the q-th target curve within the reference category; The n-th maximum value on the overall change curve of the q-th target curve within the reference category; The (n - 1)-th maximum value on the overall change curve of the q-th target curve within the reference category; The absolute value function; The change rate of the n-th maximum value on the overall change curve of the q-th target curve within the reference category; Is a second preset constant, greater than 0.

[0087] In this embodiment, Is set to 1 to avoid a zero denominator. Implementers can set the size according to the actual situation, which is not limited here.

[0088] It should be noted that when is 1, by default is 0. Thus, the peak change degree values of each target curve within the reference category are obtained.

[0089] It is known that the larger the peak change degree value, the more likely the power user corresponding to the target curve is a residential power user. Therefore, in this embodiment, the peak change degree values of each target curve within the reference category are normalized to obtain the normalized peak change degree values of each target curve within the reference category. In this embodiment, the preset peak change degree value threshold is set to 0.9. Implementers can set the size of the preset peak change degree value threshold according to the actual situation, which is not limited here. When the normalized peak change degree value is greater than the preset peak change degree value threshold, the power user corresponding to the target curve within the reference category is regarded as a residential power user; when the normalized peak change degree value is less than or equal to the preset peak change degree value threshold, the power user corresponding to the target curve within the reference category is regarded as a light industry power user; thus, the residential power users and light industry power users are accurately divided. Accurately determine the type of each power user in the urban area, and then accurately determine the heavy industry power users, light industry power users and residential power users in the urban area.

[0090] Step S4: Determine the current power grid load status according to the power load change situation of each type of power user within a preset historical time period.

[0091] Specifically, it is known that when the power load in the power grid is too large, it may cause a voltage drop, resulting in unstable voltage in the urban area. To ensure the voltage stability in the urban area at the current moment, it is necessary to determine the power load status of the power grid at the current moment, and then determine the adjustment degree of the power load of the power grid at the current moment. In this embodiment, the power load data of heavy industrial power users within a preset historical time period, the number of heavy industrial power users, the power load data of light industrial power users within a preset historical time period, the number of light industrial power users, and the power load data of residential power users within a preset historical time period, and the number of residential power users are sequentially input into the trained LSTM (Long Short-Term Memory) neural network, and the predicted power load data of the power grid at the current moment is output. The predicted power load data of the power grid at the current moment is compared with the power load data of the power grid at the current moment. If the power load data of the power grid at the current moment is larger, it means that the power load in the corresponding power grid in the urban area at the current moment is too large. To ensure the voltage stability in the urban area, the capacity or quantity of the transformer is increased to disperse the power load of the power grid at the current moment until the power load data of the power grid at the current moment is less than or equal to the predicted power load data of the power grid at the current moment, and then the increase of the capacity or quantity of the transformer is stopped, and the voltage in the urban area is ensured to be in a stable state in real time, ensuring that the electrical appliances work stably. Among them, in this embodiment, the preset historical time period is set to one month, and the end time of the preset historical time period is the previous adjacent moment of the current moment. The LSTM (Long Short-Term Memory) neural network is a well-known technology and will not be elaborated here. Thus, the power load status of the power grid at the current moment is accurately identified and accurately adjusted to ensure the voltage stability in the urban area.

[0092] In summary, in this embodiment, the power load time series curve of the power user is obtained, and the target curve is selected according to the change. According to the difference between the high power load data and the low power load data on the target curve, and the change of the target curve within a specified time period, the overall change degree is obtained; the reference category is obtained based on the overall change degree; according to the distribution of the peaks on the target curve within the reference category, the peak change degree is obtained to determine the type of power user; according to the power load change of each type of power user within a preset historical time period, the current power load status of the power grid is determined. The present invention accurately divides the power users by determining the types of power users, and then accurately analyzes the current power load status of the power grid, and accurately adjusts it to ensure the voltage stability.

[0093] Embodiment 2:

[0094] The present invention also proposes a power grid power load status identification system based on a neural network. Please refer to Figure 4, which shows the structural diagram of a power grid power load status identification system based on a neural network provided by an embodiment of the present invention. The system includes: a target curve acquisition module 10, a target curve category acquisition module 20, a power user type determination module 30, and a power grid power load status determination module 40.

[0095] The target curve acquisition module 10 is used to acquire the power load time series curve of each power user within a preset time period; and filter out the target curve according to the change situation of the power load time series curve.

[0096] The target curve category acquisition module 20 is used to determine the high power load data and low power load data on each target curve according to the size of the power load data on each target curve; obtain the overall change degree of each target curve according to the mean difference and quantity difference between the high power load data and the low power load data on each target curve, as well as the change situation of each target curve within a specified time period; and cluster the target curves based on the overall change degree to obtain the target curve categories.

[0097] The power user type determination module 30 is used to filter out the reference category according to the overall change degree of the target curves within each target curve category; wherein, the reference category is the target curve category with the largest mean value of the overall change degree; obtain the peak change degree of each target curve within the reference category according to the distribution of the peaks on each target curve within the reference category; and determine the type of each power user within the reference category based on the peak change degree.

[0098] The power grid power load status determination module 40 is used to determine the current power grid power load status according to the power load change situation of each type of power user within a preset historical time period.

[0099] It should be noted that: for the system provided in the above embodiment, only the division of the above functional modules is used for illustration. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, an embodiment of a power grid power load status identification system based on a neural network and an embodiment of a power grid power load status identification method based on a neural network provided in the above embodiment belong to the same concept, and the specific implementation process can be seen in the method embodiment, which will not be elaborated here.

[0100] Embodiment 3:

[0101] The present invention also proposes an electronic device, please refer to Figure 5, including a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and executable on the processor 402. When the processor 402 executes the computer program 403, the steps of the method according to any of the foregoing embodiments are implemented. In the embodiments of the present application, the processor is the control center of the computer system, which may be the processor of a physical machine or the processor of a virtual machine.

[0102] Embodiment 4:

[0103] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps of the method according to any of the foregoing embodiments are implemented. Among them, the computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives, and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic cards or optical cards, nanosystems (including molecular memory ICs), or any type of medium or device suitable for storing instructions and / or data.

[0104] It should be noted that the above sequence of the embodiments of the present invention is only for description and does not represent the advantages or disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0105] Each embodiment in this specification is described in a progressive manner. The same or similar parts among the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for identifying the power load status of a power grid based on a neural network, characterized in that, The method includes the following steps: Obtain the time-series curve of the power load of each power user within a preset time period; according to the change situation of the time-series curve of the power load, screen out the target curves; Determine the high-power load data and low-power load data on each target curve according to the magnitude of the power load data on each target curve; according to the mean difference and quantity difference between the high-power load data and the low-power load data on each target curve, and the change situation of each target curve within a specified time period, obtain the overall change degree of each target curve; cluster the target curves based on the overall change degree to obtain the target curve categories; Screen out the reference category according to the overall change degree of the target curves within each target curve category; wherein, the reference category is the target curve category with the largest mean value of the overall change degree; according to the distribution of the peaks on each target curve within the reference category, obtain the peak change degree of each target curve within the reference category; determine the types of each power user within the reference category based on the peak change degree; Determine the current power grid power load state according to the power load change situation of each type of power user within a preset historical time period; The method for obtaining the peak change degree is: For any target curve within the reference category, use the average curve corresponding to the upper and lower envelope curves of the target curve as the overall change curve of the target curve; According to the difference between each maximum value on the overall change curve and its previous adjacent maximum value, the change rate of each maximum value, and the number of maximum values, obtain the peak change degree of the target curve; wherein, both the difference between the maximum value and its previous adjacent maximum value and the number of maximum values are positively correlated with the peak change degree, and the change rate is negatively correlated with the peak change degree; Among them, the expression of the peak change degree value is: ; in the formula, is the peak change degree value of the q-th target curve in the reference category; is the total number of maximum values on the overall change curve of the q-th target curve in the reference category; is the n-th maximum value on the overall change curve of the q-th target curve in the reference category; is the (n - 1)-th maximum value on the overall change curve of the q-th target curve in the reference category; is the absolute value function; is the change rate of the n-th maximum value on the overall change curve of the q-th target curve in the reference category; is the second preset constant, greater than 0.

2. The method for identifying the power grid load status based on neural network according to claim 1, wherein, The method for obtaining the target curves is: Obtain the variance of the power load data on each time-series curve of the power load as the fluctuation value of each time-series curve of the power load; When the fluctuation value is greater than the preset fluctuation value threshold, use the corresponding time-series curve of the power load as the target curve.

3. The method for identifying the power grid load status based on a neural network according to claim 1, wherein, The method for obtaining the high-power load data and the low-power load data is: For any target curve, cluster the power load data on the target curve through the K-means clustering algorithm to obtain clustering clusters; wherein, the value of K in the K-means clustering algorithm is set to 2; Obtain the mean value of the power load data in each clustering cluster as the first mean value of the corresponding clustering cluster; All the power load data within the clustering cluster corresponding to the largest first mean value are used as the high-power load data of the target curve; All the power load data within the clustering cluster corresponding to the smallest first mean value are used as the low-power load data of the target curve.

4. The method for identifying the power load status of a power grid based on a neural network according to claim 1, characterized in that, The method for obtaining the overall change degree is: For any target curve, obtain the first change degree of the target curve according to the mean difference and quantity difference between the high-power load data and the low-power load data on the target curve; Obtain the second degree of change of the target curve according to the degree of fluctuation of the power load data within the first preset specified time period and the fluctuation condition of the power load data within the second preset specified time period; wherein, the degree of fluctuation is positively correlated with the second degree of change, and the fluctuation condition is negatively correlated with the second degree of change; Obtain the overall degree of change of the target curve according to the first degree of change and the second degree of change of the target curve; wherein, both the first degree of change and the second degree of change are positively correlated with the overall degree of change.

5. The method for identifying the power grid load status based on a neural network according to claim 4, characterized in that The method for obtaining the first degree of change is as follows: For any target curve, take the mean difference between the high power load data and the low power load data on the target curve as the first difference; Take the ratio of the quantity corresponding to the high power load data and the low power load data on the target curve as the first ratio; Take the difference between the first ratio and the first preset constant as the second difference; Obtain the first degree of change of the target curve according to the first difference and the second difference; wherein, the first difference is positively correlated with the first degree of change, and the second difference is negatively correlated with the first degree of change.

6. The method for identifying the power load status of a power grid based on a neural network according to claim 1, wherein The method for obtaining the change rate is as follows: For any maximum value in the overall change curve, obtain the average difference between the maximum value and each adjacent minimum value, and take them all as the reference change rate of the maximum value; Take the largest reference change rate as the change rate of the maximum value.

7. The method for identifying the power grid load status based on neural network according to claim 1, characterized in that, The method for determining the type of each power user within the reference category based on the peak change degree is as follows: When the peak change degree does not meet the preset peak change degree range, take the power user corresponding to the target curve within the reference category as a light industry power user; When the peak change degree meets the preset peak change degree range, take the power user corresponding to the target curve within the reference category as a residential power user.

8. The method for identifying the power grid load status based on a neural network according to claim 1, characterized in that, The types of power users include heavy industry power users, light industry power users and residential power users.

Images