Load data processing method and device, computer device and storage medium

Abstract

The application relates to a load data processing method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring an original load sequence corresponding to a target period; performing preliminary repair on first-type abnormal data in the original load sequence to obtain a preliminary load sequence; performing frequency domain decomposition on the preliminary load sequence, and constructing a time domain characteristic curve according to a decomposition result; determining the first-type abnormal data in the original load sequence according to a load curve corresponding to the time domain characteristic curve and the preliminary load sequence; and repairing the first-type abnormal data and second-type abnormal data in the original load sequence according to the time domain characteristic curve. The method can improve the user load abnormal data identification capability and the repair effect.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric power information, in particular to a load data processing method and device, computer equipment and storage medium. BACKGROUND

[0002] With the continuous deepening of the intelligentization of power grids and the increasing complexity of power grid structure and operation mode, the data generated by the power system gradually presents the trend of massification and high dimensionality. Among them, the abnormality, redundancy and omission of power user side data have a significant impact on the safe operation and dispatching of power grids.

[0003] An important feature of user load is that the data quality is poor and contains a large number of abnormal data. Therefore, for user load prediction, correctly identifying abnormal data is the premise and key to accurate prediction. Existing abnormal data identification and repair methods include linear interpolation method, quadratic interpolation method, etc. However, the existing methods can only identify and repair part of the obvious abnormal data (such as scattered abnormal data), and cannot achieve effective identification and repair effect for the second type of abnormal data points (such as non-obvious step points) in user load data. SUMMARY

[0004] Therefore, it is necessary to provide a load data processing method, device, computer equipment and storage medium capable of improving the abnormal data identification ability and repair effect of user load.

[0005] In a first aspect, the present application provides a load data processing method. The method comprises:

[0006] obtaining an original load sequence corresponding to a target period;

[0007] performing preliminary repair on first type abnormal data in the original load sequence to obtain a preliminary load sequence;

[0008] performing frequency domain decomposition on the preliminary load sequence, and constructing a time domain feature curve according to the decomposition result;

[0009] determining second type abnormal data in the original load sequence according to the time domain feature curve and a load curve corresponding to the preliminary load sequence; wherein the identification degree of the second type abnormal data is lower than that of the first type abnormal data;

[0010] performing repair on the first type abnormal data and the second type abnormal data in the original load sequence according to the time domain feature curve.

[0011] In one of the embodiments, the frequency domain decomposition on the preliminary load sequence and the construction of the time domain feature curve according to the decomposition result comprise:

[0012] extracting a frequency domain daily component and a frequency domain weekly component from the decomposition result;

[0013] converting the frequency domain daily component into a time domain daily component, and converting the frequency domain weekly component into a time domain weekly component;

[0014] adding the time domain daily component and the time domain weekly component to obtain a time domain feature curve.

[0015] In one of the embodiments, determining second type of abnormal data in the original load sequence according to the time domain feature curve and a load curve corresponding to the preliminary load sequence, comprises:

[0016] constructing a difference curve according to the time domain feature curve and the load curve corresponding to the preliminary load sequence;

[0017] determining the second type of abnormal data in the original load sequence according to the difference curve.

[0018] In one of the embodiments, determining the second type of abnormal data in the original load sequence according to the difference curve, comprises:

[0019] constructing a difference increment curve according to an increment of the difference curve at any two adjacent time instants within the target time period; wherein the difference increment curve is subject to a normal distribution with a mean value of 0;

[0020] determining the second type of abnormal data in the original load sequence according to a standard deviation of the difference increment curve.

[0021] In one of the embodiments, determining the second type of abnormal data in the original load sequence according to the standard deviation of the difference increment curve, comprises:

[0022] determining the standard deviation of the difference increment curve;

[0023] multiplying the standard deviation by a set multiple to obtain a screening value; wherein the set multiple is 3;

[0024] determining an abnormal point in the difference increment curve according to the screening value;

[0025] determining the second type of abnormal data in the original load sequence according to the abnormal point.

[0026] In one of the embodiments, repairing first type of abnormal data and second type of abnormal data in the original load sequence according to the time domain feature curve, comprises:

[0027] extracting a feature curve segment corresponding to the first type of abnormal data and a feature curve segment corresponding to the second type of abnormal data from the time domain feature curve;

[0028] linearly transforming the feature curve segment corresponding to the first type of abnormal data to obtain a repair value corresponding to the first type of abnormal data, and linearly transforming the feature curve segment corresponding to the second type of abnormal data to obtain a repair value corresponding to the second type of abnormal data.

[0029] In one of the embodiments, the preliminary repair of the first type of abnormal data in the original load sequence comprises:

[0030] The first type of abnormal data in the original load sequence is preliminarily repaired by linear interpolation.

[0031] In a second aspect, the present application further provides a load data processing device. The device comprises:

[0032] a sequence acquisition module configured to acquire an original load sequence corresponding to a target period;

[0033] a first repair module configured to preliminarily repair the first type of abnormal data in the original load sequence to obtain a preliminary load sequence;

[0034] a curve construction module configured to perform frequency domain decomposition on the preliminary load sequence, and construct a time domain feature curve according to the decomposition result;

[0035] a data determination module configured to determine the second type of abnormal data in the original load sequence according to the time domain feature curve and a load curve corresponding to the preliminary load sequence; wherein the recognition degree of the second type of abnormal data is lower than that of the first type of abnormal data;

[0036] a second repair module configured to repair the first type of abnormal data and the second type of abnormal data in the original load sequence according to the time domain feature curve.

[0037] In a third aspect, the present application further provides a computer device. The computer device comprises a memory and a processor, the memory stores a computer program, and the processor realizes the following steps when executing the computer program:

[0038] acquiring an original load sequence corresponding to a target period;

[0039] preliminarily repairing the first type of abnormal data in the original load sequence to obtain a preliminary load sequence;

[0040] performing frequency domain decomposition on the preliminary load sequence, and constructing a time domain feature curve according to the decomposition result;

[0041] determine second type of abnormal data in the original load sequence according to the time domain feature curve and a load curve corresponding to the preliminary load sequence; wherein the recognition degree of the second type of abnormal data is lower than that of the first type of abnormal data;

[0042] repair the first type of abnormal data and the second type of abnormal data in the original load sequence according to the time domain feature curve.

[0043] In a fourth aspect, the present application further provides a computer readable storage medium. The computer readable storage medium has a computer program stored thereon, and the computer program is executed by a processor to implement the following steps:

[0044] obtain an original load sequence corresponding to a target period;

[0045] preliminarily repair first type of abnormal data in the original load sequence to obtain a preliminary load sequence;

[0046] perform frequency domain decomposition on the preliminary load sequence, and construct a time domain feature curve according to a decomposition result;

[0047] determine second type of abnormal data in the original load sequence according to the time domain feature curve and a load curve corresponding to the preliminary load sequence; wherein the recognition degree of the second type of abnormal data is lower than that of the first type of abnormal data;

[0048] repair the first type of abnormal data and the second type of abnormal data in the original load sequence according to the time domain feature curve.

[0049] In a fifth aspect, the present application further provides a computer program product. The computer program product comprises a computer program, and the computer program is executed by a processor to implement the following steps:

[0050] obtain an original load sequence corresponding to a target period;

[0051] preliminarily repair first type of abnormal data in the original load sequence to obtain a preliminary load sequence;

[0052] perform frequency domain decomposition on the preliminary load sequence, and construct a time domain feature curve according to a decomposition result;

[0053] determine second type of abnormal data in the original load sequence according to the time domain feature curve and a load curve corresponding to the preliminary load sequence; wherein the recognition degree of the second type of abnormal data is lower than that of the first type of abnormal data;

[0054] repair the first type of abnormal data and the second type of abnormal data in the original load sequence according to the time domain feature curve.

[0055] The load data processing method, device, computer device and storage medium described above, through the preliminary repair of the first type of abnormal data in the original load sequence, i.e. the simple repair of the original load sequence, lays a foundation for the subsequent extraction of the time domain feature curve of the original load curve; further, the preliminary load sequence is decomposed in the frequency domain, and the time domain feature curve is constructed according to the decomposition result, and then the second type of abnormal data in the original load sequence can be determined according to the time domain feature curve and the load curve corresponding to the preliminary load sequence, that is, through the analysis of the user load characteristics, the accurate identification of the second type of abnormal data can be realized; thus, according to the time domain feature curve, the first type of abnormal data and the second type of abnormal data in the original load sequence can be effectively repaired. That is, the first type of abnormal data and the second type of abnormal data in the load data can be accurately identified and repaired by using the present scheme. BRIEF DESCRIPTION OF DRAWINGS

[0056] Figure 1 A flowchart of a load data processing method in an embodiment is shown;

[0057] Figure 2 A flowchart of constructing a time domain feature curve in an embodiment is shown;

[0058] Figure 3 A flowchart of determining a second type of abnormal data in an embodiment is shown;

[0059] Figure 4 A flowchart of repairing abnormal data in an embodiment is shown;

[0060] Figure 5 A flowchart of a load data processing method in another embodiment is shown;

[0061] Figure 6 A structural block diagram of a load data processing device in an embodiment is shown;

[0062] Figure 7 A structural block diagram of a load data processing device in another embodiment is shown;

[0063] Figure 8 A structural block diagram of a load data processing device in another embodiment is shown;

[0064] Figure 9 An internal structure diagram of a computer device in an embodiment is shown. DETAILED DESCRIPTION

[0065] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0066] The load data processing method provided by the embodiments of the present application can be applied to the case of processing abnormal data in user load data. Optionally, the method can be implemented by an independent server or a terminal with powerful computing capability. In an embodiment, as shown in Figure 1 FIG. 1, a load data processing method is provided, which is taken as an example of being applied to a server and includes the following steps:

[0067] S101, obtaining an original load sequence corresponding to a target period.

[0068] The target period is a period for which load abnormal data is repaired, and can be consecutive weeks, such as three consecutive weeks. The original load sequence refers to a user load sequence in the target period that has not been processed.

[0069] Optionally, the load data of a user in consecutive three weeks can be exported through a user load side management platform, or the load data of the user in consecutive three weeks can be exported based on historical load data of a user smart meter, that is, as original load data. The original load data is arranged in chronological order, and thus the original load sequence is obtained.

[0070] S102, performing preliminary repair on first type abnormal data in the original load sequence to obtain a preliminary load sequence.

[0071] The first type abnormal data, that is, the significant abnormal data in the original load sequence, can include but is not limited to zero value, null value, near-zero value, continuous constant value, and significant abnormal step value, etc.

[0072] Optionally, before repairing the first type abnormal data, different types of abnormal data need to be distinguished by a corresponding method. The specific distinguishing method is as follows:

[0073] Firstly, for the null value of the first type abnormal data, P t null is taken as a criterion. P t represents a curve corresponding to the original load sequence, where t is a time point, which can be in hours or in days. If the original load at time t is null, the data at the time point is regarded as the null value of the first type abnormal data.

[0074] Secondly, for the zero value and near-zero value of the first type abnormal data, |P t |≤ε1 is taken as a criterion. ε1 is a small number, and its size depends on the sensitivity of the sensor.

[0075] Thirdly, for the continuous constant value of the first type abnormal data, the difference between adjacent points can be used to distinguish the criterion, as shown in the following formulas (1)-(2):

[0076]

[0077] ΔP t = P t - P t-1 (2)

[0078] wherein, ΔP t is the original load difference curve of two adjacent time, P t-1 is the curve corresponding to the original load sequence at t-1 time, r is the detection window width, generally taking the value of 1, i is the current detection time, and ε2 is a small number, the size of which depends on the sensitivity of the sensor.

[0079] The fourth: for the first type of abnormal data of significant abnormal step value, the Chebyshev inequality method independent of any distribution assumption is used for identification, specifically as the following formula (3):

[0080] Prod(| ΔP t - E| < k * D) > 1-1 / k 2 (3)

[0081] wherein, Prod() is a probability function, E is the expectation of ΔP t , D is the standard deviation, and k represents the degree of sample deviation from the expectation. The meaning of the formula is that the probability of the sample deviating from the expectation k times the standard deviation is less than 1 / k 2 . Optionally, k = 5 can be taken, which means that the probability of ΔP t falling within the interval [E-5D, E+5D] is about 96%; if the original load difference falls outside this interval, then this original load difference is a significant abnormal increment, i.e. a significant abnormal step value.

[0082] It should be noted that after identifying the above four different types of first type of abnormal data, linear interpolation can be used to preliminarily repair the first type of abnormal data in the original load sequence. After repairing the first type of abnormal data in the original load sequence, the preliminary load sequence can be obtained. It can be understood that through the linear interpolation method, the identified first type of abnormal data can be repaired simply and conveniently.

[0083] S103, frequency domain decomposition is performed on the preliminary load sequence, and a time domain feature curve is constructed according to the decomposition result.

[0084] wherein, frequency domain decomposition is a data processing method based on the principle of Fourier transform, which decomposes time domain signals into signals of different frequencies, thereby obtaining frequency domain signals. The feature curve represents the overall characteristics of the curve corresponding to the load sequence, and the time domain feature curve is the feature curve of the curve corresponding to the preliminary load sequence in the time domain.

[0085] Specifically, after the preliminary load sequence is decomposed in the frequency domain, preliminary load data components of the daily cycle part (i.e., frequency domain daily components), preliminary load data components of the weekly cycle part (i.e., frequency domain weekly components), preliminary load data components of the low frequency part (i.e., frequency domain low frequency components), and preliminary load data components of the high frequency part (i.e., frequency domain high frequency components) are obtained.

[0086] Optionally, the decomposed components can be recombined, for example, the sum of the frequency domain daily components and the frequency domain weekly components can be taken as a frequency domain characteristic curve, and the frequency domain characteristic curve can be transformed into the time domain to obtain a time domain characteristic curve.

[0087] In S104, second type abnormal data in the original load sequence is determined according to the time domain characteristic curve and a load curve corresponding to the preliminary load sequence.

[0088] The load curve is drawn according to the preliminary load sequence, in which the horizontal coordinate is time and the vertical coordinate is user load. The second type abnormal data is non-significant abnormal data in the original load sequence, and can include non-significant abnormal step values. Further, the recognition degree of the second type abnormal data is lower than that of the first type abnormal data, that is, the difficulty of identifying the second type abnormal data from the original load sequence is higher than that of identifying the first type abnormal data from the original load sequence.

[0089] Optionally, after the time domain characteristic curve and the load curve corresponding to the preliminary load sequence are obtained, the time domain characteristic curve and the load curve corresponding to the preliminary load sequence are analyzed to construct a statistical condition that can reflect the change rule of the load curve, that is, a normal distribution curve with an expectation of 0 can be constructed. According to the characteristics of the normal distribution curve with an expectation of 0, the second type abnormal data can be finely identified.

[0090] In S105, the first type abnormal data and the second type abnormal data in the original load sequence are repaired according to the time domain characteristic curve.

[0091] Optionally, after the first type abnormal data and the second type abnormal data are identified, for each first type abnormal data and second type abnormal data, a time domain characteristic curve segment corresponding thereto can be found, and further, the time domain characteristic curve segment can be processed so that the breakpoints at both ends of the segment match the adjacent normal data. The time domain characteristic curve after processing is the final repair value of the first type abnormal data and the second type abnormal data.

[0092] In the load data processing method, the first type of abnormal data in the original load sequence is preliminarily repaired, that is, the original load sequence is simply repaired, thereby laying a foundation for subsequent extraction of the time domain feature curve of the original load curve. Further, the preliminary load sequence is decomposed in the frequency domain, and the time domain feature curve is constructed according to the decomposition result. Then, the second type of abnormal data in the original load sequence can be determined according to the time domain feature curve and the load curve corresponding to the preliminary load sequence, that is, the second type of abnormal data can be accurately identified by analyzing the load characteristics of the user. Therefore, the first type of abnormal data and the second type of abnormal data in the original load sequence can be effectively repaired according to the time domain feature curve. That is, the first type of abnormal data and the second type of abnormal data in the load data can be accurately identified and repaired by using the scheme.

[0093] In order to better explain the construction process of the time domain feature curve, on the basis of the above-mentioned embodiments, in one embodiment, as shown in Figure 2 , the steps include the following steps:

[0094] S201, extracting the frequency domain daily component and the frequency domain weekly component from the decomposition result.

[0095] Optionally, after the preliminary load sequence is decomposed in the frequency domain, the frequency domain daily component, the frequency domain weekly component, the frequency domain low-frequency component and the frequency domain high-frequency component of the preliminary load sequence are obtained, from which the frequency domain daily component and the frequency domain weekly component can be extracted.

[0096] S202, converting the frequency domain daily component into a time domain daily component, and converting the frequency domain weekly component into a time domain weekly component.

[0097] Optionally, the frequency domain daily component can be converted into the time domain daily component, and the frequency domain weekly component can be converted into the time domain weekly component by inverse Fourier transform.

[0098] S203, adding the time domain daily component and the time domain weekly component to obtain the time domain feature curve.

[0099] It should be noted that for most user loads, the load is composed of basic load units (including industry, commerce, residence, government, etc.), and the load of each load unit is small enough compared to the total load. According to the central limit theorem, a random event influenced by many random events should follow a normal distribution. Therefore, the load curve corresponding to the preliminary load sequence should follow a normal distribution, that is, P t ’~N(u t ,σ t ), where P t ’ represents the load curve corresponding to the preliminary load sequence, u t is the expectation of P t ’, and σ tIt is P t The variance of '; the load curve P corresponding to the initial load sequence at time t before time t. t-1 '~N(u t-1 , σ t-1 ), where u t-1 It is P t-1 'Expectation, σ t-1 It is P t-1 The variance of '. Assuming the initial load sequences at two times are independent, subtracting them yields formula (4):

[0100] ΔP t '~N(Δu t , Δσ t (4)

[0101] Where, ΔP t ' is the initial load difference curve between two adjacent time points, Δu t It is ΔP t The expected value of ', Δσ t It is ΔP t The variance of '.

[0102] Because among the four components of the load curve corresponding to the preliminary load sequence, the low-frequency component changes very little between the two time periods, while the high-frequency component generally includes measurement noise and the random behavior of users' electricity consumption. The noise is Gaussian white noise, and the random behavior follows a Gaussian distribution. Therefore, the high-frequency component is close to a Gaussian distribution with an expected value of 0. Thus, only the daily and weekly components in the time domain differ by only Δu from the expected value of the preliminary user load at adjacent times. t Therefore, in this embodiment, the sum of the daily and weekly time-domain components is used as the time-domain characteristic curve. Let the time-domain characteristic curve be represented by F(t). Thus, the difference Δu between the initial user load expectations at adjacent times... t This can be expressed as formula (5):

[0103] Δu t =F(t)-F(t-1) (5)

[0104] Wherein, F(t-1) is the time-domain characteristic curve at time t-1.

[0105] In this embodiment, by extracting the daily and weekly components from the frequency domain decomposition results, a time-domain feature curve is constructed, which lays the foundation for the subsequent identification of the second type of abnormal data and improves the accuracy of the identification of the second type of abnormal data.

[0106] To better explain the process for determining the second type of abnormal data in S104, based on the above embodiments, in one embodiment, such as... Figure 3 As shown, the specific steps include:

[0107] S301, constructing a difference curve according to the time-domain characteristic curve and the load curve corresponding to the preliminary load sequence.

[0108] The difference curve is a curve obtained by subtracting the load curve corresponding to the preliminary load sequence from the time-domain characteristic curve.

[0109] Optionally, let I(t) represent the difference curve, I(t) can be represented by formula (6):

[0110] I(t)=P t ’-F(t) (6)

[0111] S302, determining the second type of abnormal data in the original load sequence according to the difference curve.

[0112] Optionally, constructing a difference increment curve according to the increment of the difference curve at any two adjacent time points in the target period; wherein the difference increment curve is subject to a normal distribution with a mean of 0; determining the second type of abnormal data in the original load sequence according to the standard deviation of the difference increment curve.

[0113] Wherein, let D(t) represent the difference increment curve, D(t) can be represented by formula (7):

[0114] D(t)=ΔP t ’-ΔF(t) (7)

[0115] Wherein, ΔF(t) is the difference curve of the time-domain characteristic curve of the adjacent two time points.

[0116] Since the expectation of ΔP t ’ is Δu t , and the expectation of ΔF(t) is also Δu t , the difference increment curve D(t) is subject to a normal distribution with an expectation of 0. For a normal distribution with an expectation of 0, data points falling within three standard deviations are considered normal data, and data points falling outside three standard deviations are considered abnormal data. Therefore, for the difference increment curve D(t) subject to a normal distribution with an expectation of 0, the identification of the second type of abnormal data can be: determining the standard deviation of the difference increment curve; taking the product of the standard deviation and a set multiple as a screening value; wherein the set multiple is 3; determining the abnormal points in the difference increment curve according to the screening value; determining the second type of abnormal data in the original load sequence according to the abnormal points.

[0117] Optionally, the standard deviation of the difference increment curve is calculated and represented by σ, and 3σ is taken as the screening value of the second type of abnormal data. Therefore, the abnormal points exceeding the range of 3σ, wherein the abnormal points are corresponding time points, and the data corresponding to the time points are the second type of abnormal data.

[0118] It can be understood that by constructing a statistical condition that can reflect the change rule of the preliminary load curve, that is, constructing a difference increment curve subject to a normal distribution with an expectation of 0, and according to the characteristics of the normal distribution with an expectation of 0, the second type of abnormal data can be accurately identified.

[0119] In this embodiment, the second type of abnormal data in the original load sequence can be accurately identified according to the characteristics of the difference curve.

[0120] In order to better explain the repair process of the first type of abnormal data and the second type of abnormal data, on the basis of the above embodiment, in an embodiment, as shown in Figure 4 the specific steps include the following steps:

[0121] S401, extracting the feature curve segment corresponding to the first type of abnormal data and the feature curve segment corresponding to the second type of abnormal data from the time domain feature curve.

[0122] Optionally, after identifying the first type of abnormal data and the second type of abnormal data, each first type of abnormal data and second type of abnormal data corresponds to a time, and the time domain feature curve within a certain time range before and after each corresponding time can be set as the time domain feature curve segment corresponding to the time.

[0123] S402, performing linear transformation on the feature curve segment corresponding to the first type of abnormal data to obtain the repair value corresponding to the first type of abnormal data, and performing linear transformation on the feature curve segment corresponding to the second type of abnormal data to obtain the repair value corresponding to the second type of abnormal data.

[0124] It should be noted that since the preliminary repair stage only performs simple repair on the first type of abnormal data without using the information of the time domain feature curve, for each first type of abnormal data, after finding the time domain feature curve corresponding thereto, linear transformation is performed on the feature curve segment corresponding to the first type of abnormal data, that is, the time domain feature curve corresponding to the first type of abnormal data is linearly transformed to the adjacent normal time domain feature curves at both ends, and the transformed time domain feature curve is the repair value corresponding to the first type of abnormal data.

[0125] At the same time, for each second type of abnormal data, after finding the time domain feature curve corresponding thereto, linear transformation is performed on the feature curve segment corresponding to the second type of abnormal data, that is, the time domain feature curve corresponding to the second type of abnormal data is linearly transformed to the adjacent normal time domain feature curves at both ends, and the transformed time domain feature curve is the repair value corresponding to the second type of abnormal data.

[0126] In this embodiment, the time domain feature curve segments corresponding to the first type of abnormal data and the second type of abnormal data are repaired by linear transformation, thereby effectively repairing the first type of abnormal data and the second type of abnormal data.

[0127] Figure 5 For another embodiment of the flowchart of the load data processing method, on the basis of the above embodiment, the embodiment provides an optional example of the load data processing method. In combination with Figure 5 , the specific implementation process is as follows:

[0128] S501, obtaining the original load sequence corresponding to the target period.

[0129] S502, performing preliminary repair on the first type of abnormal data in the original load sequence to obtain a preliminary load sequence.

[0130] Optionally, the linear interpolation is used to perform preliminary repair on the first type of abnormal data in the original load sequence.

[0131] S503, performing frequency domain decomposition on the preliminary load sequence, and extracting a frequency domain daily component and a frequency domain weekly component from the decomposition result.

[0132] S504, converting the frequency domain daily component into a time domain daily component, and converting the frequency domain weekly component into a time domain weekly component.

[0133] S505, adding the time domain daily component and the time domain weekly component to obtain a time domain characteristic curve.

[0134] S506, constructing a difference curve according to the time domain characteristic curve and the load curve corresponding to the preliminary load sequence.

[0135] S507, constructing a difference increment curve according to the increment of the difference curve at any two adjacent time points in the target period.

[0136] Wherein, the difference increment curve is subject to a normal distribution with a mean of 0.

[0137] S508, determining the standard deviation of the difference increment curve.

[0138] S509, taking the product of the standard deviation and a set multiple as a screening value.

[0139] Wherein, the set multiple is 3.

[0140] S510, determining the abnormal points in the difference increment curve according to the screening value.

[0141] S511, determining the second type of abnormal data in the original load curve according to the abnormal points.

[0142] S512, extracting the characteristic curve segment corresponding to the first type of abnormal data and the characteristic curve segment corresponding to the second type of abnormal data from the time domain characteristic curve.

[0143] S513, performing linear transformation on the feature curve segment corresponding to the first type of abnormal data to obtain a repair value corresponding to the first type of abnormal data, and performing linear transformation on the feature curve segment corresponding to the second type of abnormal data to obtain a repair value corresponding to the second type of abnormal data.

[0144] The specific process of S501-S513 can be referred to the description of the method embodiments, which has similar implementation principles and technical effects, and will not be described here.

[0145] It should be understood that, although each step in the flowchart involved in each embodiment as described above is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless explicitly stated herein, the execution of these steps has no strict order limitation, and these steps can be executed in other orders. Moreover, at least part of the steps in the flowchart involved in each embodiment as described above can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily sequential, but can be alternately executed with other steps or steps or stages in other steps.

[0146] Based on the same inventive concept, the embodiments of the present application also provide a load data processing device for implementing the load data processing method described above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more load data processing device embodiments provided below can be referred to the limitations of the load data processing method described above, and will not be described here.

[0147] In one embodiment, as shown in Figure 6 A load data processing device 1 is provided, comprising a sequence acquisition module 100, a first repair module 200, a curve construction module 300, a data determination module 400 and a second repair module 500, wherein:

[0148] The sequence acquisition module 100 is configured to acquire an original load sequence corresponding to a target period.

[0149] The first repair module 200 is configured to preliminarily repair first type of abnormal data in the original load sequence to obtain a preliminary load sequence.

[0150] The curve construction module 300 is configured to perform frequency domain decomposition on the preliminary load sequence, and construct a time domain feature curve according to the decomposition result.

[0151] The data determination module 400 is configured to determine the second type of abnormal data in the original load sequence according to the time-domain feature curve and the load curve corresponding to the preliminary load sequence. The recognition degree of the second type of abnormal data is lower than that of the first type of abnormal data.

[0152] The second repair module 500 is configured to repair the first type of abnormal data and the second type of abnormal data in the original load sequence according to the time-domain feature curve.

[0153] The load data processing device can lay a foundation for subsequent extraction of the time-domain feature curve of the original load curve by preliminarily repairing the first type of abnormal data in the original load sequence, that is, by simply repairing the original load sequence. Further, the time-domain feature curve is constructed according to the decomposition result of the preliminary load sequence, and the second type of abnormal data in the original load sequence can be determined according to the time-domain feature curve and the load curve corresponding to the preliminary load sequence, that is, the second type of abnormal data can be accurately recognized by analyzing the load characteristics of the user. Therefore, the first type of abnormal data and the second type of abnormal data in the original load sequence can be effectively repaired according to the time-domain feature curve. That is, the first type of abnormal data and the second type of abnormal data in the load data can be accurately recognized and repaired by using the scheme.

[0154] In one embodiment, the curve construction module 300 is specifically configured to:

[0155] extract the frequency-domain daily component and the frequency-domain weekly component from the decomposition result, convert the frequency-domain daily component into a time-domain daily component, and convert the frequency-domain weekly component into a time-domain weekly component, and add the time-domain daily component and the time-domain weekly component to obtain the time-domain feature curve.

[0156] In one embodiment, on the basis of Figure 6 as shown in Figure 7 , the data determination module 400 includes:

[0157] The curve determination unit 410 is configured to construct a difference curve according to the time-domain feature curve and the load curve corresponding to the preliminary load sequence.

[0158] The data determination unit 420 is configured to determine the second type of abnormal data in the original load sequence according to the difference curve.

[0159] In one embodiment, on the basis of Figure 6 or Figure 7 as shown in Figure 8 , the data determination unit 420 includes:

[0160] The curve determining sub-unit 421 is configured to construct a difference increment curve according to the increments of the difference value curve at any two adjacent time points in the target time period, wherein the difference increment curve is subject to a normal distribution with a mean value of 0.

[0161] The data determining sub-unit 422 is configured to determine the second type of abnormal data in the original load sequence according to a standard deviation of the difference increment curve.

[0162] In an embodiment, the data determining sub-unit 422 is specifically configured to:

[0163] determine the standard deviation of the difference increment curve, take the product of the standard deviation and a set multiple as a screening value, wherein the set multiple is 3, determine an abnormal point in the difference increment curve according to the screening value, and determine the second type of abnormal data in the original load curve according to the abnormal point.

[0164] In an embodiment, the second repair module 500 is specifically configured to:

[0165] extract a feature curve segment corresponding to the first type of abnormal data and a feature curve segment corresponding to the second type of abnormal data from the time domain feature curve, perform linear transformation on the feature curve segment corresponding to the first type of abnormal data to obtain a repair value corresponding to the first type of abnormal data, and perform linear transformation on the feature curve segment corresponding to the second type of abnormal data to obtain a repair value corresponding to the second type of abnormal data.

[0166] In an embodiment, the first repair module 200 is specifically configured to:

[0167] perform preliminary repair on the first type of abnormal data in the original load sequence by using linear interpolation.

[0168] The above various modules in the load data processing apparatus can be realized by software, hardware, and combinations thereof, in whole or in part. The above various modules can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in a computer device in software form, so as to be called and executed by a processor to perform the operations corresponding to the above various modules.

[0169] In an embodiment, a computer device is provided, which can be a server, and an internal structure diagram of the computer device can be as shown in Figure 9As shown in the figure. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store user load data for several consecutive weeks. The network interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to implement a load data processing method.

[0170] Those skilled in the art can understand that, Figure 9 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0171] In one embodiment, a computer device is provided, comprising a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the following steps:

[0172] Obtaining an original load sequence corresponding to a target period;

[0173] Preliminary repairing first-type abnormal data in the original load sequence to obtain a preliminary load sequence;

[0174] Performing frequency domain decomposition on the preliminary load sequence, and constructing a time domain feature curve according to the decomposition result;

[0175] According to the time domain feature curve and the load curve corresponding to the preliminary load sequence, determining second-type abnormal data in the original load sequence; the recognition degree of the second-type abnormal data is lower than that of the first-type abnormal data;

[0176] According to the time domain feature curve, repairing the first-type abnormal data and the second-type abnormal data in the original load sequence.

[0177] In one embodiment, when the processor executes the computer program to perform frequency domain decomposition on the preliminary load sequence and construct a time domain feature curve according to the decomposition result, the following steps are also implemented:

[0178] Extracting a frequency domain daily component and a frequency domain weekly component from the decomposition result; converting the frequency domain daily component into a time domain daily component, and converting the frequency domain weekly component into a time domain weekly component; adding the time domain daily component and the time domain weekly component to obtain the time domain feature curve.

[0179] In one embodiment, when the processor executes the computer program to determine the second type of abnormal data in the original load sequence according to the time-domain feature curve and the load curve corresponding to the preliminary load sequence, the following steps are further implemented:

[0180] According to the time-domain feature curve and the load curve corresponding to the preliminary load sequence, a difference curve is constructed; and according to the difference curve, the second type of abnormal data in the original load sequence is determined.

[0181] In one embodiment, when the processor executes the computer program to determine the second type of abnormal data in the original load sequence according to the difference curve, the following steps are further implemented:

[0182] According to the increment of any two adjacent time points in the target time period in the difference curve, a difference increment curve is constructed; wherein the difference increment curve is subject to a normal distribution with a mean of 0; and according to the standard deviation of the difference increment curve, the second type of abnormal data in the original load sequence is determined.

[0183] In one embodiment, when the processor executes the computer program to determine the second type of abnormal data in the original load sequence according to the standard deviation of the difference increment curve, the following steps are further implemented:

[0184] The standard deviation of the difference increment curve is determined; the product of the standard deviation and a set multiple is taken as a screening value; wherein the set multiple is 3; according to the screening value, an abnormal point in the difference increment curve is determined; and according to the abnormal point, the second type of abnormal data in the original load curve is determined.

[0185] In one embodiment, when the processor executes the computer program to repair the first type of abnormal data and the second type of abnormal data in the original load sequence according to the time-domain feature curve, the following steps are further implemented:

[0186] From the time-domain feature curve, a feature curve segment corresponding to the first type of abnormal data and a feature curve segment corresponding to the second type of abnormal data are extracted; the feature curve segment corresponding to the first type of abnormal data is linearly transformed to obtain a repair value corresponding to the first type of abnormal data, and the feature curve segment corresponding to the second type of abnormal data is linearly transformed to obtain a repair value corresponding to the second type of abnormal data.

[0187] In one embodiment, when the processor executes the computer program to preliminarily repair the first type of abnormal data in the original load sequence, the following steps are further implemented:

[0188] The first type of abnormal data in the original load sequence is preliminarily repaired by linear interpolation.

[0189] In one embodiment, a computer readable storage medium is provided, and the computer readable storage medium stores a computer program. When the computer program is executed by a processor, the following steps are implemented:

[0190] obtaining an original load sequence corresponding to the target period;

[0191] performing preliminary repair on first-type abnormal data in the original load sequence to obtain a preliminary load sequence;

[0192] performing frequency domain decomposition on the preliminary load sequence, and constructing a time domain feature curve according to a decomposition result;

[0193] determining second-type abnormal data in the original load sequence according to the time domain feature curve and a load curve corresponding to the preliminary load sequence; the recognition degree of the second-type abnormal data is lower than that of the first-type abnormal data;

[0194] repairing the second-type abnormal data and the first-type abnormal data in the original load sequence according to the time domain feature curve.

[0195] In one embodiment, when the computer program performing frequency domain decomposition on the preliminary load sequence and constructing a time domain feature curve according to a decomposition result is executed by the processor, the following steps are further implemented:

[0196] extracting a frequency domain daily component and a frequency domain weekly component from the decomposition result; converting the frequency domain daily component into a time domain daily component, and converting the frequency domain weekly component into a time domain weekly component; adding the time domain daily component and the time domain weekly component to obtain the time domain feature curve.

[0197] In one embodiment, when the computer program determining second-type abnormal data in the original load sequence according to the time domain feature curve and a load curve corresponding to the preliminary load sequence is executed by the processor, the following steps are further implemented:

[0198] constructing a difference curve according to the time domain feature curve and the load curve corresponding to the preliminary load sequence; determining the second-type abnormal data in the original load sequence according to the difference curve.

[0199] In one embodiment, when the computer program determining second-type abnormal data in the original load sequence according to the difference curve is executed by the processor, the following steps are further implemented:

[0200] constructing a difference increment curve according to an increment of the difference curve at any two adjacent time instants within the target period; wherein the difference increment curve is subject to a normal distribution with a mean value of 0; determining the second-type abnormal data in the original load sequence according to a standard deviation of the difference increment curve.

[0201] In one embodiment, when the computer program determining second-type abnormal data in the original load sequence according to the standard deviation of the difference increment curve is executed by the processor, the following steps are further implemented:

[0202] Determine a standard deviation of the difference increment curve; take a product of the standard deviation and a set multiple as a screening value; wherein the set multiple is 3; determine an abnormal point in the difference increment curve according to the screening value; and determine second-type abnormal data in the original load curve according to the abnormal point.

[0203] In one embodiment, the computer program, when executed by the processor, further implements the following steps for repairing the first-type abnormal data and the second-type abnormal data in the original load sequence according to the time-domain feature curve:

[0204] extracts a feature curve segment corresponding to the first-type abnormal data and a feature curve segment corresponding to the second-type abnormal data from the time-domain feature curve; performs linear transformation on the feature curve segment corresponding to the first-type abnormal data to obtain a repair value corresponding to the first-type abnormal data, and performs linear transformation on the feature curve segment corresponding to the second-type abnormal data to obtain a repair value corresponding to the second-type abnormal data.

[0205] In one embodiment, the computer program, when executed by the processor, further implements the following steps for preliminarily repairing the first-type abnormal data in the original load sequence:

[0206] uses linear interpolation to preliminarily repair the first-type abnormal data in the original load sequence.

[0207] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0208] obtains an original load sequence corresponding to a target period;

[0209] preliminarily repairs first-type abnormal data in the original load sequence to obtain a preliminary load sequence;

[0210] performs frequency-domain decomposition on the preliminary load sequence, and constructs a time-domain feature curve according to a decomposition result;

[0211] determines second-type abnormal data in the original load sequence according to the time-domain feature curve and a load curve corresponding to the preliminary load sequence; the second-type abnormal data has a lower degree of recognition than the first-type abnormal data;

[0212] repaired the first-type abnormal data and the second-type abnormal data in the original load sequence according to the time-domain feature curve.

[0213] In one embodiment, the computer program, when executed by the processor, further implements the following steps for performing frequency-domain decomposition on the preliminary load sequence and constructing a time-domain feature curve according to a decomposition result:

[0214] extracting a frequency domain daily component and a frequency domain weekly component from the decomposition result; converting the frequency domain daily component into a time domain daily component, and converting the frequency domain weekly component into a time domain weekly component; and adding the time domain daily component and the time domain weekly component to obtain a time domain characteristic curve.

[0215] In one embodiment, the computer program, when executed by the processor, determines the second type of abnormal data in the original load sequence according to the time domain characteristic curve and the load curve corresponding to the preliminary load sequence, further implements the following steps:

[0216] According to the time domain characteristic curve and the load curve corresponding to the preliminary load sequence, a difference curve is constructed; and according to the difference curve, the second type of abnormal data in the original load sequence is determined.

[0217] In one embodiment, the computer program, when executed by the processor, determines the second type of abnormal data in the original load sequence according to the difference curve, further implements the following steps:

[0218] According to the increment of the difference curve at any two adjacent time points within the target time period, a difference increment curve is constructed; wherein the difference increment curve is subject to a normal distribution with a mean of 0; and according to the standard deviation of the difference increment curve, the second type of abnormal data in the original load sequence is determined.

[0219] In one embodiment, the computer program, when executed by the processor, determines the second type of abnormal data in the original load sequence according to the standard deviation of the difference increment curve, further implements the following steps:

[0220] The standard deviation of the difference increment curve is determined; the product of the standard deviation and a set multiple is taken as a screening value; wherein the set multiple is 3; according to the screening value, an abnormal point in the difference increment curve is determined; and according to the abnormal point, the second type of abnormal data in the original load curve is determined.

[0221] In one embodiment, the computer program, when executed by the processor, repairs the first type of abnormal data and the second type of abnormal data in the original load sequence according to the time domain characteristic curve, further implements the following steps:

[0222] From the time domain characteristic curve, a characteristic curve segment corresponding to the first type of abnormal data and a characteristic curve segment corresponding to the second type of abnormal data are extracted; the characteristic curve segment corresponding to the first type of abnormal data is linearly transformed to obtain a repair value corresponding to the first type of abnormal data, and the characteristic curve segment corresponding to the second type of abnormal data is linearly transformed to obtain a repair value corresponding to the second type of abnormal data.

[0223] In one embodiment, the computer program, when executed by the processor, performs preliminary repair on the first type of abnormal data in the original load sequence, further implements the following steps:

[0224] The first type of abnormal data in the original load sequence is preliminarily repaired by linear interpolation.

[0225] It should be noted that the user data (including but not limited to the user's load data) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of the related data need to comply with relevant regulations.

[0226] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments of each method. In the embodiments provided in the present application, any reference to memory, database or other medium can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (Read-Only Memory, ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetic variable memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric memory (Ferroelectric Random Access Memory, FRAM), phase change memory (Phase Change Memory, PCM), graphene memory, etc. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0227] The technical features of the above embodiments can be combined arbitrarily. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present application.

[0228] The above-described embodiments are merely illustrative of several embodiments of the present application, and the description is relatively specific and detailed, but should not be understood as a limitation on the scope of the patent. It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the scope of the present application. Therefore, the scope of protection of the present application should be subject to the appended claims.

Claims

1. A load data processing method, characterized in that, The method includes: Obtain the original load sequence corresponding to the target time period; Preliminary repair is performed on the first type of abnormal data in the original load sequence to obtain a preliminary load sequence; The initial load sequence is decomposed in the frequency domain, and a time-domain characteristic curve is constructed based on the decomposition results; A difference curve is constructed based on the time-domain characteristic curve and the load curve corresponding to the preliminary load sequence; Based on the increment of the difference curve between any two adjacent times within the target time period, a difference increment curve is constructed; wherein, the difference increment curve follows a normal distribution with a mean of 0; Based on the standard deviation of the difference increment curve, a second type of abnormal data in the original load sequence is determined; wherein the identifiability of the second type of abnormal data is lower than that of the first type of abnormal data. Based on the time-domain characteristic curve, the first and second types of abnormal data in the original load sequence are repaired.

2. The method according to claim 1, characterized in that, The step of performing frequency domain decomposition on the preliminary load sequence and constructing time domain characteristic curves based on the decomposition results includes: Extract the daily and weekly components in the frequency domain from the decomposition results; The frequency domain daily component is converted into the time domain daily component, and the frequency domain periodic component is converted into the time domain periodic component. The time-domain daily component and the time-domain weekly component are added together to obtain the time-domain characteristic curve.

3. The method according to claim 1, characterized in that, The step of determining the second type of outlier data in the original load sequence based on the standard deviation of the difference increment curve includes: Determine the standard deviation of the difference increment curve; The product of the standard deviation and a set multiple is used as the screening value; wherein, the set multiple is 3. Based on the selected values, outliers in the difference increment curve are identified; Based on the anomaly points, the second type of abnormal data in the original load curve is determined.

4. The method according to claim 1, characterized in that, The step of repairing the first and second types of abnormal data in the original load sequence based on the time-domain characteristic curve includes: Extract the feature curve segments corresponding to the first type of abnormal data and the feature curve segments corresponding to the second type of abnormal data from the time-domain feature curves; A linear transformation is performed on the feature curve segment corresponding to the first type of abnormal data to obtain the repair value corresponding to the first type of abnormal data, and a linear transformation is performed on the feature curve segment corresponding to the second type of abnormal data to obtain the repair value corresponding to the second type of abnormal data.

5. The method according to claim 1, characterized in that, The preliminary repair of the first type of anomalous data in the original load sequence includes: Linear interpolation is used to perform preliminary repair on the first type of abnormal data in the original load sequence.

6. The method according to any one of claims 1-5, characterized in that, The first category of outlier data includes zero values, null values, near-zero values, continuous constant values, and significantly anomalous step values.

7. A load data processing device, characterized in that, The device includes: The sequence acquisition module is used to acquire the original load sequence corresponding to the target time period; The first repair module is used to perform preliminary repair on the first type of abnormal data in the original load sequence to obtain a preliminary load sequence; The curve construction module is used to perform frequency domain decomposition on the preliminary load sequence and construct time domain characteristic curves based on the decomposition results. The data determination module is used to construct a difference curve based on the time-domain feature curve and the load curve corresponding to the preliminary load sequence; construct a difference increment curve based on the increment of the difference curve at any two adjacent times within the target time period; wherein the difference increment curve follows a normal distribution with a mean of 0; and determine the second type of abnormal data in the original load sequence based on the standard deviation of the difference increment curve; wherein the identifiability of the second type of abnormal data is lower than that of the first type of abnormal data. The second repair module is used to repair the first type of abnormal data and the second type of abnormal data in the original load sequence according to the time-domain characteristic curve.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images