Power consumption behavior similarity evaluation method and system based on frequency domain load data

Abstract

The invention discloses a power consumption behavior similarity evaluation method and system based on frequency domain load data, and the method comprises the steps: collecting first object data in a first mode, and carrying out the frequency domain analysis; and performing first task selection according to a frequency domain analysis result, and grouping first task selection results through an algorithm. And performing multi-dimensional evaluation on the grouping result to obtain a similarity evaluation result of the first object. According to the method, data redundancy is reduced, remarkable characteristics of power consumption behaviors are highlighted, and efficient support is provided for subsequent task selection. And according to a frequency domain analysis result, performing feature significance sorting, and realizing efficient grouping through a clustering algorithm. The step ensures that the grouping result can reflect the internal similarity of the power consumption behaviors, so that the classification accuracy of the power consumption behaviors is optimized. And the grouping result is evaluated by using a plurality of indexes, so that the scientificity and credibility of evaluation are further improved. The step can provide comprehensive evaluation of grouping quality, and provides a reliable decision basis for subsequent application.

Description

Technical Field

[0001] The present invention relates to the technical field of power data processing, and in particular to a method and system for evaluating similarity of power consumption behavior based on frequency domain load data. Background Art

[0002] With the development of smart grid and energy Internet, in-depth analysis of user electricity consumption behavior has gradually become an important direction to improve the efficiency of power grid operation and optimize the allocation of power resources. In this context, user electricity consumption behavior analysis technology based on load data has emerged. Traditional electricity consumption behavior analysis methods mainly rely on time domain data, and identify their typical electricity consumption patterns by observing user load curves, electricity consumption periodicity and load distribution characteristics. However, time domain analysis methods have limitations when processing high-dimensional and diverse data, and it is difficult to effectively reveal complex electricity consumption behavior characteristics. In contrast, frequency domain analysis can describe the core characteristics of user electricity consumption behavior with less data through harmonic feature extraction, providing a new idea for similarity evaluation.

[0003] In the prior art, the frequency-domain-based electricity consumption behavior analysis method mainly focuses on the feature extraction of a single indicator, ignoring the multidimensionality and dynamic change characteristics of frequency domain data. In addition, the grouping and evaluation methods often rely on simple clustering algorithms, lacking a sufficient and decentralized multidimensional evaluation mechanism. The defects of this method are as follows: first, it fails to fully utilize the harmonic characteristics in the frequency domain data, resulting in incomplete feature extraction; second, there is a lack of scientific evaluation of the grouped results, making the results unreliable and unexplainable. Therefore, how to extract effective features from frequency domain load data and construct a multidimensional evaluation framework based on the frequency domain has become an urgent problem to be solved in the prior art. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: In the prior art, the frequency-domain-based electricity consumption behavior analysis method mainly focuses on the feature extraction of a single indicator, ignoring the multidimensionality and dynamic change characteristics of frequency domain data. In addition, the grouping and evaluation methods often rely on simple clustering algorithms, lacking a sufficient and decentralized multidimensional evaluation mechanism. The defects of this method are as follows: first, the harmonic characteristics in the frequency domain data are not fully utilized, resulting in incomplete feature extraction; second, there is a lack of scientific evaluation of the grouped results, which makes the results lack reliability and interpretability.

[0006] The present invention uses frequency domain analysis to extract harmonic characteristics, overcoming the defect of incomplete feature extraction in traditional methods due to reliance on a single indicator. The present invention introduces multidimensional evaluation indicators to make up for the lack of scientific evaluation of grouping results in existing methods, and significantly improves the reliability and interpretability of grouping results. By combining frequency domain features with multidimensional evaluation, complex electricity consumption behavior patterns can be more accurately portrayed, providing an effective means to solve the bottleneck problem in high-dimensional electricity consumption data analysis.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: a method for evaluating the similarity of power consumption behavior based on frequency domain load data, comprising: collecting first object data in a first mode and performing frequency domain analysis.

[0008] The first task selection is performed according to the frequency domain analysis result, and the first task selection results are grouped through an algorithm.

[0009] A multi-dimensional evaluation is performed on the grouping results to obtain a similarity evaluation result of the first object.

[0010] As a preferred solution of the method for evaluating the similarity of power consumption behavior based on frequency domain load data described in the present invention, wherein: collecting the first object data in the first mode includes collecting the time interval of data sampling in the first mode and calculating the frequency.

[0011] As a preferred solution of the method for evaluating the similarity of power consumption behavior based on frequency domain load data of the present invention, the frequency domain analysis includes further calculating according to the calculated frequency to obtain the maximum frequency domain order. The frequency domain analysis is performed to obtain the frequency domain analysis result.

[0012] As a preferred solution of the method for evaluating the similarity of electricity consumption behavior based on frequency domain load data described in the present invention, the first task selection based on the frequency domain analysis results includes generating frequency domain-based features based on the frequency domain analysis results, and sorting and selecting the most significant features.

[0013] As a preferred solution of the method for evaluating similarity of electricity consumption behavior based on frequency domain load data described in the present invention, the grouping of the first task selection results by an algorithm includes grouping the first mode by an algorithm according to the most significant features selected.

[0014] As a preferred solution of the method for evaluating the similarity of power consumption behavior based on frequency domain load data described in the present invention, wherein: the first mode is a load mode, and the first object data includes but is not limited to user power consumption data.

[0015] Frequency domain analysis is harmonic analysis, and the results of frequency domain analysis are harmonic-based features.

[0016] The first task includes but is not limited to feature extraction.

[0017] As a preferred scheme of the method for evaluating the similarity of electricity consumption behavior based on frequency domain load data described in the present invention, wherein: the multi-dimensional evaluation of the grouping results to obtain the similarity evaluation result of the first object includes using multiple similarity and dispersion indicators to evaluate the clustering results to obtain the similarity evaluation result of the first object.

[0018] A power consumption behavior similarity evaluation system based on frequency domain load data, characterized by: comprising:

[0019] The data collection module collects the first object data in the first mode and performs frequency domain analysis.

[0020] The calculation module selects the first task according to the frequency domain analysis result, and groups the first task selection results through an algorithm.

[0021] The evaluation module performs multi-dimensional evaluation on the grouping results to obtain a similarity evaluation result of the first object.

[0022] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0023] A computer-readable storage medium stores a computer program, which implements the steps of the method described above when executed by a processor.

[0024] The beneficial effects of the present invention are as follows: by collecting the first object data in the first mode and performing frequency domain analysis, the harmonic-based features are extracted, thereby achieving the goal of extracting key information from high-dimensional data. This step can reduce data redundancy, highlight the significant features of electricity consumption behavior, and provide efficient support for subsequent task selection.

[0025] According to the frequency domain analysis results, the features are ranked by significance and efficiently grouped using a clustering algorithm. This step ensures that the grouping results can reflect the inherent similarities of electricity consumption behaviors, thereby optimizing the classification accuracy of electricity consumption behaviors.

[0026] The grouping results are evaluated using multiple indicators (such as MIA, CDI, SI, and DBI), which comprehensively consider similarity and dispersion, further improving the scientificity and credibility of the evaluation. This step can provide a comprehensive evaluation of the grouping quality and provide a reliable decision-making basis for subsequent applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0028] Figure 1 The first embodiment of the present invention provides an overall flow chart of a method and system for evaluating the similarity of power consumption behavior based on frequency domain load data. DETAILED DESCRIPTION

[0029] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0030] Example 1, reference Figure 1 , which is an embodiment of the present invention, provides a method for evaluating the similarity of power consumption behavior based on frequency domain load data, comprising:

[0031] S1: Collecting first object data in a first mode and performing frequency domain analysis.

[0032] In the present invention, the first mode is a load mode, and the first object data is the user's electricity consumption behavior data. The frequency domain analysis is a harmonic analysis.

[0033] Specifically, the user's electricity usage behavior data includes the following:

[0034] Load curve data:

[0035] Includes daily, weekly and monthly electricity load curves, reflecting the time distribution of users' electricity consumption patterns.

[0036] Power consumption data:

[0037] Real-time and historical power consumption data, recording users' electricity demand in different time periods.

[0038] Power quality data:

[0039] Voltage, current, power factor, etc. reflect the power quality status during the user's electricity use process.

[0040] Equipment power consumption data:

[0041] Itemized electricity consumption data of household appliances (such as air conditioners, refrigerators, electric vehicle chargers, etc.).

[0042] Behavioral characteristic data:

[0043] It includes the frequency and duration of electricity consumption, as well as the concentration of electricity consumption in different time periods (such as the difference in electricity consumption between peak hours and non-peak hours).

[0044] Collecting time domain data, consider the time interval T0 for performing load pattern sampling, corresponding to the base frequency f0 = 1 / T0. The data of the M customers considered are in the time interval T0 with event interval T sam Sampling (sampling frequency f sam =1 / T sam ). According to the Nyquist theorem, the maximum meaningful frequency that can be obtained from the sampled data is f max =f sam / 2, thus obtaining the maximum meaningful harmonic order:

[0045] h max =f max / f0=T0 / (2T sam )

[0046] Assume that the time period T0 is 1 day and the sampling time T sam 15min is the most appropriate. Perform harmonic analysis on the representative load pattern sampled, calculate the amplitude and phase of the harmonics by discrete Fourier transform, and take the harmonic order h max Then a set of harmonic-based features are extracted using the amplitude and phase information of the harmonic components.

[0047] Harmonic sorting. By evaluating each harmonic amplitude h=1,...,h max The information dispersion is used to evaluate the most significant harmonic set:

[0048]

[0049] in, Indicates the harmonic amplitude A of each customer hi Relative to its average Then the harmonic order is ranked by index The phase values ​​are not used to sort the harmonic orders because the phase variability is very wide for all harmonic orders h>0.

[0050] It should be noted that the first object includes but is not limited to user electricity usage behavior data, and may also be industrial equipment operation data and public transportation energy consumption data.

[0051] In an optional embodiment of the present invention, the first object data is the operation data of industrial equipment, and power sensors, vibration sensors and temperature sensors are installed to collect power consumption, vibration frequency and temperature changes during the operation of the industrial equipment. The data sampling time interval is set to 1 minute, and the collection frequency is set according to the equipment operation characteristics (for example, 50Hz or 100Hz). Perform a fast Fourier transform (FFT) on the collected power data to extract the main frequency components. Calculate the frequency domain features, including the main frequency of the power signal, the amplitude corresponding to the vibration frequency, and the spectral distribution of the equipment temperature fluctuations. Perform harmonic analysis on the power signal to identify the main and subharmonic components in the operation of the equipment. According to the distribution characteristics of the harmonic amplitude, screen the significant harmonic orders.

[0052] In an optional embodiment of the present invention, the first object data is public transportation energy consumption data, and energy consumption sensors and position sensors are installed to collect energy consumption data and position data of public transportation vehicles (such as electric vehicles). The data sampling interval is set to 10 seconds, and the calculation frequency is dynamically adjusted according to the driving speed and power change characteristics of the vehicle. Discrete Fourier transform (DFT) is applied to the energy consumption data to extract the frequency components of energy consumption fluctuations. The frequency domain characteristics of the vehicle under different operating modes (such as acceleration, deceleration, and steady driving) are calculated. The harmonic characteristics of the energy consumption data are analyzed to identify the main harmonic components during acceleration and deceleration. Based on the significance of the harmonic amplitude, the key harmonic orders within the frequency range are screened.

[0053] S2: Select a first task according to the frequency domain analysis result, and group the first task selection results through an algorithm.

[0054] In the present invention, the frequency domain analysis result is a harmonic feature, and the first task is the selection of the harmonic feature. The algorithm uses a clustering algorithm for grouping.

[0055] The harmonic signature is selected by including the amplitude of the zeroth order harmonic (proportional to the average power) and appropriate values ​​of the amplitude and phase correlation values ​​of some of the top harmonic orders. The amplitudes of the first n harmonics are introduced as the index V h The harmonics are extracted in descending order. Define a set Θ n , which contains the n top harmonic orders involved in the formation of the considered feature set. For example, Θ0 contains only the top-ranked harmonic orders, while Θ 10 Contains the first 10 harmonic orders, arranged in descending order of harmonic order.

[0056] The phase-related variables are not used for harmonic order sorting, but are used as features for clustering purposes. The normalized phase value is multiplied by a weight factor to define the phase-related features for a given harmonic order. The weight factor is the sum of the corresponding harmonic order amplitude and the value of the harmonic order defined by the set θ. nThe ratio of the harmonic amplitudes of all harmonic orders in is determined by the RMS value. Therefore, n The two phase-related characteristics of the hth harmonic order are:

[0057]

[0058] Among them, A j Represents Θ n The jth harmonic amplitude in α h Represents the phase angle of the hth harmonic.

[0059] Various feature sets are then defined based on the harmonic order, including phase-related features. The only exception is the phase of the zeroth order harmonic, which is always zero and is never included in a feature set. This is called feature set H n , the feature set is represented as:

[0060] f n ={(A k ,y k ′,y k ″),(k≠0)∩(k∈Θ n )}∪{(A k ),(k=0)∩(k∈Θ n )}

[0061] Among them, A k Represents Θ n The amplitude of the kth harmonic in y k ′,y k ″ represents the kth phase-related feature.

[0062] For a given value of n, the data set representing the customer daily load pattern m=1,...,M is:

[0063]

[0064] All the data of all customers are:

[0065]

[0066] It should be noted that the first task includes but is not limited to the selection of harmonic features, and may also be frequency domain anomaly detection and power consumption pattern recognition.

[0067] In an optional embodiment of the present invention, the first task is frequency domain anomaly detection, extracting the amplitude and phase characteristics of each harmonic order from the frequency domain analysis results to generate a feature vector. The standard deviation of the amplitude and the rate of change of the phase are used as preliminary indicators for anomaly detection. A statistical threshold-based method is applied, such as defining upper and lower limits of the amplitude and phase, to filter out harmonic orders that exceed the threshold. The feature vector is analyzed using an isolation forest algorithm to mark possible anomalies. A density clustering algorithm (such as DBSCAN) is applied to cluster the feature vectors of the detected abnormal harmonic orders. The clustering results are used to identify the category and distribution of abnormal patterns.

[0068] By comparing with known abnormal patterns in historical data, the abnormal clustering results are classified. After the classification is completed, the abnormal pattern category and possible causes are output.

[0069] In an optional embodiment of the present invention, the first task is to identify the power consumption pattern, and extract the key harmonic features of the daily load pattern from the frequency domain analysis results, including the main harmonic amplitude, phase distribution and frequency change trend. The power consumption periodicity indicators (such as daily peak time, load mean) are calculated in combination with the load characteristics. Support vector machine (SVM) or K-means clustering (K-Means) is used to classify the user's power consumption pattern, which is initially divided into daytime peak type, nighttime peak type, balanced type and other categories. Feature vectors are further extracted based on the classification results, such as the frequency domain feature center of each type of user.

[0070] According to the classification results, the dynamic time warping (DTW) algorithm is applied to further optimize the clustering and eliminate the classification error caused by the time sequence misalignment of power consumption behavior. The final power consumption pattern category and related feature set are generated. For each category, a specific power consumption behavior pattern label is generated for subsequent grouping and multi-dimensional evaluation.

[0071] S3: Perform multi-dimensional evaluation on the grouping results to obtain a similarity evaluation result of the first object.

[0072] The harmonic-based features defined above are used for customer classification by running a clustering algorithm to group the load patterns according to their salient features. A modified follow-the-leader procedure is adopted. The clustering process is driven by a distance threshold and involves iteratively redistributing the customer data into clusters until stable clusters are formed. More specifically, the first loop groups together the load patterns whose modified Euclidean distance does not exceed the distance threshold, automatically determines the final number of clusters and calculates their centroids. Continuous loops redistribute the load patterns to the existing clusters according to the minimum modified Euclidean distance with respect to the existing cluster centroids, updating the left cluster centroid and the inner cluster centroid each time the load pattern changes clusters. The process stops when stable cluster formation is reached.

[0073] Assume that by performing the clustering process using the features contained in the vector y, K customer classes are formed. Let us introduce a subset L containing the initial (time domain) data of k=1,...,K types of load patterns (k) The set R is represented as:

[0074] R = {r (k) ,k=1,...,K}

[0075] where r (k) is calculated by calculating the number of members belonging to the subset L (k) The weighted average of the time domain load patterns is obtained, assuming the reference power of each load pattern as a weighting factor.

[0076] Defining the similarity index requires a certain notion of distance. Consider a representative load pattern r (k) With subset L (k) Distance between:

[0077] d(r (k) ,L (k) )

[0078] Defined as r (k) With L (k) The geometric mean of the distances between each member of , and the base set average distance Defined as L (k) The geometric mean of the distances between members. Based on these distances, a variety of appropriateness indicators are used, including MIA, CDI, SI, and DBI.

[0079] Mean Index Adequacy (MIA), determined by the distance between each representative load pattern and the load pattern belonging to the corresponding cluster:

[0080]

[0081] The Clustering Dispersion Indicator (CDI) depends on the distance between load patterns in the same cluster and the distance between representative load patterns of a class:

[0082]

[0083] Scatter Index (SI):

[0084]

[0085] Where p is the collective scattering:

[0086]

[0087] The Davies-Bouldin index, which represents the system-wide average of the similarity measures of each cluster to its most similar cluster, is expressed in Euclidean form:

[0088]

[0089] A common feature of these indices is that the smaller the value, the higher the adequacy.

[0090] The computer device may be a server. The computer device includes a processor, a memory, an input / output interface (I / O for short) and a communication interface. The processor, the memory and the input / output interface are connected via a system bus, and the communication interface is connected to the system bus via the input / output interface. 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 data cluster data of a power monitoring system. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for evaluating the similarity of power consumption behavior based on frequency domain load data is implemented.

[0091] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and 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 embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0092] Embodiment 2 is an embodiment of the present invention, which provides a method and system for evaluating the similarity of power consumption behavior based on frequency domain load data. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0093] Taking the actual application scenario as an example, the specific steps of implementing the power user behavior analysis using the method of the present invention are as follows:

[0094] A dataset is used to illustrate the results of the proposed method. A stratified sampling method is applied to select 232 customers. A 15-minute data sampling is considered, so that the daily load pattern of each customer contains 96 time domain samples. The data is pre-processed to eliminate the load current data corresponding to abnormal days (e.g., bank holidays occurring on weekdays), strikes, accidents, or failures.

[0095] Perform harmonic analysis on the current data to obtain the frequency domain load feature set.

[0096] The load characteristics in time and frequency domains are comprehensively utilized to cluster the load patterns of power users.

[0097] When n=4, the clustering results of the load patterns of power users using the method of the present invention are shown in Table 1. The results of the adequacy evaluation index are shown in Table 2.

[0098] Table 1 Clustering results of power user load patterns

[0099]

[0100]

[0101] Table 2 Results of adequacy evaluation indicators

[0102] MIA CDI SI DBI 0.08 0.24 3 1.1

[0103] The advantages of the present invention in analyzing the behavior of power users can be clearly seen from the clustering results of power user load patterns in Table 1 and the results of adequacy evaluation indicators in Table 2.

[0104] Table 1 is analyzed as follows:

[0105] The method of the present invention can classify 232 customers into 16 different load pattern categories, most of which (such as categories 1, 2, 3, and 5) contain a large number of customers, while categories 11, 13, and 14 only have a small number of customers. This shows that the present invention has a high ability to distinguish in load pattern clustering and can finely identify the characteristics of customers' electricity consumption behavior. Traditional methods (such as simple time domain feature extraction methods) can usually only roughly classify customers into 3 to 5 categories, and are prone to losing fine-grained information. The present invention refines the load pattern categories and achieves higher resolution by combining frequency domain feature analysis with a clustering algorithm.

[0106] The customer distribution results show that the method of the present invention can identify a variety of electricity consumption behavior patterns, ranging from a small number of specific customers (such as individual customers in categories 9, 11, and 13) to large-scale general patterns (such as 79 customers in category 2). This feature helps power companies formulate more targeted marketing and operation and maintenance strategies. Traditional methods often fail to effectively identify the electricity consumption behavior of a small number of specific customers, which can easily lead to misjudgments or omissions. The present invention can take into account the characteristic analysis of both specific customers and mainstream customer groups.

[0107] Table 2 is analyzed as follows:

[0108] MIA (mean information entropy) is 0.08: Low information entropy indicates that the clustering results of the present invention have high certainty, and the electricity consumption behaviors of customers in each category are more similar. In contrast, the MIA of the prior art is usually above 0.2, and the clustering results are not clear enough.

[0109] CDI (Class Discrimination Index) is 0.24: A lower CDI value indicates good differentiation between clustering categories and no significant overlap between categories. The CDI value of traditional methods is usually around 0.4, with blurred boundaries between categories.

[0110] SI (Silhouette Coefficient) is 3: It means that the clustering effect is significant, the data within the category is highly compact, and the distance between categories is obvious. In contrast, the SI of traditional methods is usually below 2, indicating that the clustering quality is poor.

[0111] DBI (Davidson-Boulding Index) is 1.1: The lower the value, the more compact and reasonable the clustering is. The DBI value of the prior art is usually above 1.8, reflecting the lack of clustering compactness.

[0112] Embodiment 3 is an embodiment of the present invention, including a power consumption behavior similarity evaluation system based on frequency domain load data, specifically:

[0113] The data collection module collects the first object data in the first mode and performs frequency domain analysis.

[0114] The calculation module selects the first task according to the frequency domain analysis result, and groups the first task selection results through an algorithm.

[0115] The evaluation module performs multi-dimensional evaluation on the grouping results to obtain a similarity evaluation result of the first object.

[0116] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for evaluating the similarity of power consumption behavior based on frequency domain load data, characterized in that: include: Collecting first object data in a first mode and performing frequency domain analysis; Selecting a first task according to the frequency domain analysis result, and grouping the first task selection results by an algorithm; A multi-dimensional evaluation is performed on the grouping results to obtain a similarity evaluation result of the first object.

2. The method for evaluating the similarity of power consumption behavior based on frequency domain load data according to claim 1, characterized in that: The collecting of the first object data in the first mode includes collecting the time interval of data sampling in the first mode and calculating the frequency.

3. The method for evaluating the similarity of power consumption behavior based on frequency domain load data according to claim 2, characterized in that: The frequency domain analysis includes further calculating according to the calculated frequency to obtain the maximum frequency domain order; Perform frequency domain analysis to obtain frequency domain analysis results.

4. The method for evaluating the similarity of power consumption behavior based on frequency domain load data according to claim 3, characterized in that: The selecting the first task according to the frequency domain analysis result includes generating frequency domain-based features according to the frequency domain analysis result, and sorting and selecting the most significant features.

5. The method for evaluating the similarity of power consumption behavior based on frequency domain load data according to claim 4, characterized in that: The grouping of the first task selection results by an algorithm includes grouping the first patterns by an algorithm according to the most significant features selected.

6. The method for evaluating the similarity of power consumption behavior based on frequency domain load data according to claim 5, characterized in that: The first mode is a load mode, and the first object data includes but is not limited to user power consumption data; Frequency domain analysis is harmonic analysis, and the results of frequency domain analysis are harmonic-based features; The first task includes but is not limited to feature extraction.

7. The method for evaluating the similarity of power consumption behavior based on frequency domain load data according to claim 6, characterized in that: The performing multi-dimensional evaluation on the grouping results to obtain the similarity evaluation result of the first object includes evaluating the clustering results using multiple similarity and dispersion indicators to obtain the similarity evaluation result of the first object.

8. A system for evaluating the similarity of power consumption behavior based on frequency domain load data using the method according to any one of claims 1 to 7, characterized in that: A data collection module, collecting data of the first object in a first mode and performing frequency domain analysis; A calculation module selects a first task according to the frequency domain analysis result, and groups the first task selection results through an algorithm; The evaluation module performs multi-dimensional evaluation on the grouping results to obtain a similarity evaluation result of the first object.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images