Multi-dimensional load label definition and identification method, device and equipment of power system and medium

Abstract

The invention discloses a multi-dimensional load label definition and identification method and device of a power system, equipment and a medium, and relates to the technical field of computers, and the method comprises the steps: carrying out the noise reduction of a collected power signal of the power system, and recognizing a target signal segment with sudden power change in the power signal after noise reduction; generating a power signal space feature based on the geographic information of the denoised power signal, determining a power signal time feature according to the historical power sequence of the historical power signal, and generating a power signal electrical feature according to the denoised power signal and the target signal segment; and carrying out weighted fusion on the generated features, clustering the obtained fused feature set to obtain a load label, coding the load label through a preset coding rule, and identifying the electric power signal according to the obtained target label. Therefore, multi-dimensional characteristics of the load can be comprehensively considered, a standardized load label set is generated, and unified identification and tracking of the fixed access load and the mobile load are realized.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method, apparatus, equipment and medium for defining and identifying multi-dimensional load labels in power systems. Background Technology

[0002] With the rapid penetration of new energy sources, energy storage, electric vehicles, and other new loads into distribution networks, industrial parks, and user-side networks, loads are showing new trends of diverse access methods, complex time characteristics, and increased impact and volatility in electrical behavior: there are fixed access types (factories, industrial parks, residential communities) as well as mobile access types (agricultural well pump trucks, mobile energy storage vehicles, temporary construction power, mobile charging vehicles, etc.); there is obvious seasonality (agricultural loads), daily / weekly periodicity (commercial / office), and quasi-periodicity and randomness of loads such as charging piles; low power factor, impulsive starting current, and strong intermittency and volatility affect power quality and operational safety. Existing methods mostly classify users using coarse-grained criteria such as "user type (residential / industrial / agricultural) + electricity consumption time pattern (power curve)," lacking a unified label and joint identification mechanism for spatial access patterns (fixed / mobile and access point changes), time structure parameterization (seasonal intensity, periodic / autocorrelation structure), and fine-grained electrical characteristics (harmonics, power factor, transient impacts, flicker, etc.). This makes it difficult to meet the needs of refined operation analysis, scheduling, and demand-side management under the new power system. Summary of the Invention

[0003] In view of this, the purpose of this invention is to provide a method, apparatus, equipment, and medium for defining and identifying multi-dimensional load tags in power systems. This method can comprehensively consider the multi-dimensional characteristics of loads, generate a standardized load tag set, and achieve unified identification and tracking of both fixed-connection loads and mobile loads. The specific solution is as follows:

[0004] Firstly, this application discloses a method for defining and identifying multi-dimensional load labels in a power system, including:

[0005] The power signals from the acquired power system are denoised, and target signal segments with power abrupt changes in the denoised power signals are identified; the power signals include transformer area power signals and user power signals.

[0006] Based on the geographic information of the denoised power signal, spatial features of the power signal are generated, and temporal features of the power signal are determined according to the historical power sequence of the historical power signal. Then, electrical features of the power signal corresponding to the denoised power signal are generated based on the denoised power signal and the target signal segment.

[0007] The spatial features, temporal features, and electrical features of the power signal are weighted and fused to obtain a fused feature set.

[0008] The fused feature set is clustered to obtain the load label corresponding to the power signal, and the load label is encoded according to a preset encoding rule to identify the power signal based on the obtained target label.

[0009] Optionally, the step of denoising the acquired power system power signal and identifying target signal segments with power abrupt changes in the denoised power signal includes:

[0010] The user's power data is collected by the first acquisition device based on a preset first acquisition frequency to obtain the user's power signal, and the power data of the transformer area is collected by the second acquisition device based on a preset second acquisition frequency to obtain the transformer area's power signal.

[0011] The user power signal is denoised according to a preset bandpass filter to obtain a denoised user power signal.

[0012] Wavelet threshold denoising is performed on the power signal of the transformer area to obtain the denoised power signal of the transformer area;

[0013] The noise-reduced user power signal and the noise-reduced background power signal are used as the noise-reduced power signal, and target signal segments in the noise-reduced power signal whose power change is greater than a preset power change threshold are identified.

[0014] Optionally, generating spatial features of the power signal based on the geographic information of the denoised power signal includes:

[0015] Determine the power signal type corresponding to the noise-reduced power signal, and determine the target identifier code corresponding to the power signal type from the preset identifier code;

[0016] Determine the load access point corresponding to the noise-reduced power signal, and determine the target latitude and longitude corresponding to the load access point;

[0017] The longitude standard deviation and dimensional standard deviation of the noise-reduced power signal are calculated based on the target latitude and longitude, and the reciprocal of the sum of the longitude standard deviation, the dimensional standard deviation and the preset constant is used as the position stability index of the noise-reduced power signal.

[0018] The number of access point changes of the noise-reduced power signal within a preset period is counted, and the ratio of the number of access point changes to the preset period is taken as the access point change frequency of the noise-reduced power signal.

[0019] A first feature vector is constructed based on the target identifier code, the location stability index, and the access point change frequency, and the first feature vector is used as the spatial feature of the power signal corresponding to the noise-reduced power signal.

[0020] Optionally, determining the time characteristics of the power signal based on the historical power sequence of the historical power signal includes:

[0021] The first amplitude corresponding to the first time period, the second amplitude corresponding to the second time period, and the third amplitude corresponding to the third time period of the historical power signal are determined respectively. An amplitude vector is constructed based on the first amplitude, the second amplitude, and the third amplitude. The amplitude vector is then used as the periodic feature of the power signal. The first time period is a period in days, the second time period is a period in weeks, and the third time period is a period in years.

[0022] The historical average power, historical maximum power, and historical minimum power of the historical power signal within a preset time threshold are statistically analyzed.

[0023] The ratio between the historical average power and the historical maximum power is used as the power signal load factor, and the ratio between the difference between the historical maximum power and the historical minimum power and the historical average power is used as the power signal peak-valley characteristic.

[0024] The average electricity consumption percentage of the historical power signal in each time period within the first time period is statistically analyzed, and the Shannon entropy corresponding to the historical power signal in the first time period is calculated based on the average electricity consumption percentage. Then, the Shannon entropy is used as the time pattern entropy corresponding to the historical power signal.

[0025] The second feature vector is constructed based on the periodic characteristics of the power signal, the load factor of the power signal, the peak and valley characteristics of the power signal, and the time pattern entropy, and the second feature vector is used as the time feature of the power signal corresponding to the denoised power signal.

[0026] Optionally, generating the electrical characteristics of the power signal corresponding to the denoised power signal based on the denoised power signal and the target signal segment includes:

[0027] The rated active power, average active power, average reactive power, and apparent power corresponding to the noise-reduced power signal are determined, and the ratio between the average active power and the apparent power is used as the power factor.

[0028] A power characteristic vector is constructed based on the rated active power, the average active power, the average reactive power, and the power factor.

[0029] The transient energy characteristics of the target signal segment are calculated based on a preset energy integration function, and the harmonic characteristics of the noise-reduced power signal are determined.

[0030] A third feature vector is constructed based on the power characteristic vector, the transient energy characteristics, and the harmonic characteristics, and the third feature vector is used as the electrical characteristics of the power signal corresponding to the noise-reduced power signal.

[0031] Optionally, the weighted fusion of the spatial features, temporal features, and electrical features of the power signal to obtain a fused feature set includes:

[0032] The spatial features, temporal features, and electrical features of the power signal are standardized or normalized, and the processed feature signals are weighted and fused according to preset weights to obtain a fused feature set.

[0033] Optionally, the step of clustering the fused feature set to obtain the load label corresponding to the power signal includes:

[0034] The fused feature set is clustered using a preset clustering algorithm to obtain several clusters;

[0035] Calculate the mean spatial features, mean temporal features, and mean electrical features corresponding to the plurality of clusters, and determine the number of samples within each cluster and the category confidence level for each cluster; the category confidence level is the average posterior probability of the samples within each cluster corresponding to the plurality of clusters.

[0036] Load labels are generated based on the mean of spatial features, the mean of temporal features, the mean of electrical features, the number of samples within each cluster, and the category confidence level corresponding to each cluster.

[0037] Secondly, this application discloses a multi-dimensional load label definition and identification device for a power system, comprising:

[0038] The signal processing module is used to perform noise reduction processing on the power signals of the acquired power system and identify target signal segments with power abrupt changes in the noise-reduced power signals; the power signals include transformer area power signals and user power signals;

[0039] The feature generation module is used to generate spatial features of the power signal based on the geographic information of the denoised power signal, determine the temporal features of the power signal based on the historical power sequence of the historical power signal, and then generate electrical features of the power signal corresponding to the denoised power signal based on the denoised power signal and the target signal segment.

[0040] The feature fusion module is used to perform weighted fusion of the spatial features, temporal features, and electrical features of the power signal to obtain a fused feature set.

[0041] The tag identification module is used to perform clustering processing on the fused feature set to obtain the load tag corresponding to the power signal, and to encode the load tag according to the preset encoding rule so as to identify the power signal according to the obtained target tag.

[0042] Thirdly, this application discloses an electronic device, comprising:

[0043] Memory, used to store computer programs;

[0044] A processor is used to execute the computer program to implement the multi-dimensional load label definition and identification method for power systems as described above.

[0045] Fourthly, this application discloses a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the aforementioned method for defining and identifying multi-dimensional load tags in a power system.

[0046] In this application, noise reduction processing can be performed on the collected power system power signals, and target signal segments with power abrupt changes in the noise-reduced power signals can be identified. The power signals include transformer area power signals and user power signals. Based on the geographical information of the noise-reduced power signals, spatial features of the power signals are generated, and temporal features of the power signals are determined according to the historical power sequences of historical power signals. Then, electrical features of the power signals corresponding to the noise-reduced power signals are generated according to the noise-reduced power signals and the target signal segments. The spatial features, temporal features, and electrical features of the power signals are weighted and fused to obtain a fused feature set. The fused feature set is clustered to obtain load labels corresponding to the power signals, and the load labels are encoded according to preset encoding rules to identify the power signals based on the obtained target labels. Therefore, the method of this application can generate corresponding multi-dimensional features based on the collected power signals, such as spatial features, temporal features, and electrical features of the power signals. These three features are then weighted and fused according to preset weights, and the resulting fused feature set is clustered to obtain corresponding load labels. Finally, the load labels are encoded using preset coding rules to identify the power signals based on the obtained target labels. This approach comprehensively considers the spatial access attributes, temporal regularity, and electrical characteristics of the load. By simultaneously collecting power data from both the transformer substation and user sides, and through denoising and feature extraction, a multi-dimensional load feature vector is formed. Then, through clustering and label mapping mechanisms, a standardized load label set is generated, enabling unified identification and tracking of both fixed-access loads and mobile loads. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0048] Figure 1 This application discloses a flowchart of a method for defining and identifying multi-dimensional load labels in a power system.

[0049] Figure 2 This is a schematic diagram of the structure of a multi-dimensional load label definition and identification device for a power system disclosed in this application;

[0050] Figure 3 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] Existing technologies often use coarse-grained classification based on user type and electricity consumption time patterns, lacking a unified label and joint identification mechanism for spatial access patterns, temporal structure parameterization, and fine-grained electrical characteristics. This makes it difficult to meet the needs of refined operation analysis, scheduling, and demand-side management under the new power system.

[0053] To overcome the aforementioned technical problems, this application discloses a method, device, equipment, and medium for defining and identifying multi-dimensional load tags in a power system. This method can comprehensively consider the multi-dimensional characteristics of loads, generate a standardized load tag set, and achieve unified identification and tracking of fixed-access loads and mobile loads.

[0054] See Figure 1 As shown in the figure, this invention discloses a method for defining and identifying multi-dimensional load labels in a power system, including:

[0055] Step S11: Perform noise reduction processing on the collected power system power signals and identify target signal segments with power abrupt changes in the noise-reduced power signals; the power signals include transformer area power signals and user power signals.

[0056] In this embodiment, the collected power system signal first needs to be denoised to obtain a denoised power signal. Specifically, a first acquisition device collects user power data at a preset first acquisition frequency to obtain the user power signal, and a second acquisition device collects transformer area power data at a preset second acquisition frequency to obtain the transformer area power signal. It should be noted that the first acquisition device is a HEMS (Home Energy Management System) acquisition terminal installed on the user side, i.e., the user terminal, which synchronously samples two voltage / current signals at a sampling rate of 4kHz. The second acquisition device is a PQ (Power Quality) monitoring device installed on the low-voltage bus of the distribution transformer, which synchronously samples three-phase voltage / current signals at a sampling rate of 1kHz. Furthermore, all devices uniformly use GPS (Global Positioning System) time synchronization and local NTP (Network Time Protocol) time synchronization to control time deviation within a specified range. Within.

[0057] Furthermore, the user power signal needs to be denoised using a preset bandpass filter to obtain the denoised user power signal. Specifically, this involves denoising the 4kHz voltage data from the user terminal. Current data After bandpass filtering After removing power frequency and high-frequency noise, the noise-reduced user power signal can be represented as follows: For the power signal of the transformer substation, wavelet threshold denoising is required to obtain the denoised power signal of the transformer substation. Specifically, let the original signal be... wavelet basis is The denoised signal is then represented as: in, For wavelet transform operators, For threshold function, It is the inverse transform operator.

[0058] Finally, the denoised user power signal and the denoised background power signal need to be used as the denoised power signal, and target signal segments in the denoised power signal whose power changes exceed a preset power change threshold need to be identified. It should be noted that after denoising, event detection and slicing are performed on the signal stream. An event slice refers to a time segment of load start-up / shutdown or significant power change, defined as: ,in, This refers to the power mutation amount. The power change threshold, The width of the time window. This is the threshold for the width of the time window.

[0059] Step S12: Generate spatial features of the power signal based on the geographic information of the denoised power signal, determine the temporal features of the power signal based on the historical power sequence of the historical power signal, and then generate electrical features of the power signal corresponding to the denoised power signal based on the denoised power signal and the target signal segment.

[0060] In this embodiment, it is necessary to generate spatial features of the power signal based on the geographic information of the denoised power signal, determine the temporal features of the power signal based on the historical power sequence of the historical power signal, and then generate the electrical features of the power signal corresponding to the denoised power signal based on the denoised power signal and the target signal segment. It should be noted that the spatial features are used to characterize the dynamic attributes and access mode differences of the load at the geographic access level. For fixed and mobile access loads, the system records the geographic coordinates, node number, and connection status change information of the load at the acquisition end, and forms the spatial features of the power signal through multi-index quantification. The specific steps for generating spatial features of the power signal based on the geographic information of the denoised power signal are as follows:

[0061] Determine the power signal type corresponding to the denoised power signal, and identify the target identifier code corresponding to the power signal type from the preset identifier codes. ,in, The representation is as follows:

[0062] ;

[0063] Fixed-access loads, such as residential areas and factories, have stable access locations, while mobile-access loads, such as mobile energy storage vehicles and temporary charging piles, have access locations that change over time.

[0064] Next, it is necessary to determine the load access point corresponding to the noise-reduced power signal, and determine the target latitude and longitude of the load access point. Based on the target latitude and longitude, calculate the longitude standard deviation and latitude standard deviation of the noise-reduced power signal, and use the reciprocal of the sum of the longitude standard deviation, latitude standard deviation, and a preset constant as the position stability index of the noise-reduced power signal. Furthermore, the location stability index is used to measure the geographical stability of the load connection point over a certain period. It is defined as the reciprocal of the standard deviation of the latitude and longitude variation of the load connection point:

[0065] ;

[0066] in, , These are the standard deviations of the longitude and latitude of the load connection point, respectively. To prevent extremely small constants with a denominator of zero.

[0067] Furthermore, it is necessary to count the number of access point changes of the noise-reduced power signal within a preset period, and use the ratio of the number of access point changes to the preset period as the access point change frequency of the noise-reduced power signal. It should be noted that the access point change frequency is used to describe the spatial access activity of mobile loads, and is defined as the number of times the access point changes per unit time:

[0068] ;

[0069] in, In the time window The number of access point changes detected internally; The statistical period can be one week or one month. When A higher value indicates that the load is frequently connected between different nodes, which is typical of a high-mobility load.

[0070] Finally, a first feature vector needs to be constructed based on the target identifier code, the location stability index, and the access point change frequency. This first feature vector will then be used as the spatial feature of the power signal corresponding to the denoised power signal. Therefore, it can Represented as .

[0071] Furthermore, the temporal characteristics aim to quantify the regularity and stability of load electricity consumption behavior over time, reflecting its operating patterns at different time scales. This is achieved through analysis of historical power sequences. It can extract indicators such as periodicity, load stability, and randomness to form a time feature vector. The specific steps for determining the time characteristics of power signals based on historical power sequences are as follows:

[0072] The first amplitude corresponding to the first time period, the second amplitude corresponding to the second time period, and the third amplitude corresponding to the third time period of the historical power signal are determined respectively. An amplitude vector is constructed based on these three amplitudes, and this amplitude vector is then used as the periodic feature of the power signal. The first time period is a daily period, the second time period is a weekly period, and the third time period is a yearly period. It should be noted that autocorrelation analysis is used to extract the amplitude of the dominant periodic component from the long-term power series to reflect the regularity of the load at different time scales.

[0073] ;

[0074] in, This represents the dominant first amplitude corresponding to the first time period (24 hours), reflecting the diurnal peak and valley characteristics; This is the second amplitude corresponding to the second time period (168 hours), reflecting the difference between weekdays and weekends; This represents the third amplitude corresponding to the third time period (8760 hours), reflecting seasonal electricity consumption patterns. The higher the period amplitude, the more significant the temporal regularity of the load. When the dominant frequency component is dispersed, it indicates that the electricity consumption behavior has a strong randomness.

[0075] Then, the historical average power, historical maximum power, and historical minimum power of the historical power signal within a preset time threshold are statistically analyzed. The ratio between the historical average power and the historical maximum power is taken as the power signal load rate, and the ratio between the difference between the historical maximum power and the historical minimum power and the historical average power is taken as the peak-valley characteristic of the power signal.

[0076] It should be noted that the load factor is used to describe the smoothness of load operation and is defined as the ratio of the average load to the maximum load within a specific statistical period.

[0077] ;

[0078] in, The historical average power over the period; This is the highest historical power output within the same period. When... A higher value indicates that the load is operating relatively smoothly; conversely, a lower value indicates that the load has significant intermittency or peak-valley differences.

[0079] It should be noted that the peak-valley characteristic is an indicator that quantifies the volatility of the load curve, and can be expressed as:

[0080] ;

[0081] in This represents the historical minimum power during that period. When... A large value indicates that the load curve fluctuates violently, which has a significant impact on power quality and dispatch balance. This indicator can be used to identify high-impact loads.

[0082] Next, it is necessary to statistically analyze the average electricity consumption percentage of historical power signals for each time period within the first time period, and calculate the Shannon entropy corresponding to the historical power signals within the first time period based on the average electricity consumption percentage. This Shannon entropy is then used as the time pattern entropy corresponding to the historical power signals. It should be noted that, to characterize the randomness of load distribution throughout the day, a day is divided into... Calculate the average electricity consumption percentage for each time period (e.g., 96 15-minute intervals). And based on Shannon entropy, time pattern entropy is defined:

[0083] ;

[0084] in Entropy The higher the value, the more random and less regular the load behavior.

[0085] Finally, a second feature vector is constructed based on the periodic characteristics of the power signal, the load factor of the power signal, the peak and valley characteristics of the power signal, and the time pattern entropy. This second feature vector is then used as the time feature of the power signal after denoising. And it needs to be explained that It can be represented as .

[0086] Furthermore, electrical characteristics are used to quantify the inherent electrical properties and dynamic operating characteristics of the load, serving as the core basis for reflecting the essential differences in equipment and identifying load types. By analyzing the collected voltage, current waveforms, and power sequences, indicators such as power, harmonics, fluctuations, and transient characteristics are extracted to construct electrical characteristics. The specific steps for generating the electrical characteristics of the power signal corresponding to the denoised power signal based on the denoised power signal and the target signal segment are as follows:

[0087] The rated active power, average active power, average reactive power, and apparent power of the power signal after noise reduction are determined. The ratio between the average active power and apparent power is used as the power factor. Then, a power characteristic vector is constructed based on the rated active power, average active power, average reactive power, and power factor. It should be noted that the power characteristic reflects the power behavior of the load under steady-state operation and mainly includes parameters such as rated power, average power, and power factor.

[0088] ;

[0089] in, The rated active power of the equipment is derived from the nameplate or a long-term statistical average. This represents the average active power of the load during the monitoring period; is the average reactive power; is the power factor. It reflects the ratio of active power to apparent power.

[0090] Then, the transient energy characteristics corresponding to the target signal segment need to be calculated based on a preset energy integration function, and the harmonic characteristics (THD, Total Harmonic Distortion) of the power signal after noise reduction need to be determined. It should be noted that, to identify the dynamic characteristics at the moment of equipment start-up and shutdown, transient energy can be defined based on the determined event slices:

[0091] ;

[0092] in , These represent time windows before and after the instantaneous start-up and shutdown points of the equipment. Indicates at the sampling time The active power value, This represents the average active power of the load within the statistical window.

[0093] Harmonic characteristics reflect the degree of impact of load on power quality. By performing a Fast Fourier Transform on the current signal, its total harmonic distortion rate and the content of major harmonic components can be calculated.

[0094] ;

[0095] in This is the effective value of the basic current. Let be the effective value of the nth harmonic current, and N be the maximum harmonic order considered.

[0096] Finally, a third feature vector is constructed based on the power characteristic vector, transient energy characteristics, and harmonic characteristics, and this third feature vector is used as the electrical feature of the power signal corresponding to the denoised power signal. ,and It can be represented as .

[0097] This allows for a comprehensive consideration of the spatial access attributes, temporal regularity, and electrical characteristics of the load, thereby improving the accuracy of multi-dimensional load identification.

[0098] Step S13: Perform weighted fusion of the spatial features, temporal features, and electrical features of the power signal to obtain a fused feature set.

[0099] In this embodiment, the spatial, temporal, and electrical features of the power signal need to be weighted and fused to obtain a fused feature set. Specifically, the spatial, temporal, and electrical features of the power signal need to be standardized and normalized, and then the processed feature signals are weighted and fused according to preset weights to obtain the fused feature set. It should be noted that because the spatial, temporal, and electrical features have different dimensions (e.g., power is measured in watts, location in latitude and longitude, and periodicity is a dimensionless coefficient), direct concatenation will lead to numerical bias between features. Therefore, Z-score normalization or Min-Max normalization methods can be used for feature scaling. Z-score normalization can be expressed as follows:

[0100] ;

[0101] in, Let i represent the i-th data value in the original data. This represents the new data value after standardization. Let be the mean of the i-th feature. Its standard deviation. This method is applicable to feature data that follows an approximately normal distribution.

[0102] Min–Max normalization can be expressed as follows:

[0103] ;

[0104] in, Let i represent the i-th data value in the original data. This represents the new data value after standardization. Represents the maximum value in the original data. This represents the minimum value in the original data. This method is suitable for features with a finite numerical range and uneven distribution. The normalized feature values ​​are uniformly mapped to the [0,1] interval, eliminating the dimensional differences between different features.

[0105] Furthermore, to achieve a comprehensive expression of multi-dimensional features, weights are assigned to the three types of features based on business objectives and scenario requirements. And perform weighted fusion:

[0106] ;

[0107] And the weights satisfy the normalization condition:

[0108] ;

[0109] This weighted fusion mechanism is adjustable and task-oriented, automatically optimizing weight allocation according to different scenarios. The multidimensional feature set after normalization and weighted fusion can be represented as follows: ,and It can be represented as: .

[0110] Step S14: Perform clustering processing on the fused feature set to obtain the load label corresponding to the power signal, and encode the load label according to the preset encoding rule to identify the power signal based on the obtained target label.

[0111] In this embodiment, the fused feature set can be clustered to obtain load labels corresponding to the power signals. These load labels are then encoded using a preset encoding rule to identify the power signals based on the obtained target labels. Specifically, a preset clustering algorithm is used to cluster the fused feature set to obtain several clusters. It should be noted that the K-Means++ clustering algorithm is used to achieve self-organizing classification of load features, clustering the fused feature vector set. The objective function of the K-Means++ clustering algorithm is to minimize the sum of squared Euclidean distances between samples within a cluster and the cluster center.

[0112] ;

[0113] in, This is the mean vector. The number of clusters K is pre-defined based on the load structure and business objectives of the power system, such as residential, industrial, charging pile, and agricultural types.

[0114] After clustering is completed, each cluster To correspond to a load category, it is necessary to generate a multidimensional load label with physical meaning for each category based on the statistical means of samples within each cluster in the spatial, temporal, and electrical dimensions. Therefore, it is necessary to calculate the spatial, temporal, and electrical feature means for several clusters separately, and determine the number of samples within each cluster and the category confidence for each cluster. The category confidence is the average posterior probability of the samples within each cluster. Finally, based on the spatial, temporal, and electrical feature means, the number of samples within each cluster, and the category confidence, load labels for each cluster are generated. The multidimensional load label can be represented as: .

[0115] in, The mean value of spatial characteristics within the cluster represents the typical access type (fixed / mobile) and location stability; The mean of the time characteristics within the cluster represents a typical periodic pattern (such as diurnal, seasonal, or random). The mean electrical characteristics within the cluster represent typical power levels, harmonic characteristics, and fluctuations. This represents the number of samples within the cluster. The class confidence score, defined by the average posterior probability of samples within the cluster, can be expressed as:

[0116] .

[0117] Furthermore, the payload tags need to be encoded using preset encoding rules, and to facilitate cross-system identification and subsequent retrieval, all generated tags adopt a unified standardized encoding rule:

[0118] ;

[0119] L, K, and Type represent the three dimensions of space, time, and type, respectively.

[0120] It should be noted that the tags can be continuously learned and dynamically updated. When a new load type is detected or the characteristic offset of an existing load exceeds a threshold, the tags can be updated accordingly. At this time, the re-clustering and label update process can be automatically triggered:

[0121] ;

[0122] in, For the current load label The previous load labels are used as the basis for the label evolution record. At the same time, the clustering model parameters for each clustering are saved in the cloud, which supports long-term tracking and system adaptive optimization.

[0123] Since the same rules are used to generate tags for the models collected in the power system, the power signals can be identified based on the target tags obtained after encoding.

[0124] In this embodiment, the collected power system power signals can be denoised, and target signal segments with power abrupt changes in the denoised power signals can be identified. The power signals include transformer area power signals and user power signals. Spatial features of the power signals are generated based on the geographical information of the denoised power signals, and temporal features of the power signals are determined based on the historical power sequences of historical power signals. Then, electrical features of the power signals corresponding to the denoised power signals are generated based on the denoised power signals and the target signal segments. The spatial features, temporal features, and electrical features of the power signals are weighted and fused to obtain a fused feature set. The fused feature set is clustered to obtain load labels corresponding to the power signals, and the load labels are encoded using preset encoding rules to identify the power signals based on the obtained target labels. Therefore, the method of this application can generate corresponding multi-dimensional features based on the collected power signals, such as spatial features, temporal features, and electrical features of the power signals. These three features are then weighted and fused according to preset weights, and the resulting fused feature set is clustered to obtain corresponding load labels. Finally, the load labels are encoded using preset coding rules to identify the power signals based on the obtained target labels. This approach comprehensively considers the spatial access attributes, temporal regularity, and electrical characteristics of the load. By simultaneously collecting power data from both the transformer substation and user sides, and through denoising and feature extraction, a multi-dimensional load feature vector is formed. Then, through clustering and label mapping mechanisms, a standardized load label set is generated, enabling unified identification and tracking of both fixed-access loads and mobile loads.

[0125] See Figure 2 As shown in the figure, an embodiment of the present invention discloses a multi-dimensional load label definition and identification device for a power system, comprising:

[0126] The signal processing module 11 is used to perform noise reduction processing on the power signals of the acquired power system and identify target signal segments with power abrupt changes in the noise-reduced power signals; the power signals include transformer area power signals and user power signals.

[0127] The feature generation module 12 is used to generate spatial features of the power signal based on the geographic information of the denoised power signal, determine the temporal features of the power signal based on the historical power sequence of the historical power signal, and then generate electrical features of the power signal corresponding to the denoised power signal based on the denoised power signal and the target signal segment.

[0128] The feature fusion module 13 is used to perform weighted fusion of the spatial features, temporal features, and electrical features of the power signal to obtain a fused feature set.

[0129] The tag identification module 14 is used to perform clustering processing on the fused feature set to obtain the load tag corresponding to the power signal, and to encode the load tag according to a preset encoding rule so as to identify the power signal according to the obtained target tag.

[0130] In this embodiment, the collected power system power signals can be denoised, and target signal segments with power abrupt changes in the denoised power signals can be identified. The power signals include transformer area power signals and user power signals. Spatial features of the power signals are generated based on the geographical information of the denoised power signals, and temporal features of the power signals are determined based on the historical power sequences of historical power signals. Then, electrical features of the power signals corresponding to the denoised power signals are generated based on the denoised power signals and the target signal segments. The spatial features, temporal features, and electrical features of the power signals are weighted and fused to obtain a fused feature set. The fused feature set is clustered to obtain load labels corresponding to the power signals, and the load labels are encoded using preset encoding rules to identify the power signals based on the obtained target labels. Therefore, the method of this application can generate corresponding multi-dimensional features based on the collected power signals, such as spatial features, temporal features, and electrical features of the power signals. These three features are then weighted and fused according to preset weights, and the resulting fused feature set is clustered to obtain corresponding load labels. Finally, the load labels are encoded using preset coding rules to identify the power signals based on the obtained target labels. This approach comprehensively considers the spatial access attributes, temporal regularity, and electrical characteristics of the load. By simultaneously collecting power data from both the transformer substation and user sides, and through denoising and feature extraction, a multi-dimensional load feature vector is formed. Then, through clustering and label mapping mechanisms, a standardized load label set is generated, enabling unified identification and tracking of both fixed-access loads and mobile loads.

[0131] In some embodiments, the signal processing module 11 may specifically include:

[0132] The signal acquisition unit is used to acquire the user's power data through the first acquisition device based on the preset first acquisition frequency to obtain the user's power signal, and to acquire the transformer area's power data through the second acquisition device based on the preset second acquisition frequency to obtain the transformer area's power signal.

[0133] The first signal noise reduction unit is used to perform noise reduction processing on the user power signal according to a preset bandpass filter to obtain the noise-reduced user power signal.

[0134] The second signal noise reduction unit is used to perform wavelet threshold noise reduction on the power signal of the transformer area to obtain the noise-reduced power signal of the transformer area.

[0135] The signal segment determination unit is used to take the noise-reduced user power signal and the noise-reduced background area power signal as the noise-reduced power signal, and identify the target signal segment in the noise-reduced power signal whose power change is greater than a preset power change threshold.

[0136] In some embodiments, the feature generation module 12 may specifically include:

[0137] The identifier code determination unit is used to determine the power signal type corresponding to the noise-reduced power signal, and to determine the target identifier code corresponding to the power signal type from the preset identifier codes;

[0138] The latitude and longitude determination unit is used to determine the load access point corresponding to the noise-reduced power signal and to determine the target latitude and longitude corresponding to the load access point;

[0139] The index determination unit is used to calculate the longitude standard deviation and the dimensional standard deviation of the noise-reduced power signal based on the target latitude and longitude, and to use the reciprocal of the sum of the longitude standard deviation, the dimensional standard deviation and the preset constant as the position stability index of the noise-reduced power signal.

[0140] The frequency change determination unit is used to count the number of access point changes of the noise-reduced power signal within a preset period, and to use the ratio of the number of access point changes to the preset period as the access point change frequency of the noise-reduced power signal.

[0141] The spatial feature construction unit is used to construct a first feature vector based on the target identification code, the location stability index, and the access point change frequency, and to use the first feature vector as the spatial feature of the power signal corresponding to the noise-reduced power signal.

[0142] In some embodiments, the feature generation module 12 may specifically include:

[0143] The periodic feature determination unit is used to determine the first amplitude corresponding to the first time period, the second amplitude corresponding to the second time period, and the third amplitude corresponding to the third time period of the historical power signal, respectively, and to construct an amplitude vector based on the first amplitude, the second amplitude, and the third amplitude, and then use the amplitude vector as the periodic feature of the power signal; wherein, the first time period is a period in days, the second time period is a period in weeks, and the third time period is a period in years;

[0144] A power determination unit is used to calculate the historical average power, historical maximum power, and historical minimum power of the historical power signal within a preset time threshold.

[0145] The power characteristic determination unit is used to take the ratio between the historical average power and the historical maximum power as the power signal load rate, and the ratio between the difference between the historical maximum power and the historical minimum power and the historical average power as the power signal peak-valley characteristics.

[0146] The Shannon entropy calculation unit is used to calculate the average electricity consumption ratio of the historical power signal in each time period within the first time period, and calculate the Shannon entropy corresponding to the historical power signal in the first time period based on the average electricity consumption ratio, and then use the Shannon entropy as the time pattern entropy corresponding to the historical power signal.

[0147] The time feature construction unit is used to construct a second feature vector based on the power signal periodicity, the power signal load factor, the power signal peak-valley characteristics, and the time pattern entropy, and to use the second feature vector as the power signal time feature corresponding to the denoised power signal.

[0148] In some embodiments, the feature generation module 12 may specifically include:

[0149] The power factor calculation unit is used to determine the rated active power, average active power, average reactive power and apparent power corresponding to the noise-reduced power signal, and to use the ratio between the average active power and the apparent power as the power factor.

[0150] A power characteristic vector determination unit is used to construct a power characteristic vector based on the rated active power, the average active power, the average reactive power, and the power factor.

[0151] The harmonic characteristic determination unit is used to calculate the transient energy characteristics corresponding to the target signal segment based on a preset energy integration function, and to determine the harmonic characteristics corresponding to the noise-reduced power signal.

[0152] An electrical feature construction unit is used to construct a third feature vector based on the power characteristic vector, the transient energy feature, and the harmonic characteristic, and to use the third feature vector as the electrical feature of the power signal corresponding to the noise-reduced power signal.

[0153] In some embodiments, the feature fusion module 13 may specifically include:

[0154] The feature fusion unit is used to standardize or normalize the spatial features, temporal features, and electrical features of the power signal, and to perform weighted fusion of the processed feature signals according to preset weights to obtain a fused feature set.

[0155] In some embodiments, the label recognition module 14 may specifically include:

[0156] The feature clustering unit is used to perform clustering processing on the fused feature set using a preset clustering algorithm to obtain several clusters;

[0157] The confidence calculation unit is used to calculate the mean of spatial features, the mean of temporal features, and the mean of electrical features corresponding to the plurality of clusters, and to determine the number of samples within each cluster and the category confidence for each cluster; the category confidence is the average posterior probability of the samples within each cluster corresponding to the plurality of clusters.

[0158] The label generation unit is used to generate load labels corresponding to the several clusters based on the mean of spatial features, the mean of temporal features, the mean of electrical features, the number of samples within the cluster, and the category confidence level.

[0159] Furthermore, embodiments of this application also disclose an electronic device, Figure 3 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0160] Figure 3 This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the multi-dimensional load tag definition and identification method for power systems disclosed in any of the foregoing embodiments. Alternatively, the electronic device 20 in this embodiment may specifically be a computer.

[0161] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0162] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0163] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the multi-dimensional load tag definition and identification method for the power system executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs capable of performing other specific tasks.

[0164] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned method for defining and identifying multi-dimensional load tags in a power system. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0165] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0166] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0167] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0168] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0169] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for defining and identifying multi-dimensional load labels in a power system, characterized in that, include: The power signals from the acquired power system are denoised, and target signal segments with power abrupt changes in the denoised power signals are identified; the power signals include transformer area power signals and user power signals. Based on the geographic information of the denoised power signal, spatial features of the power signal are generated, and temporal features of the power signal are determined according to the historical power sequence of the historical power signal. Then, electrical features of the power signal corresponding to the denoised power signal are generated based on the denoised power signal and the target signal segment. The spatial features, temporal features, and electrical features of the power signal are weighted and fused to obtain a fused feature set. The fused feature set is clustered to obtain the load label corresponding to the power signal, and the load label is encoded according to a preset encoding rule to identify the power signal based on the obtained target label.

2. The method for defining and identifying multi-dimensional load labels in a power system according to claim 1, characterized in that, The process of denoising the acquired power system signal and identifying target signal segments with power abrupt changes in the denoised power signal includes: The user's power data is collected by the first acquisition device based on a preset first acquisition frequency to obtain the user's power signal, and the power data of the transformer area is collected by the second acquisition device based on a preset second acquisition frequency to obtain the transformer area's power signal. The user power signal is denoised according to a preset bandpass filter to obtain a denoised user power signal. Wavelet threshold denoising is performed on the power signal of the transformer area to obtain the denoised power signal of the transformer area; The noise-reduced user power signal and the noise-reduced background power signal are used as the noise-reduced power signal, and target signal segments in the noise-reduced power signal whose power change is greater than a preset power change threshold are identified.

3. The method for defining and identifying multi-dimensional load tags in a power system according to claim 1, characterized in that, The generation of spatial features of the power signal based on the geographic information of the denoised power signal includes: Determine the power signal type corresponding to the noise-reduced power signal, and determine the target identifier code corresponding to the power signal type from the preset identifier code; Determine the load access point corresponding to the noise-reduced power signal, and determine the target latitude and longitude corresponding to the load access point; The longitude standard deviation and dimensional standard deviation of the noise-reduced power signal are calculated based on the target latitude and longitude, and the reciprocal of the sum of the longitude standard deviation, the dimensional standard deviation and the preset constant is used as the position stability index of the noise-reduced power signal. The number of access point changes of the noise-reduced power signal within a preset period is counted, and the ratio of the number of access point changes to the preset period is taken as the access point change frequency of the noise-reduced power signal. A first feature vector is constructed based on the target identifier code, the location stability index, and the access point change frequency, and the first feature vector is used as the spatial feature of the power signal corresponding to the noise-reduced power signal.

4. The method for defining and identifying multi-dimensional load tags in a power system according to claim 1, characterized in that, The method of determining the time characteristics of power signals based on historical power sequences includes: The first amplitude corresponding to the first time period, the second amplitude corresponding to the second time period, and the third amplitude corresponding to the third time period of the historical power signal are determined respectively. An amplitude vector is constructed based on the first amplitude, the second amplitude, and the third amplitude. The amplitude vector is then used as the periodic feature of the power signal. The first time period is a period in days, the second time period is a period in weeks, and the third time period is a period in years. The historical average power, historical maximum power, and historical minimum power of the historical power signal within a preset time threshold are statistically analyzed. The ratio between the historical average power and the historical maximum power is used as the power signal load factor, and the ratio between the difference between the historical maximum power and the historical minimum power and the historical average power is used as the power signal peak-valley characteristic. The average electricity consumption percentage of the historical power signal in each time period within the first time period is statistically analyzed, and the Shannon entropy corresponding to the historical power signal in the first time period is calculated based on the average electricity consumption percentage. Then, the Shannon entropy is used as the time pattern entropy corresponding to the historical power signal. The second feature vector is constructed based on the periodic characteristics of the power signal, the load factor of the power signal, the peak and valley characteristics of the power signal, and the time pattern entropy, and the second feature vector is used as the time feature of the power signal corresponding to the denoised power signal.

5. The method for defining and identifying multi-dimensional load tags in a power system according to claim 1, characterized in that, The step of generating the electrical characteristics of the power signal corresponding to the denoised power signal based on the denoised power signal and the target signal segment includes: The rated active power, average active power, average reactive power, and apparent power corresponding to the noise-reduced power signal are determined, and the ratio between the average active power and the apparent power is used as the power factor. A power characteristic vector is constructed based on the rated active power, the average active power, the average reactive power, and the power factor. The transient energy characteristics of the target signal segment are calculated based on a preset energy integration function, and the harmonic characteristics of the noise-reduced power signal are determined. A third feature vector is constructed based on the power characteristic vector, the transient energy characteristics, and the harmonic characteristics, and the third feature vector is used as the electrical characteristics of the power signal corresponding to the noise-reduced power signal.

6. The method for defining and identifying multi-dimensional load tags in a power system according to claim 1, characterized in that, The weighted fusion of the spatial features, temporal features, and electrical features of the power signal to obtain a fused feature set includes: The spatial features, temporal features, and electrical features of the power signal are standardized or normalized, and the processed feature signals are weighted and fused according to preset weights to obtain a fused feature set.

7. The method for defining and identifying multi-dimensional load labels in a power system according to any one of claims 1 to 6, characterized in that, The process of clustering the fused feature set to obtain the load label corresponding to the power signal includes: The fused feature set is clustered using a preset clustering algorithm to obtain several clusters; Calculate the mean spatial features, mean temporal features, and mean electrical features corresponding to the plurality of clusters, and determine the number of samples within each cluster and the category confidence level for each cluster; the category confidence level is the average posterior probability of the samples within each cluster corresponding to the plurality of clusters. Load labels are generated based on the mean of spatial features, the mean of temporal features, the mean of electrical features, the number of samples within each cluster, and the category confidence level corresponding to each cluster.

8. A multi-dimensional load label definition and identification device for a power system, characterized in that, include: The signal processing module is used to perform noise reduction processing on the power signals of the acquired power system and identify target signal segments with power abrupt changes in the noise-reduced power signals; the power signals include transformer area power signals and user power signals; The feature generation module is used to generate spatial features of the power signal based on the geographic information of the denoised power signal, determine the temporal features of the power signal based on the historical power sequence of the historical power signal, and then generate electrical features of the power signal corresponding to the denoised power signal based on the denoised power signal and the target signal segment. The feature fusion module is used to perform weighted fusion of the spatial features, temporal features, and electrical features of the power signal to obtain a fused feature set. The tag identification module is used to perform clustering processing on the fused feature set to obtain the load tag corresponding to the power signal, and to encode the load tag according to the preset encoding rule so as to identify the power signal according to the obtained target tag.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the multi-dimensional load label definition and identification method for a power system as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store a computer program, wherein the computer program, when executed by a processor, implements the multi-dimensional load label definition and identification method for a power system as described in any one of claims 1 to 7.

Images