Power line background noise identification method and device based on G3-PLC communication

Abstract

The invention relates to the technical field of intelligent power grids, in particular to a power line background noise recognition method and device based on G3-PLC communication, and the method comprises the steps: obtaining the signal-to-noise ratio time sequence data of a G3-PLC communication subcarrier, calculating a background noise feature value according to the signal-to-noise ratio time sequence data, constructing a background noise feature vector according to the background noise feature value, and carrying out the recognition of the background noise of the G3-PLC communication subcarrier. A K-Means algorithm is used for carrying out clustering analysis on the background noise feature vectors to form a noise mode cluster set with physical significance, and then each cluster obtained by clustering is mapped into a power line background noise type based on a preset noise type mapping rule, so that the accuracy and reliability of power line background noise recognition are remarkably improved, and the power line background noise recognition efficiency is improved. The method is suitable for an actual power grid operation environment.

Description

Technical Field

[0001] The present application relates to the field of smart grid technology, and in particular to a method and device for identifying power line background noise based on G3-PLC communication. Background Art

[0002] G3-PLC is a power line communication (PLC) technology widely used in smart grids, smart meters, industrial automation, and the Internet of Things (IoT). Its core is to use existing power lines for data transmission. However, power lines are essentially designed for power transmission rather than communication, so they have a complex and harsh noise environment. Accurate measurement and analysis of power line background noise is a key prerequisite for ensuring the reliability and stability of PLC communication links. Typical power line background noise can be divided into the following categories: (1) Pulse noise: instantaneous high-amplitude spike interference caused by switching operations (such as appliance start-stop, relay operation, etc.); (2) Narrowband interference: narrow spectrum interference from broadcast signals, wireless communication equipment, or harmonics of power electronic equipment; (3) Periodic noise: regular interference signals generated by power electronic devices such as rectifiers and new energy photovoltaic inverters.

[0003] The existing patent application with publication number CN117411516A proposes a method and system for constructing a low-voltage power line channel noise model. This method separates background noise and impulse noise by reading multiple sets of measured noise data collected at the same sampling point of a specific low-voltage power line communication channel to be modeled, and then determines the type of channel noise model and the corresponding noise generation method. However, the limitation of this method is that its modeling process is mainly based on noise testing and analysis within a specific modeling channel, and fails to fully cover the complexity and diversity of the actual power grid operating environment. Therefore, its measurement results can usually only reflect the typical characteristics under specific modeling conditions, and it is difficult to ensure universality and reliability in the actual power grid environment. Summary of the Invention

[0004] Based on this, it is necessary to provide a power line background noise identification method and device based on G3-PLC communication to address the problem of low reliability of existing power line background noise measurement and analysis methods.

[0005] In a first aspect, the present application provides a method for identifying power line background noise based on G3-PLC communication. The method comprises:

[0006] Step S1, obtaining the signal-to-noise ratio time series data of the G3-PLC communication subcarrier;

[0007] Step S2, calculating a background noise characteristic value according to the signal-to-noise ratio time series data, and constructing a background noise characteristic vector according to the background noise characteristic value;

[0008] Step S3: performing cluster analysis on the background noise feature vector using a K-Means algorithm, and mapping each cluster obtained by clustering to a power line background noise type based on a preset noise type mapping rule.

[0009] Furthermore, step S3 includes:

[0010] Step S31, using the K-Means++ algorithm to select M points from a sample point set as initial cluster centroids; the sample points in the sample point set are the background noise feature vectors;

[0011] Step S32, traversing all sample points, calculating the distances from the sample points to the current M centroids, and reallocating the sample points to the cluster represented by the centroid closest to the sample points;

[0012] Step S33: for each cluster, calculate the average value of all sample points in the cluster in each feature dimension to obtain a new centroid;

[0013] Step S34, repeatedly executing steps S32 to S33 until a preset convergence condition is met, thereby obtaining the noise clustering result.

[0014] Furthermore, the background noise characteristic values ​​include mean, variance, kurtosis, maximum instantaneous drop and power frequency harmonic energy.

[0015] Furthermore, the preset noise type mapping rule includes: if the centroid feature of the cluster satisfies , C impulse Indicates the cluster whose noise type is impulse noise, γ k is the kurtosis of the kth G3-PLC communication subcarrier, then the power line background noise type mapped by the cluster is impulse noise; if the centroid characteristics of the cluster satisfy , C periodic Indicates that the noise type is the power frequency harmonic interference cluster, E periodic，k is the power frequency harmonic energy of the kth G3-PLC communication subcarrier, max(E) represents the maximum value of the power frequency harmonic energy, then the power line background noise type mapped by this cluster is power frequency harmonic interference; if the centroid characteristics of the cluster meet , C narrowband Indicates that the noise type is a cluster of narrowband interference, σ k 2 is the variance of the kth G3-PLC communication subcarrier, then the power line background noise type mapped by this cluster is narrowband interference.

[0016] Furthermore, the power line background noise identification method based on G3-PLC communication also includes:

[0017] Step S4, for different types of power line background noise, background noise suppression is performed according to a preset communication dynamic optimization strategy; the preset communication dynamic optimization strategy includes a spectrum notching strategy, a transmit power adaptive adjustment strategy, and a time domain interleaving depth adaptive adjustment strategy.

[0018] Furthermore, the step S2 includes:

[0019] Step S21, calculating a background noise characteristic value according to the signal-to-noise ratio time series data;

[0020] Step S22, periodically capturing and storing the background noise characteristic value based on a curve configuration object; the attributes of the curve configuration object include a logical name, a buffer, a capture object, a capture period, a sorting method, a sorting object, a number of entries used, and a maximum number of entries; and the methods of the curve configuration object include reset and capture;

[0021] Step S23: constructing a background noise feature vector according to the background noise feature value.

[0022] Furthermore, the background noise characteristic value is stored via a predefined 1-byte data class ID and a 6-byte data identification ID.

[0023] Furthermore, after step S2 and before step S3, the method further includes:

[0024] Step S30: normalize the background noise feature vector.

[0025] In a second aspect, the present application further provides a device for identifying power line background noise based on G3-PLC communication. The device comprises:

[0026] A data acquisition module is used to obtain the signal-to-noise ratio time series data of the G3-PLC communication subcarrier;

[0027] a feature vector construction module, configured to calculate a background noise feature value according to the signal-to-noise ratio time series data, and construct a background noise feature vector according to the background noise feature value;

[0028] The background noise identification module is used to perform cluster analysis on the background noise feature vector using a K-Means algorithm, and map each cluster obtained by clustering to a power line background noise type based on a preset noise type mapping rule.

[0029] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are performed:

[0030] Step S1, obtaining the signal-to-noise ratio time series data of the G3-PLC communication subcarrier;

[0031] Step S2, calculating a background noise characteristic value according to the signal-to-noise ratio time series data, and constructing a background noise characteristic vector according to the background noise characteristic value;

[0032] Step S3: performing cluster analysis on the background noise feature vector using a K-Means algorithm, and mapping each cluster obtained by clustering to a power line background noise type based on a preset noise type mapping rule.

[0033] The above-mentioned power line background noise identification method, device and computer equipment based on G3-PLC communication obtain the signal-to-noise ratio time series data of the G3-PLC communication subcarrier, calculate the background noise characteristic value according to the signal-to-noise ratio time series data, and construct the background noise characteristic vector according to the background noise characteristic value. The K-Means algorithm is used to perform cluster analysis on the background noise characteristic vector to form a noise pattern cluster with physical significance, and each cluster obtained by clustering is mapped to a power line background noise type based on a preset noise type mapping rule, which significantly improves the accuracy and reliability of power line background noise identification and is suitable for actual power grid operation environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 Flowchart of a method for identifying power line background noise based on G3-PLC communication in one embodiment;

[0035] Figure 2 A flow chart of a method for identifying power line background noise based on G3-PLC communication in another embodiment;

[0036] Figure 3 FIG. 4 is a structural block diagram of a power line background noise identification device based on G3-PLC communication in one embodiment. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0038] Example 1

[0039] like Figure 1 As shown, this embodiment provides a method for identifying power line background noise based on G3-PLC communication, comprising the following steps:

[0040] Step S1, obtaining the signal-to-noise ratio time series data of the G3-PLC communication subcarrier.

[0041] Among them, the G3-PLC communication system includes K subcarriers, and the number of subcarriers K varies according to the frequency band. For example, the CENELEC A band has 36 subcarriers and the FCC band has 72 subcarriers. The signal-to-noise ratio time series data includes the signal-to-noise ratio time series of each G3-PLC communication subcarrier. Specifically, the G3-PLC communication subcarrier is obtained from the G3-PLC communication module integrated in the smart electricity meter, and for each G3-PLC communication subcarrier (k=1, 2, ..., K), N signal-to-noise ratio (SNR) samples are collected within the specified time window T. Therefore, the signal-to-noise ratio time series S of the kth G3-PLC communication subcarrier is obtained. k It can be expressed as:

[0042] S k =[SNR k (t1),SNR k (t2),…,SNR k (t i ),…,SNR k (t N )] T ∈R N (1)

[0043] Where, t i is the i-th sampling time point, SNR k (ti) represents the kth G3-PLC communication subcarrier at time t i In a preferred embodiment, the time window T is 15 minutes.

[0044] Step S2: Calculate the background noise characteristic value according to the signal-to-noise ratio time series data, and construct a background noise characteristic vector according to the background noise characteristic value.

[0045] The background noise characteristic values ​​include mean, variance, kurtosis, maximum instantaneous drop, and power frequency harmonic energy. The mean value of the signal-to-noise ratio can reflect the average noise level of the G3-PLC communication subcarrier within the time window T. Its specific calculation formula is shown in Equation (2):

[0046] (2)

[0047] Where μ k Represents the mean signal-to-noise ratio of the kth G3-PLC communication subcarrier.

[0048] The variance of the signal-to-noise ratio can characterize the fluctuation intensity of the signal-to-noise ratio of the G3-PLC communication subcarrier around the mean. Its specific calculation formula is shown in formula (3):

[0049] (3)

[0050] Where, σk 2 represents the variance of the signal-to-noise ratio of the kth G3-PLC communication subcarrier.

[0051] The kurtosis of the signal-to-noise ratio can indicate the sharpness of the impulse noise. Its specific calculation formula is shown in formula (4):

[0052] (4)

[0053] Where, γ k It represents the kurtosis of the signal-to-noise ratio of the kth G3-PLC communication subcarrier.

[0054] The maximum instantaneous drop in the signal-to-noise ratio is used to capture sudden drops in the signal-to-noise ratio, such as strong pulse interference. This value is calculated by calculating the drop in the signal-to-noise ratio between all adjacent sampling points and taking the maximum value. The specific calculation formula is shown in Equation (5):

[0055] (5)

[0056] Where, Δ max，k The maximum instantaneous drop in the signal-to-noise ratio of the kth G3-PLC communication subcarrier, SNR k (t i+1 ) represents the kth G3-PLC communication subcarrier at time t i+1 signal-to-noise ratio.

[0057] The power frequency harmonic energy of the signal-to-noise ratio can quantify the energy intensity of periodic noise, especially the power frequency and its harmonics, on the G3-PLC communication subcarrier. The specific calculation formula is shown in formula (6):

[0058] (6)

[0059] Where, E periodic,k The power frequency harmonic energy of the kth G3-PLC communication subcarrier signal-to-noise ratio is represented, f0 is the grid fundamental frequency, M is the highest harmonic order, f m Indicates the frequency of the mth power frequency harmonic, F is the discrete Fourier transform (DFT), F{S k} represents the discrete Fourier transform of the signal-to-noise ratio time series of the kth G3-PLC communication subcarrier, |F{S k}(f m )| 2 Indicates that at frequency point f m The square of the DFT coefficient amplitude at , that is, the energy of the frequency component.

[0060] Specifically, for each G3-PLC communication subcarrier, the five calculated background noise eigenvalues ​​are used to construct a 5-dimensional background noise feature vector as shown in formula (7):

[0061] f k =[μ k ,σ k 2 ,γ k ,Δ max,k ,E periodic,k ] T ∈R 5 (7)

[0062] Where, f k Represents the background noise feature vector of the kth G3-PLC communication subcarrier.

[0063] Furthermore, steps S1 and S2 can be performed by a smart energy meter. After the smart energy meter calculates the background noise characteristic values, this embodiment also predefines a 1-byte data class ID and a 6-byte data identification ID to store the background noise characteristic values ​​to facilitate storage and capture of the background noise characteristic values. The storage identification definition and storage format definition of the background noise characteristic values ​​are shown in Table 1.

[0064] Table 1

[0065]

[0066] The data class ID enables rapid data screening. For example, the concentrator only needs to read messages with data class ID = 3 to extract all background noise feature values. The data identifier ID precisely identifies the type of specific feature value, allowing the concentrator to read specific background noise feature values ​​in a targeted manner.

[0067] This embodiment implements structured management of background noise characteristic values ​​by defining data class IDs and data identification IDs. The two together constitute the index skeleton of standardized storage, which can greatly improve the efficiency of data organization and retrieval, and solve the technical problem that traditional electricity meters store all data in a pile, resulting in low retrieval efficiency.

[0068] Furthermore, this embodiment also predefines a curve configuration object in the smart energy meter to achieve periodic capture and storage of background noise characteristic values. Specifically, step S2 includes:

[0069] Step S21: Calculate the background noise characteristic value according to the signal-to-noise ratio time series data.

[0070] Step S22: Periodically capture and store background noise characteristic values ​​based on the curve configuration object.

[0071] The curve configuration object follows a standardized data structure model and includes properties such as logical name, buffer, capture object, capture cycle, sorting method, sorting object, number of entries used, maximum number of entries, as well as methods such as reset and capture. The specific definitions of each property and method are as follows:

[0072] Logical name (logical_name): A unique identifier for an object instance. In this embodiment, it is defined as 0.0.72.99.0.255 (using OBIS format).

[0073] Buffer: A queue that stores capture entries. Each entry contains all the data obtained by a capture operation and supports index or range access.

[0074] Capture object (Capture_object): defines the target object and its attributes to be recorded by the capture operation. When the method "capture(data)" is called or the automatic periodic capture is performed, the values ​​of the selected attributes of these capture objects will be copied to the curve buffer. The capture objects in this embodiment are 5 background noise feature values: μ k , σ k 2 ,γ k , Δ max，k , k E periodic,k , (corresponding to OBIS: {1.0.32.17.0.255}, {1.0.42.17.0.255}, {1.0.52.17.0.255}, {1.0.62.17.0.255}, {1.0.72.17.0.255} k and a timestamp of the capture moment (format: year:month:day:hour:minute:second).

[0075] Capture_period: The interval (in seconds) that triggers automatic capture. This is controlled by the smart meter's internal clock, with a capture performed at the end of each period. If the capture period (in seconds) is set to 1 or greater, automatic capture is used. The default capture period, T, is 15 minutes, and can be configured as needed by those skilled in the art.

[0076] Sort method (sort_method): Defines how new entries are added to the buffer. Sorting methods are categorized into two main types: sorted and unsorted. Unsorted methods include fifo (first in first out) and lifo (last in first out). The first in first out strategy overwrites the oldest entry when the buffer is full, while the last in first out strategy overwrites the newest entry when the buffer is full.

[0077] sort_object: When the sort method is sort, the sort object is the specified ordering, which specifies a capture object attribute, such as a timestamp, to be used to sort the buffer entries.

[0078] entries_in_use: Indicates the number of entries stored in the buffer. Initially, or after calling the reset(data) method, it is 0, indicating that there is no entry in the buffer. Each time the capture(data) method is successfully called, entries_in_use increases by 1 until it reaches profile_entries.

[0079] Maximum number of entries (profile_entries): The attribute "profile_entries" specifies the maximum number of entries that can be stored in the buffer. The physical limit of the buffer depends on the capture object. When "profile_entries" changes, the buffer will also be adjusted accordingly and will reject entries that would exceed the maximum buffer size.

[0080] Reset (reset(data)): Executing the "reset(data)" method clears the buffer, causing "entries_in_use" to become 0, indicating no valid entries in the buffer. Calling this method does not trigger any additional operations on the capture object; in particular, it does not reset any capture object properties. The "reset(data)" method is automatically called to clear the buffer when the "capture period" and / or "capture object" properties are modified to 1.

[0081] Capture (capture(data)): Executing the "capture(data)" method reads the relevant attribute values ​​of each captured object and copies them into the buffer. Depending on the "sort_method" and the actual state of the buffer, calling this method will either create a new entry or replace an already less important entry. Whenever the buffer's maximum number of entries is not full, the "entries_in_use" attribute is incremented by 1. Calling this method does not trigger any additional operations on the captured objects; in particular, it does not reset any of the captured object's attributes.

[0082] After the smart energy meter completes the acquisition of its own G3-PLC communication subcarrier signal-to-noise ratio time series data and the construction of the background noise feature vector, it periodically triggers the capture(data) method according to the capture_period configured in the curve configuration object. Each capture stores the five background noise feature values ​​calculated at the current moment together with the timestamp as an entry in the local curve configuration object buffer. The sorting method (such as fifo) ensures that the buffer automatically manages old data when the profile_entries limit is reached.

[0083] The concentrator within the distribution network substation periodically reads the background noise characteristic curve data (i.e., the sequence of entries in the buffer) stored in each connected smart energy meter via the communication network. The concentrator stores each meter's curve data in its local database. The concentrator's database summarizes the historical background noise characteristic curve data for all smart energy meters within the substation, providing a data foundation for subsequent cluster analysis of power line background noise. Specifically, in a preferred embodiment, the background noise characteristic curve data stored in the concentrator database can adopt the structured format shown in Table 2:

[0084] Table 2

[0085]

[0086] In Table 2, the feature value record n ​​contains the complete data of a capture entry, including the timestamp and five background noise feature values.

[0087] Step S23: constructing a background noise feature vector according to the background noise feature value.

[0088] Furthermore, after step S2 and before step S3, the method further includes:

[0089] Step S30: normalize the background noise feature vector.

[0090] Specifically, each background noise feature vector can form a background noise feature matrix as shown in formula (8), and then the background noise feature matrix is ​​standardized according to the formula shown in formula (9) to ensure that the background noise feature vector after normalization obeys the distribution with a mean of 0 and a variance of 1.

[0091] F=[f1,f2,…,f k ] T ∈R K×5 (8)

[0092] , , (9)

[0093] Where F represents the background noise feature matrix, f k is the original background noise feature vector, f k ’ is the background noise feature vector after normalization, μ F is the mean of the original background noise feature vector, σ F is the variance of the original background noise feature vector.

[0094] This embodiment can reduce the dimensional differences and numerical scale deviations between multi-dimensional features by standardizing the background noise feature vector, ensuring that the contributions of all features can be more fairly considered in subsequent clustering analysis, thereby significantly improving the accuracy and reliability of noise classification.

[0095] Step S3: performing cluster analysis on the background noise feature vector using a K-Means algorithm, and mapping each cluster obtained by clustering to a power line background noise type based on a preset noise type mapping rule.

[0096] The objective function in the K-Means algorithm is to minimize the intra-class square error function, which is expressed as shown in formula (10):

[0097] (10)

[0098] Where C i represents the i-th cluster, μ i represents the centroid of the ith cluster, and K represents the number of clusters.

[0099] Specifically, the iterative steps of the K-Means algorithm include:

[0100] Step S31: Use the K-Means++ algorithm to select M points from the sample point set as the initial cluster centroids.

[0101] The sample points in the sample point set are the original background noise feature vectors or the background noise feature vectors after normalization. In particular, this embodiment uses the K-Means++ algorithm to select the initial centroid. The specific method is as follows: 1. Randomly select a sample from the sample point set as the initial centroid; 2. Calculate the shortest distance between each sample and the current centroid, represented by D(x); 3. Calculate the probability of each sample being selected as the next centroid. The probability calculation formula is: , x represents the sample point, X represents the sample set, and the next centroid is selected according to the roulette wheel method. 4. Repeat steps 2 and 3 until M centroids are selected. The M centroids finally obtained are the initial cluster centroids of step S31. The expression of the initial cluster centroid is .

[0102] Step S32, traverse all sample points, calculate the distance between the sample point and the current M centroids, and redistribute the sample point to the cluster represented by the centroid closest to the sample point. The specific calculation process is shown in formula (11):

[0103] (11)

[0104] Where C i (t) represents the i-th cluster at the t-th iteration, μ j (t) represents the centroid of the jth cluster at the tth iteration, ||f k -μ j (t) || is the sample point f k and the center of mass μ j (t) The Euclidean distance between j Indicates the index j for finding the minimum value.

[0105] In step S33, for each cluster, the average value of all sample points in each feature dimension in the cluster is calculated to obtain a new centroid. The specific calculation process is shown in formula (12):

[0106] (12)

[0107] Where μ i (t+1) represents the centroid of the i-th cluster at the t+1-th iteration.

[0108] Step S34, repeating steps S32 to S33 until a preset convergence condition is met, and obtaining a noise clustering result.

[0109] By repeatedly executing steps S32 and S33, the value of the objective function can be gradually reduced until convergence, thereby achieving optimal partitioning of the feature space. Optionally, the preset convergence condition can be that the centroid change is less than a preset convergence threshold or that a maximum number of iterations, such as 100, is reached. When the preset convergence condition is that the centroid change is less than a preset convergence threshold, its expression is shown in formula (13):

[0110] (13)

[0111] Where, ||μ i (t+1) -μ i (t) || represents the Euclidean distance between the new centroid and the old centroid of the i-th cluster, and ε is the preset convergence threshold. In a preferred embodiment, ε=10 -4 .

[0112] The noise clustering result finally obtained includes the cluster label to which each sample point belongs, m clusters and the set of sample points contained in each cluster. In this embodiment, the number of clusters is 3. The preset noise type mapping rules include: if the centroid feature of the cluster satisfies , C impulse Indicates the cluster whose noise type is impulse noise, γ k is the kurtosis of the kth G3-PLC communication subcarrier, then the power line background noise type mapped by this cluster is impulse noise (such as motor start and stop); if the centroid characteristics of the cluster satisfy , C periodic Indicates that the noise type is the power frequency harmonic interference cluster, E periodic，k is the power frequency harmonic energy of the kth G3-PLC communication subcarrier, max(E) represents the maximum value of the power frequency harmonic energy, then the power line background noise type mapped by this cluster is power frequency harmonic interference; if the centroid characteristics of the cluster meet , C narrowband Indicates that the noise type is a cluster of narrowband interference, σ k 2 is the variance of the kth G3-PLC communication subcarrier, then the power line background noise type mapped by this cluster is narrowband interference, see Table 3 for details.

[0113] Table 3

[0114]

[0115] Furthermore, if Figure 2 As shown, the power line background noise identification method based on G3-PLC communication in this embodiment also includes:

[0116] Step S4: for different types of power line background noise, background noise suppression is performed according to a preset communication dynamic optimization strategy.

[0117] The preset communication dynamic optimization strategy includes a spectrum notching strategy, a transmit power adaptive adjustment strategy, and a time domain interleaving depth adaptive adjustment strategy. The spectrum notching strategy specifically disables the G3-PLC communication subcarriers contained in clusters with power frequency harmonic interference and clusters with narrowband interference. The expression for the disabled subcarrier set is shown in Equation (14):

[0118] K disabled =C periodic ∪C narrowband (14)

[0119] Where K disabled Indicates a set of disabled subcarriers. Using a spectrum notching strategy can directly avoid frequency bands contaminated by strong interference, improve the signal-to-noise ratio of effective subcarriers, and thus enhance communication quality.

[0120] The transmit power adaptive adjustment strategy is to increase the transmit power of the G3-PLC communication subcarriers in the cluster where the noise type is narrowband interference, and maintain the normal transmit power of the other G3-PLC communication subcarriers, as shown in formula (15):

[0121] (15)

[0122] Where, P tx (k) represents the transmission power of the kth G3-PLC communication subcarrier, ΔP represents the preset power compensation value, P nominal Indicates the normal transmit power. Adaptive transmit power adjustment can compensate for deep fading caused by narrowband interference, enhancing signal penetration and improving communication quality.

[0123] The adaptive adjustment strategy of time domain interleaving depth is to adjust the time domain interleaving depth of G3-PLC communication subcarriers in the cluster with impulse noise to 32, and the time domain interleaving depth of G3-PLC communication subcarriers in other clusters to 16, as shown in formula (16):

[0124] (16)

[0125] Where D interleave The adaptive adjustment strategy of time-domain interleaving depth increases time redundancy and disperses burst noise over multiple symbol periods for error correction, significantly improving the signal's anti-interference capability and thus enhancing communication quality.

[0126] The power line background noise identification method based on G3-PLC communication in this embodiment obtains the signal-to-noise ratio time series data of the G3-PLC communication subcarrier, calculates the background noise characteristic value based on the signal-to-noise ratio time series data, and constructs a background noise characteristic vector based on the background noise characteristic value. The background noise characteristic vector is clustered using the K-Means algorithm to form a noise pattern cluster with physical significance, and each cluster obtained by clustering is mapped to a power line background noise type based on a preset noise type mapping rule. This significantly improves the accuracy and reliability of power line background noise identification and is suitable for actual power grid operation environments.

[0127] Example 2

[0128] Based on the same inventive concept, this embodiment also provides a power line background noise identification device based on G3-PLC communication for implementing the power line background noise identification method based on G3-PLC communication mentioned above. The implementation solution provided by this device is similar to the implementation solution described in the above method. Therefore, the specific limitations in one or more embodiments of the power line background noise identification device based on G3-PLC communication provided below can refer to the limitations of the power line background noise identification method based on G3-PLC communication above, and will not be repeated here. Specifically, Figure 3 As shown, this embodiment provides a power line background noise identification device based on G3-PLC communication, including: a data acquisition module, a feature vector construction module and a background noise identification module, wherein:

[0129] A data acquisition module is used to obtain the signal-to-noise ratio time series data of the G3-PLC communication subcarrier;

[0130] a feature vector construction module, configured to calculate a background noise feature value according to the signal-to-noise ratio time series data, and construct a background noise feature vector according to the background noise feature value;

[0131] The background noise identification module is used to perform cluster analysis on the background noise feature vector using a K-Means algorithm, and map each cluster obtained by clustering to a power line background noise type based on a preset noise type mapping rule.

[0132] Furthermore, the background noise recognition module is also used to use the K-Means++ algorithm to select M points from the sample point set as the initial cluster centroids; traverse all sample points, calculate the distances from the sample points to the current M centroids, and redistribute the sample points to the clusters represented by the centroids closest to the sample points; for each cluster, calculate the average value of all sample points in the cluster on each feature dimension to obtain a new centroid; step S34, repeat the centroid allocation step (i.e., step S32 in Example 1) and the centroid update step (i.e., step S33 in Example 1) until the preset convergence condition is met to obtain the noise clustering result.

[0133] Furthermore, the power line background noise identification device based on G3-PLC communication also includes a communication dynamic optimization module, which is used to suppress background noise according to a preset communication dynamic optimization strategy for different power line background noise types.

[0134] Furthermore, the feature vector construction module includes a feature value calculation unit, a feature value storage unit and a feature vector construction unit, wherein:

[0135] an eigenvalue calculation unit, configured to calculate a background noise eigenvalue based on the signal-to-noise ratio time series data;

[0136] an eigenvalue storage unit, for periodically capturing and storing background noise eigenvalues ​​based on a curve configuration object;

[0137] The feature vector construction unit is used to construct a background noise feature vector according to the background noise feature value.

[0138] Furthermore, the feature vector construction module also includes a vector normalization processing unit for performing normalization processing on the background noise feature vector.

[0139] Example 3

[0140] This embodiment further provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0141] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0142] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for identifying power line background noise based on G3-PLC communication, characterized in that: The method comprises: Step S1, obtaining the signal-to-noise ratio time series data of the G3-PLC communication subcarrier; Step S2, calculating a background noise characteristic value according to the signal-to-noise ratio time series data, and constructing a background noise characteristic vector according to the background noise characteristic value; Step S3: performing cluster analysis on the background noise feature vector using a K-Means algorithm, and mapping each cluster obtained by clustering to a power line background noise type based on a preset noise type mapping rule.

2. The power line background noise identification method based on G3-PLC communication according to claim 1, characterized in that: The step S3 comprises: Step S31, using the K-Means++ algorithm to select M points from a sample point set as initial cluster centroids; the sample points in the sample point set are the background noise feature vectors; Step S32, traversing all sample points, calculating the distances from the sample points to the current M centroids, and reallocating the sample points to the cluster represented by the centroid closest to the sample points; Step S33: for each cluster, calculate the average value of all sample points in the cluster in each feature dimension to obtain a new centroid; Step S34, repeatedly executing steps S32 to S33 until a preset convergence condition is met, thereby obtaining the noise clustering result.

3. The power line background noise identification method based on G3-PLC communication according to claim 1, characterized in that: The background noise characteristic values ​​include mean, variance, kurtosis, maximum instantaneous drop and power frequency harmonic energy.

4. The method for identifying power line background noise based on G3-PLC communication according to claim 1 or 3, characterized in that: The preset noise type mapping rule includes: if the centroid feature of the cluster satisfies , C impulse Indicates the cluster whose noise type is impulse noise, γ k is the kurtosis of the kth G3-PLC communication subcarrier, then the power line background noise type mapped by the cluster is impulse noise; if the centroid characteristics of the cluster satisfy , C periodic Indicates that the noise type is the power frequency harmonic interference cluster, E periodic，k is the power frequency harmonic energy of the kth G3-PLC communication subcarrier, max(E) represents the maximum value of the power frequency harmonic energy, then the power line background noise type mapped by this cluster is power frequency harmonic interference; if the centroid characteristics of the cluster meet , C narrowband Indicates that the noise type is a cluster of narrowband interference, σ k 2 is the variance of the kth G3-PLC communication subcarrier, then the power line background noise type mapped by this cluster is narrowband interference.

5. The method for identifying power line background noise based on G3-PLC communication according to claim 1, characterized in that: The method further comprises: Step S4, for different types of power line background noise, background noise suppression is performed according to a preset communication dynamic optimization strategy; the preset communication dynamic optimization strategy includes a spectrum notching strategy, a transmit power adaptive adjustment strategy, and a time domain interleaving depth adaptive adjustment strategy.

6. The method for identifying power line background noise based on G3-PLC communication according to claim 1, characterized in that: The step S2 comprises: Step S21, calculating a background noise characteristic value according to the signal-to-noise ratio time series data; Step S22, periodically capturing and storing the background noise characteristic value based on a curve configuration object; the attributes of the curve configuration object include a logical name, a buffer, a capture object, a capture period, a sorting method, a sorting object, a number of entries used, and a maximum number of entries; and the methods of the curve configuration object include reset and capture; Step S23: constructing a background noise feature vector according to the background noise feature value.

7. The method for identifying power line background noise based on G3-PLC communication according to claim 6, characterized in that: The background noise characteristic value is stored via a predefined 1-byte data class ID and a 6-byte data identification ID.

8. The method for identifying power line background noise based on G3-PLC communication according to claim 1, characterized in that: After step S2 and before step S3, the method further includes: Step S30: normalize the background noise feature vector.

9. A power line background noise identification device based on G3-PLC communication, characterized in that: The device comprises: A data acquisition module is used to obtain the signal-to-noise ratio time series data of the G3-PLC communication subcarrier; a feature vector construction module, configured to calculate a background noise feature value according to the signal-to-noise ratio time series data, and construct a background noise feature vector according to the background noise feature value; The background noise identification module is used to perform cluster analysis on the background noise feature vector using a K-Means algorithm, and map each cluster obtained by clustering to a power line background noise type based on a preset noise type mapping rule.

10. 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 8 are implemented.

Images