Power line background noise identification method and device based on g3-plc communication

Abstract

The application relates to the technical field of smart grids, in particular to a power line background noise identification method and device based on G3-PLC communication, which comprises the following steps: acquiring signal-to-noise ratio time sequence data of G3-PLC communication subcarriers, calculating background noise characteristic values according to the signal-to-noise ratio time sequence data, constructing a background noise feature vector according to the background noise characteristic values, performing clustering analysis on the background noise feature vector by using a K-Means algorithm, forming a noise mode cluster set with physical significance, and then mapping each cluster obtained by clustering 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 identification are significantly improved, and the method is suitable for actual power grid operation environments.

Description

Technical Field

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

[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 lies in using existing power lines for data transmission. However, power lines are designed for power transmission rather than communication, and therefore 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 mainly divided into the following categories: (1) Impulse noise: instantaneous high-amplitude spike interference caused by switching operations (such as starting and stopping electrical appliances, 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] A patent application with publication number CN117411516A proposes a method and system for constructing a low-voltage power line channel noise model. This method reads multiple sets of measured noise data collected at the same sampling point in a specific low-voltage power line communication channel to be modeled, separates background noise from impulse noise, 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, failing to fully cover the complexity and diversity of the actual power grid operating environment. Therefore, its measurement results usually only reflect typical characteristics under specific modeling conditions, making it difficult to guarantee universality and reliability in the actual power grid environment. Summary of the Invention

[0004] Therefore, 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] Firstly, this application provides a method for identifying power line background noise based on G3-PLC communication. The method includes:

[0006] Step S1: Obtain the signal-to-noise ratio timing data of the G3-PLC communication subcarrier;

[0007] Step S2: Calculate background noise feature values ​​based on the signal-to-noise ratio time series data, and construct a background noise feature vector based on the background noise feature values;

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

[0009] Further, step S3 includes:

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

[0011] Step S32: Traverse all sample points, calculate the distance from the sample point to the current M centroids, and reassign the sample point to the cluster represented by the centroid closest to the sample point;

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

[0013] Step S34: Repeat steps S32 to S33 until the preset convergence condition is met to obtain the noise clustering result.

[0014] Furthermore, the background noise characteristics 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 This represents a cluster whose noise type is impulse noise, γ k If the kurtosis of the k-th G3-PLC communication subcarrier is given, then the power line background noise type mapped by this cluster is impulse noise; if the centroid characteristics of the cluster satisfy... C periodic This indicates a cluster whose noise type is power frequency harmonic interference, E periodic，k Let max(E) represent the maximum value of the power frequency harmonic energy of the k-th G3-PLC communication subcarrier. Then, the power line background noise type mapped by this cluster is power frequency harmonic interference. If the centroid characteristics of the cluster satisfy... C narrowband This indicates a cluster with narrowband interference as the noise type, σ k 2 If the variance of the k-th G3-PLC communication subcarrier is given, 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 notch strategy, an adaptive adjustment strategy for transmit power, and an adaptive adjustment strategy for time domain interleaving depth.

[0018] Further, step S2 includes:

[0019] Step S21: Calculate the background noise characteristic value based on the signal-to-noise ratio time series data;

[0020] Step S22: Periodically capture and store the background noise feature values ​​based on the curve configuration object; the attributes of the curve configuration object include logical name, buffer, capture object, capture period, sorting method, sorting object, number of entries used, and maximum number of entries; the methods of the curve configuration object include reset and capture.

[0021] Step S23: Construct a background noise feature vector based on the background noise feature values.

[0022] Furthermore, the background noise feature values ​​are stored using a predefined 1-byte data class ID and a 6-byte data identifier ID.

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

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

[0025] Secondly, this application also provides a power line background noise identification device based on G3-PLC communication. The device includes:

[0026] The data acquisition module is used to acquire the signal-to-noise ratio timing data of the G3-PLC communication subcarriers;

[0027] The feature vector construction module is used to calculate background noise feature values ​​based on the signal-to-noise ratio time series data, and construct a background noise feature vector based on the background noise feature values;

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

[0029] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps:

[0030] Step S1: Obtain the signal-to-noise ratio timing data of the G3-PLC communication subcarrier;

[0031] Step S2: Calculate background noise feature values ​​based on the signal-to-noise ratio time series data, and construct a background noise feature vector based on the background noise feature values;

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

[0033] The aforementioned power line background noise identification method, device, and computer equipment based on G3-PLC communication acquires the signal-to-noise ratio (SNR) time-series data of the G3-PLC communication subcarriers, calculates background noise feature values ​​based on the SNR time-series data, constructs background noise feature vectors based on the background noise feature values, uses the K-Means algorithm to perform cluster analysis on the background noise feature vectors to form noise pattern clusters with physical meaning, and maps each cluster obtained from clustering 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 applicable to actual power grid operating environments. Attached Figure Description

[0034] Figure 1 This is a flowchart of a power line background noise identification method based on G3-PLC communication in one embodiment;

[0035] Figure 2 This is a flowchart of a power line background noise identification method based on G3-PLC communication in another embodiment;

[0036] Figure 3 This is a structural block diagram of a power line background noise identification device based on G3-PLC communication in one embodiment. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of 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, including the following steps:

[0040] Step S1: Obtain the signal-to-noise ratio timing data of the G3-PLC communication subcarrier.

[0041] The G3-PLC communication system comprises K subcarriers, the number of which varies depending on the frequency band. For example, the CENELEC A band has 36 subcarriers, and the FCC band has 72 subcarriers. The signal-to-noise ratio (SNR) time-series data includes the SNR time series of each G3-PLC communication subcarrier. Specifically, G3-PLC communication subcarriers are obtained from the integrated G3-PLC communication module within the smart meter. For each G3-PLC communication subcarrier (k=1, 2, ..., K), N SNR samples are collected within a specified time window T. Thus, the SNR time series S of the k-th G3-PLC communication subcarrier is obtained. k It can be represented as:

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

[0043] In the formula, t i For the i-th sampling time point, SNR k (ti) represents the k-th G3-PLC communication subcarrier at time t. i The signal-to-noise ratio. In a preferred embodiment, the time window T is 15 minutes.

[0044] Step S2: Calculate the background noise feature value based on the signal-to-noise ratio time series data, and construct the background noise feature vector based on the background noise feature value.

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

[0046] (2)

[0047] In the formula, μ k This represents the average signal-to-noise ratio of the k-th G3-PLC communication subcarrier.

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

[0049] (3)

[0050] In the formula, σk 2 This represents the variance of the signal-to-noise ratio of the k-th G3-PLC communication subcarrier.

[0051] The kurtosis of the signal-to-noise ratio can represent the sharpness of impulse noise, and its specific calculation formula is shown in equation (4):

[0052] (4)

[0053] In the formula, γ k This represents the kurtosis of the signal-to-noise ratio of the k-th G3-PLC communication subcarrier.

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

[0055] (5)

[0056] In the formula, Δ max，k The SNR represents the maximum instantaneous decrease in the signal-to-noise ratio (SNR) of the k-th G3-PLC communication subcarrier. k (t i+1 ) indicates that the k-th G3-PLC communication subcarrier is at time t i+1 The signal-to-noise ratio.

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

[0058] (6)

[0059] In the formula, E periodic,k The power frequency harmonic energy represents the signal-to-noise ratio of the k-th G3-PLC communication subcarrier, where f0 is the fundamental frequency of the power grid, M is the highest harmonic order, and f m Let F represent the frequency of the m-th power frequency harmonic, and F be the Discrete Fourier Transform (DFT). k} represents the discrete Fourier transform of the signal-to-noise ratio time series of the k-th G3-PLC communication subcarrier, |F{S k}(f m )| 2 This indicates the frequency point f. m The square of the DFT coefficient amplitude at that point represents the energy of that frequency component.

[0060] Specifically, for each G3-PLC communication subcarrier, a 5-dimensional background noise feature vector is constructed using the five calculated background noise feature values, as shown in equation (7):

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

[0062] In the formula, f k This represents the background noise feature vector of the k-th G3-PLC communication subcarrier.

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

[0064] Table 1

[0065]

[0066] Among them, the data class ID enables rapid data filtering; for example, the concentrator only needs to read the message with data class ID=3 to extract all background noise feature values. The data identifier ID is used to accurately identify the type of specific feature value, and the concentrator can selectively read specific background noise feature values ​​through the data identifier ID.

[0067] This embodiment achieves structured management of background noise feature values ​​by defining data class ID and data identifier ID. Together, they form a standardized storage index skeleton, which can significantly improve the efficiency of data organization and retrieval, and solve the technical problem of traditional electricity meters storing all data in a haphazard way, resulting in low retrieval efficiency.

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

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

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

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

[0072] Logical name (logical_name): A unique identifier for an object instance. In this example, 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 acquired in a single capture operation and supports indexed or range access to entries.

[0074] Capture object: Defines the target object and its attributes to be recorded during the capture operation. When the method "capture(data)" is called or automatic periodic capture is performed, the values ​​of the selected attributes of these capture objects will be copied into the curve buffer. In this example, the capture objects are five background noise feature values: μ k , σ k 2 ,γ k Δ max，k , k E periodic,k (These correspond to the following OBIS values: {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 the timestamp of the capture time (format: year: month: day: hour: minute: second).

[0075] Capture period (Capture_period): The time interval (in seconds) at which automatic capture is triggered, controlled by the internal clock of the smart meter, and a capture is performed once at the end of each period. If the capture period (in seconds) is set to 1 or more, automatic capture is used. The default capture period T = 15 minutes, which can be configured as needed by those skilled in the art.

[0076] The `sort_method` defines how new entries are added to the buffer. Sorting methods are divided into two main categories: "sorted" and "unsorted." Unsorted methods include FIFO (First In First Out) and LIFO (Last In First Out). FIFO overwrites the oldest entry when the buffer is full, while LIFO overwrites the newest entry when the buffer is full.

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

[0078] Entries in use (entries_in_use): This indicates the number of entries stored in the buffer. It is initially 0 or after calling the reset(data) method, meaning there are no entries in the buffer. Each successful call to the capture(data) method increments entries_in_use 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 limits of the buffer depend primarily on the capture object. When "profile_entries" changes, the buffer must also be adjusted accordingly, and the maximum number of entries that would cause the buffer size to be exceeded will be rejected.

[0080] Reset (reset(data)): Executing the method "reset(data)" clears the buffer, setting "entries_in_use" to 0, and removing all valid entries from the buffer. Calling this method does not trigger any additional operations on the captured object, and in particular, it does not reset any attributes of the captured object. The method "reset(data)" is automatically called to clear the buffer when the attribute "capture period" and / or the attribute "capture object" is modified by 1.

[0081] Capture (capture(data)): The method "capture(data)" reads the relevant attribute values ​​for each captured object and copies them to the buffer. Depending on "sort_method" and the actual state of the buffer, calling this method will generate a new entry or replace an entry that is no longer important. The attribute "entries_in_use" is incremented by 1 as long as the maximum number of entries in the buffer is not full. Calling this method does not trigger any additional operations on the captured object; in particular, it does not reset any of the captured object's attributes.

[0082] After the smart energy meter completes the acquisition of signal-to-noise ratio timing data of its own G3-PLC communication subcarrier and the construction of background noise feature vectors, 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 time along 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 concentrators within the distribution network periodically read the background noise characteristic value curve data (i.e., the sequence of entries in the buffer) stored in each smart energy meter connected to them via the communication network. The concentrators store the curve data of each energy meter in their local database. The concentrator's database summarizes the historical background noise characteristic value curve data of all smart energy meters within the distribution area, providing a data foundation for subsequent cluster analysis of power line background noise. Specifically, in a preferred embodiment, the background noise characteristic value curve data stored in the concentrator's database can adopt a structured format as shown in Table 2:

[0084] Table 2

[0085]

[0086] In Table 2, the feature record n ​​contains complete data within a capture entry, including a timestamp and 5 background noise features.

[0087] Step S23: Construct a background noise feature vector based on the background noise feature values.

[0088] Furthermore, after step S2 and before step S3, the following steps are also included:

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

[0090] Specifically, each background noise feature vector can form a background noise feature matrix as shown in Equation (8), and then the background noise feature matrix is ​​standardized according to the formula shown in Equation (9) to ensure that the background noise feature vector after standardization follows a 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] In the formula, F represents the background noise feature matrix, f k f is the original background noise feature vector. k ’ μ is the standardized background noise feature vector. F σ is the mean of the original background noise feature vector. F This represents the variance of the original background noise feature vector.

[0094] This embodiment standardizes the background noise feature vector, which reduces the dimensional differences and numerical scale biases among multi-dimensional features. This ensures that the contribution of all features can be considered more fairly in subsequent cluster analysis, thereby significantly improving the accuracy and reliability of noise classification.

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

[0096] The objective function in the K-Means algorithm is to minimize the within-class squared error function, and its expression is shown in equation (10):

[0097] (10)

[0098] In the formula, C i Let μ represent the i-th cluster. i Let K represent the centroid of the i-th cluster, and K represent 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] Here, the sample points in the sample point set are either the original background noise feature vectors or the background noise feature vectors after standardization. Specifically, 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, denoted by D(x); 3. Calculate the probability that each sample will be selected as the next centroid. The probability calculation formula is: Let x represent a sample point and X represent the sample set. The next centroid is selected using the roulette wheel method. 4. Repeat steps 2 and 3 until M centroids are selected. The final M centroids are the initial cluster centroids from step S31. The expression for the initial cluster centroids is: .

[0102] Step S32: Traverse all sample points, calculate the distance from each sample point to the current M centroids, and reassign the sample points to the cluster represented by the centroid closest to each sample point. The specific calculation process is shown in equation (11):

[0103] (11)

[0104] In the formula, C i (t) μ represents the i-th cluster in the t-th iteration. j (t) Let ||f| denote the centroid of the j-th cluster at the t-th iteration. k -μ j (t) || is the sample point f k With the center of mass μ j (t) The Euclidean distance between them, argmin j The index j represents the index of the minimum value.

[0105] Step S33: For each cluster, calculate the average value of all sample points within the cluster on each feature dimension to obtain the new centroid. The specific calculation process is shown in equation (12):

[0106] (12)

[0107] In the formula, μ i (t+1) Let represent the centroid of the i-th cluster at the (t+1)-th iteration.

[0108] Step S34: Repeat steps S32 to S33 until the preset convergence condition is met to obtain the noise clustering result.

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

[0110] (13)

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

[0112] The final noise clustering result includes the cluster label of each sample point, 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 rule includes: if the centroid feature of the cluster satisfies C impulse This represents a cluster whose noise type is impulse noise, γ k If the kurtosis of the k-th G3-PLC communication subcarrier is given, then the power line background noise type mapped by this cluster is impulse noise (such as motor start-stop); if the centroid characteristics of the cluster satisfy... C periodic This indicates a cluster whose noise type is power frequency harmonic interference, E periodic，k Let max(E) represent the maximum value of the power frequency harmonic energy of the k-th G3-PLC communication subcarrier. Then, the power line background noise type mapped by this cluster is power frequency harmonic interference. If the centroid characteristics of the cluster satisfy... C narrowband This indicates a cluster with narrowband interference as the noise type, σ k 2 If the variance of the k-th G3-PLC communication subcarrier is given, then the power line background noise type mapped by this cluster is narrowband interference, as shown in Table 3.

[0113] Table 3

[0114]

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

[0116] Step S4: For different types of power line background noise, suppress background noise according to the preset communication dynamic optimization strategy.

[0117] The preset communication dynamic optimization strategies include a spectrum notch strategy, an adaptive transmit power adjustment strategy, and an adaptive time-domain interleaving depth adjustment strategy. Specifically, the spectrum notch strategy disables G3-PLC communication subcarriers contained in clusters with power frequency harmonic interference and clusters with narrowband interference. The expression for the set of disabled subcarriers is shown in equation (14):

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

[0119] In the formula, K disabled This indicates that the set of subcarriers is disabled. Using a spectrum notch filter strategy can directly avoid frequency bands contaminated by strong interference, improve the signal-to-noise ratio of effective subcarriers, and thus improve communication quality.

[0120] The adaptive transmission power adjustment strategy specifically involves increasing the transmission power of the G3-PLC communication subcarriers in the cluster with narrowband interference noise, while maintaining the normal transmission power for the remaining G3-PLC communication subcarriers, as shown in equation (15):

[0121] (15)

[0122] In the formula, P tx (k) represents the transmit power of the k-th G3-PLC communication subcarrier, and ΔP represents the preset power compensation value, P nominal This represents the standard transmit power. Adaptive transmit power adjustment strategies can specifically compensate for deep fading caused by narrowband interference, enhance signal penetration, and thus improve communication quality.

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

[0124] (16)

[0125] In the formula, D interleave The time-domain interleaving depth is defined as the interleaving depth. The adaptive adjustment strategy for time-domain interleaving depth increases time redundancy, distributing burst impulse noise across 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 acquires the signal-to-noise ratio (SNR) time-series data of the G3-PLC communication subcarriers, calculates background noise feature values ​​based on the SNR time-series data, constructs background noise feature vectors based on the background noise feature values, uses the K-Means algorithm to perform cluster analysis on the background noise feature vectors to form noise pattern clusters with physical meaning, and maps each cluster obtained by clustering 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 applicable to actual power grid operating 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 aforementioned power line background noise identification method based on G3-PLC communication. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations of one or more embodiments of the power line background noise identification device based on G3-PLC communication provided below can be found in the limitations of the power line background noise identification method based on G3-PLC communication described above, and will not be repeated here. Specifically, as... 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] The data acquisition module is used to acquire the signal-to-noise ratio timing data of the G3-PLC communication subcarriers;

[0130] The feature vector construction module is used to calculate background noise feature values ​​based on the signal-to-noise ratio time series data, and construct a background noise feature vector based on the background noise feature values;

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

[0132] Furthermore, the background noise identification module is also used to select M points from the sample point set as initial cluster centroids using the K-Means++ algorithm; traverse all sample points, calculate the distance from the sample point to the current M centroids, and reassign the sample points to the cluster represented by the centroid closest to the sample point; for each cluster, calculate the average value of all sample points in the cluster on each feature dimension to obtain a new centroid; in 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 types of power line background noise.

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

[0135] The eigenvalue calculation unit is used to calculate the background noise eigenvalues ​​based on the signal-to-noise ratio time series data.

[0136] The feature value storage unit is used for periodic capture and storage of background noise feature values ​​based on the curve configuration object;

[0137] The feature vector construction unit is used to construct a background noise feature vector based on the background noise feature values.

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

[0139] Example 3

[0140] This embodiment also provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0141] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above 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 embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should 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 includes: Step S1: Obtain the signal-to-noise ratio timing data of the G3-PLC communication subcarrier; Step S2: Calculate background noise feature values ​​based on the signal-to-noise ratio time-series data, and construct a background noise feature vector based on the background noise feature values; specifically: Background noise feature values ​​are constructed using the mean, variance, kurtosis, maximum instantaneous descent, and power frequency harmonic energy of the signal-to-noise ratio time-series data. A background noise feature vector is then constructed based on these feature values. The background noise feature vector is as follows: ； in, This is the background noise feature vector. For the first The mean of the signal-to-noise ratio (SNR) time-series data of each G3-PLC communication subcarrier. For the first The variance of the signal-to-noise ratio timing data of each G3-PLC communication subcarrier For the first Kurtosis of the signal-to-noise ratio time-series data of the G3-PLC communication subcarrier For the first The maximum instantaneous drop in the signal-to-noise ratio (SNR) timing data of each G3-PLC communication subcarrier. For the first Power frequency harmonic energy of the signal-to-noise ratio timing data of each G3-PLC communication subcarrier; Step S3: Use the K-Means algorithm to perform cluster analysis on the background noise feature vector, and map each cluster obtained from the clustering to the power line background noise type based on a preset noise type mapping rule; the preset noise type mapping rule includes: if the centroid feature of the cluster satisfies C impulse This represents a cluster whose noise type is impulse noise, γ k For the first If the kurtosis of the signal-to-noise ratio time-series data of the G3-PLC communication subcarriers is a signal-to-noise ratio, then the power line background noise type mapped by the cluster is impulse noise; if the centroid characteristics of the cluster satisfy... C periodic This indicates a cluster whose noise type is power frequency harmonic interference. For the first If the power frequency harmonic energy of the G3-PLC communication subcarrier signal-to-noise ratio time sequence data is max(E), which 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 satisfy C narrowband This indicates a cluster whose noise type is narrowband interference. For the first If the variance of the signal-to-noise ratio time-series data of each G3-PLC communication subcarrier is used, then the power line background noise type mapped by this cluster is narrowband interference.

2. The method for identifying power line background noise based on G3-PLC communication according to claim 1, characterized in that, Step S3 includes: Step S31: Use the K-Means++ algorithm to select M points from the sample point set as the initial cluster centroids; the sample points in the sample point set are the background noise feature vectors. Step S32: Traverse all sample points, calculate the distance from the sample point to the current M centroids, and reassign the sample point to the cluster represented by the centroid closest to the sample point; Step S33: For each cluster, calculate the average value of all sample points within the cluster on each feature dimension to obtain the new centroid; Step S34: Repeat steps S32 to S33 until the preset convergence condition is met to obtain the noise clustering result.

3. The method for identifying power line background noise based on G3-PLC communication according to claim 1, characterized in that, The background noise characteristics 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, characterized in that, The method further includes: 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 notch strategy, an adaptive adjustment strategy for transmit power, and an adaptive adjustment strategy for time domain interleaving depth.

5. The method for identifying power line background noise based on G3-PLC communication according to claim 1, characterized in that, Step S2 includes: Step S21: Calculate the background noise characteristic value based on the signal-to-noise ratio time series data; Step S22: Periodically capture and store the background noise feature values ​​based on the curve configuration object; the attributes of the curve configuration object include logical name, buffer, capture object, capture period, sorting method, sorting object, number of entries used, and maximum number of entries; the methods of the curve configuration object include reset and capture. Step S23: Construct a background noise feature vector based on the background noise feature values.

6. The power line background noise identification method based on G3-PLC communication according to claim 5, characterized in that, The background noise feature values ​​are stored using a predefined 1-byte data class ID and a 6-byte data identifier ID.

7. 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: Standardize the background noise feature vector.

8. A power line background noise identification device based on G3-PLC communication, characterized in that, The device includes: The data acquisition module is used to acquire the signal-to-noise ratio timing data of the G3-PLC communication subcarriers; The feature vector construction module is used to calculate background noise feature values ​​based on the signal-to-noise ratio time-series data, and to construct a background noise feature vector based on the background noise feature values; specifically: Background noise feature values ​​are constructed using the mean, variance, kurtosis, maximum instantaneous descent, and power frequency harmonic energy of the signal-to-noise ratio time-series data. A background noise feature vector is then constructed based on these feature values. The background noise feature vector is as follows: ； in, This is the background noise feature vector. For the first The mean of the signal-to-noise ratio (SNR) time-series data of each G3-PLC communication subcarrier. For the first The variance of the signal-to-noise ratio timing data of each G3-PLC communication subcarrier for Kurtosis of the signal-to-noise ratio time-series data of the G3-PLC communication subcarrier For the first The maximum instantaneous drop in the signal-to-noise ratio (SNR) timing data of each G3-PLC communication subcarrier. For the first Power frequency harmonic energy of the signal-to-noise ratio timing data of each G3-PLC communication subcarrier; The background noise identification module is used to perform cluster analysis on the background noise feature vector using the K-Means algorithm, and to map each cluster obtained from the clustering to the power line background noise type based on a preset noise type mapping rule; the preset noise type mapping rule includes: if the centroid feature of the cluster satisfies C impulse This represents a cluster whose noise type is impulse noise, γ k For the first If the kurtosis of the signal-to-noise ratio time-series data of the G3-PLC communication subcarriers is a signal-to-noise ratio, then the power line background noise type mapped by the cluster is impulse noise; if the centroid characteristics of the cluster satisfy... C periodic This indicates a cluster whose noise type is power frequency harmonic interference. For the first If the power frequency harmonic energy of the G3-PLC communication subcarrier signal-to-noise ratio time sequence data is max(E), which 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 satisfy C narrowband This indicates a cluster whose noise type is narrowband interference. For the first If the variance of the signal-to-noise ratio time-series data of each G3-PLC communication subcarrier is used, then the power line background noise type mapped by this cluster is narrowband interference.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

Images