A physical topology identification method for a transformer area based on dual-mode communication and characteristic current

Abstract

The application discloses a transformer area physical topology identification method based on dual-mode communication and characteristic current, and relates to the technical field of power line carrier communication. The existing transformer area topology identification mainly relies on manual filing or voltage correlation analysis. In order to accurately establish a four-level physical topology of 'transformer-substation-box-house', a dual-mode communication module with a house transformer function is installed at all nodes, each node constructs a physical fingerprint vector by measuring the dual-mode signal strength, unsupervised clustering is performed on the electric meter with the meter box switch as the clustering center, and the attribution relationship between the meter box and the electric meter is identified. After clustering is completed, the CCO commands the modules in the meter box switch to inject PWM characteristic current into the power line one by one, the branch gateway identifies the characteristic current source by matching filtering demodulation, and the belonging branch of the meter box switch is determined. The CCO correlates and fuses the clustering result and the demodulation result, generates a four-level topology, and reports the four-level topology to the concentrator, so that the automatic identification of the transformer area physical topology of the low-voltage distribution network is realized.

Description

Technical Field

[0001] This application relates to the field of power line dual-mode carrier communication technology, and in particular to a method for identifying the physical topology of transformer substations based on dual-mode communication and characteristic current. Background Technology

[0002] The physical topology of a distribution network area describes the electrical connections and hierarchical structure between transformers, branches, meter boxes, and user meters, and is one of the fundamental data sets for power grid operation and management. Accurately obtaining the physical topology information of a distribution area provides support for business scenarios such as fault location, outage range analysis, and line loss calculation.

[0003] Currently, physical topology identification of low-voltage distribution areas mostly uses HPLC or RF communication alone, without effectively integrating and complementing the advantages of both. The signal attenuation characteristics of HPLC within the meter box can be used for short-range clustering, while the rapid attenuation of RF signal strength with distance can be used to distinguish different meter boxes. Combining the two can significantly improve identification accuracy, but existing technologies have not fully utilized the characteristics of both.

[0004] Under the same branch, traditional methods usually use line impedance calculation or voltage correlation analysis. However, line impedance is easily affected by changes in ambient temperature and load, and voltage curves have high similarity within the transformer area, resulting in low identification accuracy and inaccurate determination of meter box ownership.

[0005] In existing characteristic current injection schemes, signal conflicts are likely to occur when multiple nodes inject characteristic current simultaneously, which can cause upper-layer devices to fail to demodulate and identify accurately, thus affecting the accuracy of identification.

[0006] Most existing solutions can only identify the two-level relationship between "transformer and household". Therefore, a physical topology identification method for transformer areas is needed to establish a complete hierarchical relationship between "branch gateway - meter box switch - meter" and thus realize a complete four-level physical topology. Summary of the Invention

[0007] The purpose of this application is to overcome the shortcomings of the prior art and provide a method for identifying the physical topology of distribution areas based on dual-mode communication and characteristic current, so as to realize the automatic identification and verification of the four-level physical topology of "transformer-branch gateway-meter box switch-user" in low-voltage distribution network.

[0008] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: a four-level hierarchical communication network including transformers, branch gateways, meter box switches and meters is constructed in the low-voltage distribution area. Each level node is equipped with a unified hardware platform that integrates HPLC and RF dual-mode communication modules and a household transformer identification function module.

[0009] A method for identifying the physical topology of transformer substations based on dual-mode communication and characteristic current includes the following steps:

[0010] Step 1: After each meter and meter box switch module is connected to the network, the CCO enables physical topology identification. The meter and meter box switch modules measure the signal strength value (RSSI) with surrounding nodes through two channels: HPLC and RF. HPLC and RSSI RF The two values ​​are then combined to form a two-dimensional fingerprint vector and reported to the CCO.

[0011] Step 2: After the CCO collects the fingerprint vectors of all nodes, it uses the meter box switch as the cluster center and uses the DBSCAN algorithm or the mean drift algorithm to cluster the meters. After the clustering is completed, the meters in the same cluster are determined to be in the same meter box, and the correspondence between the meters and the meter boxes is obtained.

[0012] Step 3: The CCO sends characteristic current commands to each module in the meter box switch. The selected meter box switch module injects a series of characteristic currents in the form of PWM pulses into the power line by controlling the switching of the internal load resistor within its allocated time slot. The unique ID of the meter box switch is encoded in the current pulse train.

[0013] Step 4: After the module of a certain meter box switch has sent the characteristic current, each branch gateway continues to sample the current on the power line. When the above PWM characteristic current is detected, the meter box ID carried in it is demodulated by matched filtering and related despreading. The branch gateway that can demodulate the ID is the branch to which the meter box switch belongs.

[0014] Step 5: The CCO executes Step 3 and Step 4 sequentially on all modules in the meter box switches, and merges the box-household relationship obtained in Step 2 with the branch-box relationship obtained in Step 4 to obtain a four-level physical topology from transformer to branch, from branch to meter box, and from meter box to meter.

[0015] Furthermore, the fingerprint vector used in step two for clustering is first normalized in actual operation, and then the similarity is measured by Mahalanobis distance or Euclidean distance between the meters. The two key parameters of the clustering algorithm, neighborhood radius and minimum number of samples, are set according to the empirical value of the average spacing between meters in the field.

[0016] Furthermore, in step three, when the module in the meter box switch generates the PWM characteristic current, a time-division and code-division hybrid approach is used to avoid conflicts. That is, each meter box switch module is allocated an independent injection time slot, and each PWM pulse sequence is embedded with mutually orthogonal PN codes or GOLD codes. Different meter boxes use different code sequences.

[0017] Furthermore, the PWM characteristic current injection in step three is strictly synchronized within a 1-millisecond time window before and after the zero-crossing point of the power line voltage. At this time, the line current near the zero-crossing point is close to zero, and the interference of the power frequency load on the characteristic current is minimized.

[0018] Furthermore, before issuing the characteristic current transmission command in step three, the CCO first queries the current load of the switch currently connected to the meter box. If the load exceeds 80% of the rated value, the injection is postponed until the load drops, so as to avoid the characteristic current causing a perceptible impact on normal power consumption.

[0019] Furthermore, in step three, if a module of a certain switch box does not respond with a characteristic current transmission command, the CCO determines that the HPLC channel communication is abnormal and automatically switches to the RF channel to resend the command, thereby improving the recognition success rate under poor channel conditions by utilizing the dual-mode characteristics.

[0020] Furthermore, the specific steps for demodulating the characteristic current of the branch gateway in step four are as follows: First, the current waveform is acquired by the current transformer (CT) at a sampling rate of not less than 50kHz. After bandpass filtering and amplification, the PN code stored locally is used to perform a sliding correlation operation with the sampled signal. When the correlation peak value exceeds the preset adaptive threshold value and meets the peak-to-average power ratio (PAPR) determination condition, it is determined to be a valid signal and the meter box ID is decoded. The adaptive threshold value is dynamically adjusted according to the background noise level of the current line.

[0021] Furthermore, in step four, if the PWM signal emitted by the module in a meter box switch is simultaneously demodulated and identified by two or more branch gateways, the CCO will compare the signal strength values ​​reported by different branch gateways and consider the gateway with the strongest signal as the subordinate branch of the meter box; if the difference between the strongest and second strongest signals is less than a preset threshold, the module in the meter box switch will be asked to resend the characteristic current to eliminate ambiguity.

[0022] Furthermore, step five includes a verification step: the concentrator sends a query to each meter based on the generated topology tree, allowing the meter module to measure the dual-mode communication quality from itself to its corresponding meter box switch. If the measured value deviates from the actual value in the topology tree by more than a certain range, the CCO will resend the clustering or characteristic current injection operation to that meter or meter box switch.

[0023] Furthermore, after the physical topology is generated in step five, it enters dynamic maintenance mode: when a new node is added to the network, only the local identification of steps one to four is performed on the new node, and the result is merged into the existing physical topology; when a node goes offline or its file is deleted, the node is directly removed from the physical topology without the need for overall re-identification.

[0024] To address the aforementioned technical problems, the present invention also provides a physical topology identification system for transformer substations, comprising:

[0025] The concentrator, equipped with a CCO, is used to issue box clustering commands and characteristic current transmission commands, receive reported information, and fuse them to generate a physical topology.

[0026] Each branch gateway has a current sampling unit and a digital signal processing unit, which is used to demodulate the characteristic current signal locally and only reports the demodulation result, without uploading the original waveform.

[0027] Meter box switches, each equipped with a dual-mode communication module and a programmable PWM generator unit, are used to inject a characteristic current carrying its own ID into the power line after receiving a command;

[0028] The electricity meter, each equipped with a dual-mode communication module, is used to measure and report dual-mode RSSI values ​​to the CCO;

[0029] The dual-mode communication modules of the aforementioned branch gateways, meter box switches, and electricity meters adopt a unified hardware platform, and the household transformer identification function is implemented through software configuration, without the need for additional hardware.

[0030] Furthermore, the digital signal processing unit of the branch gateway is implemented using DSP or FPGA, and has the ability to complete all calculations such as matched filtering, correlation despreading and peak decision locally, only reporting the identification results and corresponding timestamps to the CCO.

[0031] This invention utilizes a fusion of HPLC and RF dual-mode RSSI for meter box clustering, fully leveraging the complementary characteristics of minimal attenuation of HPLC signals within the meter box and rapid attenuation of RF signals with distance. A two-dimensional fingerprint vector is generated through feature mapping of HPLC and RF signal intensity, which is then processed by a clustering algorithm to determine which meters belong to the same meter box, significantly improving the accuracy of meter box identification. Simultaneously, this invention employs a TDMA+CDMA hybrid coding PWM characteristic current injection scheme, effectively avoiding signal conflicts when multiple nodes inject simultaneously, thus improving the reliability of branch identification. In this scheme, the branch gateway uses an edge computing architecture to demodulate the characteristic current signal locally, only reporting the identification results to the CCO, significantly reducing the computational burden and communication bandwidth usage on the CCO. Through a two-stage identification process of "box-household" clustering and "branch-box" characteristic current injection, combined with a box-by-box polling iteration mechanism, this invention ultimately achieves automatic identification of the four-level physical topology of "transformer-branch-box-household". Attached Figure Description

[0032] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a schematic diagram of the four-level hierarchical architecture of the low-voltage distribution area physical topology automatic identification system according to an embodiment of the present invention.

[0034] Figure 2 This is an overall flowchart of the physical topology identification method for transformer substations as described in an embodiment of the present invention.

[0035] Figure 3 This is a block diagram illustrating the structural composition of the dual-mode communication module described in an embodiment of the present invention.

[0036] Figure 4 This is a schematic diagram illustrating the principle of meter box identification based on dual-mode RSSI clustering as described in an embodiment of the present invention.

[0037] Figure 5 This is a schematic diagram of the PWM characteristic current signal described in an embodiment of the present invention.

[0038] Figure 6 This is a block diagram illustrating the principle of demodulating the PWM characteristic current signal by the branch gateway according to an embodiment of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0040] The core of this application is to provide a method for identifying the physical topology of distribution areas based on dual-mode communication and characteristic current, so as to realize the automatic identification and verification of the four-level physical topology of "transformer-branch gateway-meter box switch-user" in low-voltage distribution networks.

[0041] To enable those skilled in the art to better understand the present application, the structure, features, and effects of the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0042] The present invention discloses a method for identifying the physical topology of transformer substations based on dual-mode communication and characteristic current, which is applied to, for example... Figure 1 In the system shown, the low-voltage distribution area physical topology automatic identification system adopts a four-level hierarchical architecture, which includes, from top to bottom: a concentrator (CCO), several branch gateways, several meter box switches, and several meters (STA).

[0043] The concentrator is installed on the transformer side and serves as the management core of the entire distribution network. The CCO is used to initiate meter box clustering commands and characteristic current injection commands, and to receive the clustering and demodulation results reported by each node, and to generate a four-level physical topology of "transformer-branch-box-household".

[0044] The branch gateway is installed at the beginning of each branch line. Each branch gateway is equipped with a current sampling unit and a characteristic current demodulation unit, which are used to perform distributed demodulation and decoding of the PWM characteristic current signal on the power line, identify the ID information of the meter box switch from which the characteristic current originates, and report the identification result to the CCO.

[0045] The meter box switch is installed at the main switch of each meter box. Each meter box switch is equipped with a dual-mode communication module and a household transformer identification function unit integrated in the dual-mode communication module, which is used to inject a PWM characteristic current signal carrying its own ID information into the power line in response to the CCO command.

[0046] The electricity meters are installed on each user's side, and each meter is equipped with a dual-mode communication module for measuring and reporting the HPLC signal intensity value (RSSI) between itself and its adjacent nodes to the CCO. HPLC and RF signal strength value RSSI RF .

[0047] Furthermore, in this embodiment, the characteristic current demodulation unit of the branch gateway adopts an edge computing architecture, including an analog-to-digital converter, a digital signal processor, or a field-programmable gate array, for performing matched filtering, correlation despreading, and peak decision operations. The branch gateway only reports the successfully demodulated meter box switch ID information and the corresponding timestamp to the CCO; the original current sampling waveform data is not uploaded to reduce the CCO's computing power requirements and communication bandwidth usage.

[0048] Furthermore, in the low-voltage distribution area described in this embodiment, the branch gateway, meter box switch, and electricity meter are equipped with a unified hardware platform, namely a dual-mode communication module. For example... Figure 3 As shown, its structure includes: an HPLC communication unit, an RF communication unit, an MCU main control unit, and a transformer identification function unit integrated within the dual-mode communication module. The transformer identification function unit is equipped with a programmable PWM signal generator, which can generate a current pulse train with constant amplitude and adjustable width by controlling the switching time of the load resistor inside the software control module.

[0049] This invention provides a method for identifying the physical topology of transformer substations based on dual-mode communication and characteristic current. The specific implementation process is as follows: Figure 2 As shown, it includes the following steps:

[0050] S1: The dual-mode communication module completes the power-on and network connection operation. The CCO initiates physical topology identification and measures the HPLC signal intensity RSSI between each node in real time.HPLC and RF signal strength RSSI RF .

[0051] S2: Based on the dual-mode physical signal features measured in step S1, construct a dual-mode physical fingerprint vector. Use unsupervised clustering algorithms such as DBSCAN or MeanShift to perform cluster analysis on the STA nodes. This step effectively identifies the "box-household" mapping relationship between the electricity meter and its associated meter box.

[0052] S3: The CCO sends a characteristic current transmission command to the module in the designated meter box switch. This command includes parameters such as the target meter box switch ID, injection time slot number, PN code sequence, and injection duration. Upon receiving the command, the meter box switch module, according to the command parameters, injects a characteristic current with a specific PN code sequence into the power line during the designated injection time slot using PWM technology. This characteristic current exhibits the following characteristics: Figure 5 The waveform shown is an alternating pattern of positive and negative pulses.

[0053] S4: The branch gateway acquires current signals from the power line in real time and performs high-speed sampling. This stage includes: sequentially performing pre-amplification and DC blocking processing, bandpass filtering, and A / D conversion on the sampled signal, followed by input to the digital signal processing module; using a matched filter to perform sliding correlation calculation between the received signal and the local PN code sequence, and performing peak detection and threshold decision; through PN code despreading and decoding, extracting the meter box switch ID and timestamp carried by the current characteristic current. Thus, the "branch-box" affiliation relationship between the current branch gateway and the corresponding meter box is identified.

[0054] S5: The CCO polls and checks all meter box switch modules to be identified in the distribution area to determine whether all meter box switches have completed the injection and identification of characteristic current. If there are meter boxes that have not been injected, the re-injection and abnormal handling strategy is executed (in this embodiment, a maximum of three retries are set), and then step S5 is repeated to continue polling; if all meter box switches have completed the injection, the process jumps to step S6.

[0055] S6: The CCO integrates the previously identified "box-customer" mapping relationship with the "branch-box" affiliation relationship, automatically constructs a complete four-level physical topology of "transformer-branch-box-customer", and performs consistency verification on the topology model to ensure that the hierarchy from transformer to branch, then to meter box, and finally to user is accurate.

[0056] S7: After the topology generation is completed and the process ends, the system enters a continuous monitoring state to monitor minor changes in the connection, disconnection, or communication location of nodes in the distribution area's power grid. When a local change in the topology is detected, it is not necessary to re-execute the processes from S1 to S6 for the entire distribution area. Instead, incremental identification is performed only for the changed local area, and the topology tree is updated locally to reduce the system's computational load and achieve long-term dynamic maintenance of the distribution area's topology.

[0057] Furthermore, in this embodiment, steps S1 and S2 implement meter box identification based on dual-mode RSSI clustering, the principle of which is as follows: Figure 4 As shown.

[0058] In this embodiment, the CCO first collects dual-mode RSSI data between all meter modules and each meter box switch module in the entire network, where the i-th meter STA i With the j-th meter box switch SW j The HPLC and RF signal intensities are denoted as RSSI, respectively. HPLC (i,j) and RSSI RF (i,j); Based on this, the RSSI values ​​of each meter relative to all n meter box switches are concatenated in a fixed order to construct the dual-mode physical fingerprint vector F of the meter. i = [RSSI HPLC (i,1), RSSI RF (i,1), RSSI HPLC (i,2),RSSI RF (i,2), …, RSSI HPLC (i,n), RSSI RF [i,n], and the F of all m meters in the network i A dual-mode physical fingerprint vector matrix F is formed by stacking rows. The matrix F is then normalized, and cluster analysis is performed based on the similarity between the dual-mode physical fingerprint vectors of each meter. Unsupervised clustering is conducted using DBSCAN or Mean Shift clustering algorithms to output multiple meter clusters. Meters within each cluster are identified as having similar physical spatial affiliation characteristics due to their similar HPLC and RF signal distribution patterns. Furthermore, representative features of each meter cluster are calculated and matched with the dual-mode communication features corresponding to each meter box switch module. Based on the matching results, the meter box switch module corresponding to that meter cluster is determined, thereby establishing the affiliation relationship between the meter and the meter box and achieving accurate identification of the meter and meter box affiliation relationship.

[0059] The specific implementation method of the table box clustering will be described in further detail below.

[0060] First, perform distance mapping for RSSI features.

[0061] Each meter module will measure the dual-mode physical fingerprint F i The vector is reported to the CCO, which performs distance mapping on the HPLC signal intensity characteristics and RF signal intensity characteristics of each meter relative to all meter box switches to obtain HPLC distance characteristics and RF distance characteristics that can characterize the degree of communication attenuation between the meter and each meter box switch.

[0062] In this embodiment, the RSSI feature mapping of the HPLC channel is performed using the following formula:

[0063] d HPLC (i,j) = α HPLC / (RSSI HPLC (i,j) + β HPLC )

[0064] Where, d HPLC (i,j) represents the HPLC distance measurement between meter i and meter box switch j, α HPLC β is the proportionality constant. HPLC As the compensation constant, α HPLC and β HPLC This is a preset empirical constant. Its value is determined as follows: several reference points are selected at the deployment site, and least-squares fitting is performed using measured RSSI values ​​of known distances to minimize the mean square error between the mapped distance and the actual physical distance. For typical low-voltage distribution area environments, α is recommended. HPLC The value range of β is 80 to 120. HPLC The value range is 10 to 30.

[0065] The RF channel RSSI feature mapping uses the following formula:

[0066] d RF (i,j) = α RF × exp(-γ × RSSI RF (i,j))

[0067] Where, d RF (i,j) represents the RF distance measurement between meter i and meter box switch j, α RF α is the amplitude scaling factor, and γ is the attenuation factor; both are preset empirical constants. RF The value of α determines the initial magnitude of the distance metric, while the value of γ controls the sensitivity of signal intensity changes to the distance metric. Both are obtained through offline calibration experiments, i.e., collecting RSSI data at different distances in free space or typical transformer area environments and fitting empirical curves. α is recommended. RF The value of is in the range of 80 to 120, and the value of γ is in the range of 0.15 to 0.35.

[0068] Through the above mapping, the HPLC distance feature matrix D is obtained respectively. HPLC and RF distance feature matrix D RF The data is then spliced ​​together according to the same meter box switch sequence as the dual-mode physical fingerprint vector Fi to form the dual-mode distance feature vector D corresponding to each meter. i = [d HPLC (i,1), d RF (i,1), d HPLC (i,2), d RF (i,2), …, d HPLC [(i,n), dRF(i,n)], which in turn forms the dual-mode distance feature matrix D corresponding to all the meters in the network.

[0069] Next, we will implement the bimodal clustering algorithm.

[0070] CCO uses the dual-mode distance feature vector corresponding to each meter module as clustering samples, and performs unsupervised clustering on the dual-mode distance feature matrix D using a preset clustering algorithm. The clustering algorithm is either a density-based clustering algorithm or a mean-shifting algorithm.

[0071] In this embodiment, the DBSCAN algorithm is used for bin clustering. The specific steps are as follows:

[0072] S21: Set the neighborhood radius parameter ε and the minimum sample size parameter MinPts. The neighborhood radius parameter ε is determined based on the dual-mode distance characteristic distribution between meters in the same meter box and the dual-mode distance characteristic distribution between meters in different meter boxes, through on-site calibration data or historical statistical data, so that meters with similar communication characteristics in the same meter box have a high probability of falling into the same ε-neighborhood; the minimum sample size parameter MinPts is determined based on the statistical characteristics of the number of meters corresponding to the meter boxes in the transformer area and the preset minimum clustering size.

[0073] S22: For each meter data point, calculate the number of meters in its ε-neighborhood. If it is greater than or equal to MinPts, mark the point as a core point. For data points that do not meet the core point condition but are located in the ε-neighborhood of a core point, treat them as boundary points. For data points that do not belong to either the core point or the ε-neighborhood of any core point, mark them as noise points.

[0074] S23: Starting from any unvisited core point, expand the meter data points in its neighborhood through density accessibility and density reachability, and group meter data points with similar dual-mode distance characteristics into the same cluster to form a meter cluster.

[0075] S24: Repeat step S23 until all core points have been visited, and calculate the corresponding representative features for each meter cluster formed; match the representative features with the dual-mode communication features corresponding to each meter box switch module, and determine the meter box switch module corresponding to each meter cluster based on the matching result with the smallest feature distance or the highest similarity, thereby establishing a "box-household" mapping relationship.

[0076] The dual-mode physical fingerprint vector Fi or dual-mode distance feature vector Di needs to be normalized before clustering, and Mahalanobis distance or Euclidean distance is used as a similarity metric. By jointly characterizing the HPLC signal attenuation characteristics and RF field strength characteristics, meter modules with similar dual-mode communication characteristics are clustered into the same cluster. Meters in the same cluster are considered to belong to the same or similar physical meter box. Furthermore, the specific meter box is determined by matching the representative features of the meter cluster with the corresponding features of each meter box switch module, thereby realizing the identification of the meter and meter box affiliation.

[0077] To further improve the accuracy of clustering, the input data for the clustering algorithm also includes the voltage effective value fluctuation sequence and phase angle offset sequence of each meter within a preset time period. The clustering algorithm performs clustering based on a multi-dimensional joint feature vector composed of a dual-mode physical fingerprint vector and a voltage fluctuation sequence. The comprehensive similarity is calculated by weighted fusion of each dimension of the multi-dimensional joint feature vector, wherein the weight coefficient of the dual-mode physical fingerprint vector is greater than the weight coefficient of the voltage fluctuation sequence.

[0078] Furthermore, in this embodiment, steps S3 and S4 implement branch identification based on PWM characteristic current injection, which will be referred to below. Figure 5 Figure 6 This will be explained in detail from two aspects: characteristic current signal generation and distributed demodulation by branch gateways.

[0079] First, the characteristic current signal is generated.

[0080] Figure 5 To generate a characteristic current waveform, the CCO sends a characteristic current transmission command to the module in the designated meter box switch. This command includes parameters such as the target meter box switch ID, injection time slot number, PN code sequence, and injection duration.

[0081] After receiving the instruction, the meter box switch module controls the user transformer identification function unit built into its dual-mode communication module to inject a PWM characteristic current signal carrying the unique identification information ID box of the meter box switch into the power line within a preset time slot window.

[0082] In this embodiment, the PWM characteristic current signal is encoded using a hybrid encoding method of time division multiple access and code division multiple access. Each meter box switch is allocated independent time slot resources, and the pulse width modulation sequence of the PWM characteristic current is embedded with the pseudo-random PN code or GOLD code sequence of the meter box switch. The PN code sequences of different meter box switches are orthogonal to each other.

[0083] In this embodiment, the PWM characteristic current signal is generated by converting the ID information of the meter box switch into a binary feature identification code, generating a GFSK signal through Gaussian frequency shift keying modulation, and then processing it through pulse width modulation to output a characteristic modulated current with good spectral and transmission characteristics.

[0084] In this embodiment, the time slot window for injecting the PWM characteristic current by the meter box switch module is synchronized with the zero-crossing point of the power line voltage. Specifically, the meter box switch detects the zero-crossing signal of the power line voltage through the zero-crossing detection module and initiates the injection of the PWM characteristic current within a ±1ms window before and after the zero-crossing point. Injecting the characteristic current at the zero-crossing point effectively utilizes the characteristic that the line current is close to zero at this time, effectively reducing the interference of the power frequency load current on the characteristic current signal.

[0085] Next, complete the distributed demodulation of the branch gateway.

[0086] Each branch gateway maintains a current sampling wake-up state and continuously samples the phase line or neutral line current signal through its built-in current transformer at a preset sampling frequency. In this embodiment, the preset sampling frequency is 10kHz~100kHz.

[0087] Figure 6 This is a block diagram illustrating the principle of the branch gateway demodulating the PWM characteristic current signal in this embodiment. Figure 6 As shown, the demodulation algorithm for the PWM characteristic current signal by the branch gateway includes the following steps:

[0088] S41: Perform pre-amplification and DC blocking on the sampled signal to block strong current and amplify weak current.

[0089] S42: Bandpass filtering is then performed to remove power frequency and harmonic components, retaining the frequency band of the characteristic current signal. In this embodiment, the frequency band of the characteristic current signal is 500Hz~5kHz.

[0090] S43: Amplify the filtered signal to match the signal amplitude to the input range of the analog-to-digital converter, and then perform digital signal processing.

[0091] S44: A sliding correlation operation is performed on the sampled signal and the preset PN code using a local matched filter and a sliding correlator. Specifically, let the local preset PN code sequence be C(n) and the sampled signal sequence be S(n), then the correlation function R(τ) is: R(τ) = Σ n S(n) × C(n-τ).

[0092] S45: When the relevant peak value exceeds the preset adaptive threshold and meets the peak-to-average power ratio (PAPR) determination condition, it is determined to be a valid characteristic current signal, and the corresponding meter box switch ID information is extracted. The adaptive threshold value is dynamically adjusted according to the current background noise level of the line. In this embodiment, when the relevant peak value exceeds 4 times the current background noise level, it is determined to be a valid signal and the meter box ID is decoded.

[0093] As a preferred implementation, the S3 characteristic current injection stage also includes a dual-mode retransmission mechanism: if a certain meter box switch module does not respond to the characteristic current transmission command within a preset time window, the CCO determines that the meter box switch module is in an abnormal HPLC communication state, executes the dual-mode retransmission mechanism, switches to the RF wireless channel to retransmit the characteristic current transmission command, and effectively improves the topology identification success rate under single-channel abnormal conditions.

[0094] In a preferred embodiment, before the CCO sends the PWM characteristic current transmission command to the meter box switch module in step S3 (characteristic current injection stage), a load status sensing step is also included. The CCO obtains the real-time load current value associated with the meter box switch through the dual-mode communication module. If the real-time load current value is greater than a preset load threshold, the characteristic current injection of the meter box switch module is postponed until the load current value falls below the preset load threshold. In this embodiment, the preset load threshold is set to 80% of the rated current. This step can avoid the characteristic current injection from affecting the normal power supply to users and achieve coordinated scheduling of power supply reliability and topology identification.

[0095] As a preferred implementation, this embodiment also includes a conflict avoidance and arbitration mechanism.

[0096] In this embodiment, different meter box switches under the same branch gateway are assigned different injection time windows T. slot A protection interval T is set between adjacent time slots. guard (In this embodiment, it is 2ms~5ms); the meter box switches under different branch gateways are frequency domain isolated using PWM carriers with different center frequencies.

[0097] In this embodiment, when the PWM characteristic current emitted by a meter box switch module is successfully demodulated by two or more branch gateways simultaneously, an arbitration mechanism is triggered: the CCO receives the demodulation results and corresponding signal arrival times T reported by each branch gateway. arrivaland signal strength RSSI 电流 The branch gateway with the strongest demodulated signal strength is determined to be the branch to which the meter box switch belongs; if the signal strength difference is less than a preset threshold, the characteristic current is re-injected.

[0098] Furthermore, steps S5 and S6 of the method described in this embodiment implement box-by-box polling and topology generation. The specific polling strategy and topology tree generation will be explained in detail below.

[0099] In this embodiment, a dynamic priority polling strategy is adopted. Specifically, the CCO sequentially polls and sends characteristic current transmission commands to all meter box switch modules. The CCO sorts the meter box switches that have not completed branch identification according to the number of hops to the CCO from nearest to farthest, and prioritizes sending injection commands to meter box switches with fewer hops. For meter box switches that fail to identify branches, the CCO increases their priority and enters the next round to resend, until all meter box switches complete branch identification. The maximum number of retry rounds is 3.

[0100] A complete polling cycle should include the following steps:

[0101] S51: CCO sends a command to the first meter box switch module to start timer T. wait .

[0102] S52: The first meter box switch module sends the PWM characteristic current.

[0103] S53: Each branch gateway performs demodulation detection within a preset time window.

[0104] S54: After demodulating to the characteristic current, the branch gateway will report the results (box ID, signal strength) to the CCO.

[0105] S55: CCO records the branch to which the first meter box switch belongs.

[0106] S56: Repeat steps S51 to S55 to process the 2nd to Nth meter box switches in sequence.

[0107] In this embodiment, the CCO (Concentrator Control Operator) integrates the "box-household" mapping relationship obtained in the meter box clustering stage with the "branch-box" affiliation relationship obtained in the branch identification stage to generate a complete physical topology relationship containing four levels: "transformer-branch-box-household". The method for generating the topology tree is as follows: with the concentrator as the root node, each branch gateway as the first-level child node, each meter box switch as the second-level child node, and each meter as the third-level child node, the topology tree is constructed according to the parent-child relationship of "concentrator-branch gateway-meter box switch-meter".

[0108] In this embodiment, after the topology tree is generated, a topology consistency verification step is also included: the CCO sends a query command to each meter module according to the generated "transformer-branch-box-household" topology structure. Each meter module measures the dual-mode communication quality from itself to the switch of the meter box to which its cluster belongs. If the measurement result does not match the topology structure, the meter module is re-clustered or the characteristic current is resent to the switch module of the meter box.

[0109] Furthermore, in step S7 of this embodiment, a dynamic topology update step is included, specifically: the CCO monitors the addition or deletion of nodes in real time. When a new meter or meter box switch module is detected to be joining the network, only the new node undergoes dual-mode RSSI acquisition, clustering, identification, and characteristic current transmission, and the local identification results are incorporated into the existing physical topology. When a node is detected to be offline, the offline node is removed from the physical topology. This incremental update mechanism avoids performing overall identification every time the topology changes, significantly reducing system overhead.

[0110] To further verify the effectiveness of the method, a typical low-voltage distribution area was selected for testing. This area includes one concentrator, six branch gateways, 19 meter box switches, and 260 meters. The branch gateways, meter box switches, and meters are all equipped with a unified hardware platform integrating a dual-mode communication module and a transformer identification function unit. In this test, the CCO implemented four-level physical topology identification according to the following steps.

[0111] Step 1: After each meter and meter box switch is powered on and connected to the network, the CCO initiates physical topology identification and measures the RSSI with adjacent nodes through the dual-mode communication module. HPLC and RSSI RF The value is calculated and the measurement results are reported to the CCO. The CCO collects the dual-mode physical fingerprint vectors of all nodes.

[0112] Step 2: The CCO uses the DBSCAN clustering algorithm, setting ε=3dB and MinPts=3, and uses 19 meter box switches as candidate cluster centers to perform meter box clustering on 260 meters. The clustering results divide the 260 meters into 19 clusters, with each cluster corresponding to one meter box, achieving accurate identification of the "box-household" relationship.

[0113] Step 3: The CCO sends PWM characteristic current sending commands to the 19 meter box switch modules in sequence according to the dynamic priority strategy; each meter box switch module receives the command and injects a PWM characteristic current signal carrying its own ID into the power line during the voltage zero-crossing window.

[0114] Step 4: The 6 branch gateways maintain current sampling (sampling rate 50kHz), demodulate the sampled signal through matched filtering and sliding correlator, and each branch gateway successfully demodulates the ID information of its subordinate meter box switch and reports the result to CCO.

[0115] Step 5: The CCO (Container-Household) clustering results are correlated and fused with the "Support-Container" identification results to generate a complete four-level physical topology of "Variable-Support-Container-Household". After on-site manual verification, the topology identification accuracy reached over 99%.

[0116] As described above, the physical topology identification method for transformer substations based on dual-mode communication and characteristic current provided in this application mainly achieves automatic identification of the four-level physical topology of "transformer-branch-box-household" from the following aspects, thereby improving the accuracy and reliability of topology identification: First, a two-dimensional fingerprint vector is generated by mapping the intensity features of HPLC and RF signals, and then the "box-household" relationship is identified through a clustering algorithm. This fully utilizes the complementary characteristics of minimal attenuation of HPLC signals within the meter box and rapid attenuation of RF signals with distance, significantly improving the accuracy of meter box identification. Second, by adopting a PWM characteristic current injection scheme with time-division and code-division hybrid coding, the signal conflict problem when multiple nodes inject simultaneously is effectively avoided, improving the reliability of branch identification. Third, an edge computing architecture is used to demodulate the characteristic current signal locally, and only the identification results are reported to the CCO, greatly reducing the computing power pressure and communication bandwidth occupation of the CCO. Fourth, by associating and fusing the "box-household" clustering results with the "branch-box" branch identification results, the complete automatic identification of the four-level physical topology of "transformer-branch-box-household" is finally achieved.

[0117] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any person skilled in the art can make some modifications or alterations to the above-described technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of patent protection of the present invention.

Claims

1. A method for identifying the physical topology of transformer substations based on dual-mode communication and characteristic current, characterized in that, Includes the following steps: Step 1: After each meter and meter box switch module is connected to the network, the CCO enables physical topology identification. The meter and meter box switch modules measure the signal strength value (RSSI) with surrounding nodes through two channels: HPLC and RF. HPLC and RSSI RF The two values ​​are then combined to form a two-dimensional fingerprint vector and reported to the CCO. Step 2: After collecting the fingerprint vectors of all nodes, the CCO performs distance mapping on the HPLC signal intensity features and RF signal intensity features of each meter relative to all meter box switches, respectively, to obtain HPLC distance features and RF distance features that characterize the degree of communication attenuation between the meter and each meter box switch. The CCO uses the dual-mode distance feature vector corresponding to each meter module as clustering samples, with the meter box switch as the cluster center, and uses the DBSCAN algorithm or mean drift algorithm to cluster the meters. After clustering, meters in the same cluster are identified as belonging to the same meter box, thus obtaining the correspondence between the meter and the meter box. Among these, DBSCAN... The AN algorithm includes: setting the neighborhood radius parameter ε and the minimum number of samples parameter MinPts; for each meter data point, calculating the number of meters in its ε-neighborhood, and if it is greater than or equal to MinPts, then marking the point as a core point; for data points that do not meet the core point condition but are located in the ε-neighborhood of a core point, treating them as boundary points; for data points that do not belong to either a core point or any core point's ε-neighborhood, marking them as noise points; starting from any unvisited core point, grouping meter data points with similar dual-mode distance characteristics into the same cluster to form a meter cluster, repeating this process until all core points have been visited; Step 3: The CCO sends characteristic current commands to each module in the meter box switch. The selected meter box switch module injects a series of characteristic currents in the form of PWM pulses into the power line by controlling the switching of the internal load resistor within its allocated time slot. The unique ID of the meter box switch is encoded in the current pulse train. Step 4: After the module of a certain meter box switch has sent the characteristic current, each branch gateway continues to sample the current on the power line. When the above PWM characteristic current is detected, the meter box ID carried in it is demodulated by matched filtering and related despreading. The branch gateway that can demodulate the ID is the branch to which the meter box switch belongs. Step 5: The CCO executes Step 3 and Step 4 sequentially on all modules in the meter box switches, and merges the box-household relationship obtained in Step 2 with the branch-box relationship obtained in Step 4 to obtain a four-level physical topology from transformer to branch, from branch to meter box, and from meter box to meter.

2. The method for identifying the physical topology of a transformer substation according to claim 1, characterized in that, In step two, the fingerprint vector used for clustering is first normalized in practice, and then the similarity is measured by Mahalanobis distance or Euclidean distance between the meters. The two key parameters of the clustering algorithm, neighborhood radius and minimum number of samples, are set according to the empirical value of the average spacing between meters in the field.

3. The method for identifying the physical topology of a transformer substation according to claim 1, characterized in that, In step three, when the module in the meter box switch generates the PWM characteristic current, a time-division and code-division hybrid approach is used to avoid conflicts. That is, each meter box switch module is allocated an independent injection time slot, and each PWM pulse sequence is embedded with mutually orthogonal PN codes or GOLD codes. Different meter boxes use different code sequences.

4. The method for identifying the physical topology of a transformer substation according to claim 1, characterized in that, Before issuing the characteristic current transmission command in step three, the CCO first checks the current load of the switch in the meter box. If the load exceeds 80% of the rated value, the injection is postponed until the load drops, so as to avoid the characteristic current causing a perceptible impact on normal power consumption.

5. The method for identifying the physical topology of a transformer substation according to claim 1, characterized in that, In step three, if a module of a certain switch box does not respond to the characteristic current transmission command, the CCO determines that the HPLC channel communication is abnormal and automatically switches to the RF channel to resend the command, thereby improving the recognition success rate under poor channel conditions by utilizing the dual-mode characteristics.

6. The method for identifying the physical topology of a transformer substation according to claim 1, characterized in that, The specific steps for demodulating the characteristic current in step four of the branch gateway are as follows: First, the current waveform is acquired by the current transformer (CT) at a sampling rate of not less than 50kHz. After bandpass filtering and amplification, the PN code stored locally is used to perform a sliding correlation operation with the sampled signal. When the correlation peak value exceeds the preset adaptive threshold value and meets the peak-to-average power ratio (PAPR) determination condition, it is determined to be a valid signal and the meter box ID is decoded. The adaptive threshold value is dynamically adjusted according to the background noise level of the current line.

7. The method for identifying the physical topology of a transformer substation according to claim 1, characterized in that, In step four, if the PWM signal emitted by the module in a meter box switch is simultaneously demodulated and identified by two or more branch gateways, the CCO will compare the signal strength values ​​reported by different branch gateways and consider the gateway with the strongest signal as the subordinate branch of the meter box. If the difference between the strongest and second strongest signals is less than a preset threshold, the module in the meter box switch will be asked to resend the characteristic current to eliminate ambiguity.

8. The method for identifying the physical topology of a transformer substation according to claim 1, characterized in that, Step five also includes a verification process: the concentrator sends a query to each meter based on the generated topology tree, allowing the meter module to measure the dual-mode communication quality from itself to its corresponding meter box switch. If the measured value deviates from the actual value in the topology tree by more than a certain range, the CCO will resend the clustering or characteristic current injection operation to that meter or meter box switch.

9. The method for identifying the physical topology of a transformer substation according to claim 1, characterized in that, After the physical topology is generated in step five, it enters dynamic maintenance mode: when a new node is added to the network, only the local identification of steps one to four is performed on the new node, and the result is merged into the existing physical topology; when a node is offline or its file is deleted, the node is directly removed from the physical topology without the need for overall re-identification.

10. A transformer area physical topology identification system, applicable to the transformer area physical topology identification method as described in any one of claims 1 to 9, characterized in that, include: The concentrator, equipped with a CCO, is used to issue box clustering commands and characteristic current transmission commands, receive reported information, and fuse them to generate a physical topology. Each branch gateway has a current sampling unit and a digital signal processing unit, which is used to demodulate the characteristic current signal locally. It has the ability to complete all calculations such as matched filtering, correlation despreading and peak decision locally, and only reports the demodulation results without uploading the original waveform. Meter box switches, each equipped with a dual-mode communication module and a programmable PWM generator unit, are used to inject a characteristic current carrying its own ID into the power line after receiving a command; The electricity meter, each equipped with a dual-mode communication module, is used to measure and report dual-mode RSSI values ​​to the CCO; The dual-mode communication modules of the aforementioned branch gateways, meter box switches, and electricity meters adopt a unified hardware platform, and the household transformer identification function is implemented through software configuration, without the need for additional hardware.

Images