A transformer area electrical topology self-identification method and system

Abstract

The application discloses a kind of transformer area electrical topology self-identification method and system, it is related to power distribution network automation technical field, method includes: by transformer area intelligent fusion terminal acquisition electrical quantity time series data;Obtain each monitoring point fluctuation time series data;According to fluctuation time series data, time series correlation analysis is carried out between each monitoring point, the fluctuation propagation time delay characteristics between any two monitoring points are extracted by cross-correlation function sub-sampling interpolation, and fluctuation attenuation characteristics are determined according to cross-correlation peak value amplitude;Fluctuation propagation time delay characteristics are monitored between monitoring points Clock bias self-estimation and elimination are carried out, and statistical fusion is carried out, to obtain the fluctuation propagation time delay characteristics after fusion;According to the fluctuation propagation time delay characteristics after fusion and fluctuation attenuation characteristics, determine the electrical topology structure information of transformer area.The system applies the above method.The application extracts propagation time delay characteristics and attenuation characteristics by time series correlation analysis of natural fluctuation, and automatically identifies the complete electrical topology structure of transformer area.

Description

Technical Field

[0001] This application relates to the field of power distribution network automation technology, and in particular to a method and system for self-identification of electrical topology in transformer substations. Background Technology

[0002] With the deepening of the digital transformation of distribution networks, the accurate topology information of distribution transformer areas, as the final power supply units in the power system, is the fundamental data support for line loss analysis, fault location, and power quality management. However, many distribution transformer areas suffer from missing topology maps or inaccurate maps, which seriously hinders the improvement of the level of refined management of distribution networks.

[0003] The existing transformer area topology identification mainly adopts the following technical solutions: manual on-site inspection and verification method, signal injection method, power outage-power restoration step-by-step confirmation method, and voltage correlation clustering method.

[0004] Among the relevant transformer substation topology identification technologies, manual surveying is costly, inefficient, and unable to provide continuous dynamic tracking; signal injection requires specialized equipment for on-site operation and cannot be remotely automated; power outage confirmation affects power supply reliability; voltage correlation clustering only utilizes the static similarity of voltage amplitudes, which can only determine the affiliation of transformers within a household and cannot accurately determine the specific connection relationships, branch node locations, and electrical distances between lines within the substation area. None of the above solutions utilize the dynamic time-domain characteristics of electrical quantity fluctuations propagating and responding in the network to infer the topology. Summary of the Invention

[0005] In view of the above problems, this application provides a method and system for self-identification of electrical topology in a transformer substation. It can automatically identify the complete electrical topology of the transformer substation by utilizing only the existing voltage and current acquisition capabilities of the transformer substation fusion terminal and extracting propagation delay and attenuation characteristics through time-series correlation analysis of natural fluctuations.

[0006] In some embodiments, the voltage and / or current of each monitoring point within the transformer substation are synchronously collected by the intelligent fusion terminal of the transformer substation to obtain electrical quantity time-series data; natural fluctuation events generated by the normal operation of the transformer substation are detected from the time-series data, and the signal quality of each detected natural fluctuation event is evaluated to screen high-quality fluctuation events with effective bandwidth exceeding a preset bandwidth threshold; time-series correlation analysis is performed on the fluctuation data between each monitoring point, and fluctuation propagation delay features are extracted by subsampling interpolation of cross-correlation function, and attenuation features are determined based on the peak amplitude of cross-correlation; the fluctuation propagation delay features extracted from multiple high-quality fluctuation events are self-estimated and eliminated by clock deviation between monitoring points, and multiple sets of propagation delay values ​​after eliminating clock deviation are statistically fused to obtain fused fluctuation propagation delay features; using propagation feature-topology association mapping information, electrical distance is determined based on fused fluctuation propagation delay features, branch connection relationships are determined based on attenuation features, and the electrical topology of the transformer substation is reconstructed using a topology inference algorithm.

[0007] In this embodiment, topology identification is achieved entirely based on passive observation by using the natural fluctuations generated during the normal operation of the transformer area as a topology detection signal source. No special signal injection or power outage operation is required. Moreover, the topology information contained in the propagation delay and attenuation features used far exceeds the voltage static similarity.

[0008] In some embodiments, temporal correlation analysis is achieved by calculating a cross-correlation function, determining the time offset corresponding to the peak value of the cross-correlation function as the propagation delay feature, determining the attenuation feature based on the peak amplitude, and using the peak significance threshold to determine the effective propagation correlation path.

[0009] In this embodiment, the cross-correlation function combined with sub-sampling period interpolation technology can extract time offsets from fluctuating signals with an accuracy far superior to that of the sampling interval; by using event groups with fluctuating sources located on both sides of the monitoring point pair to self-estimate and eliminate clock deviations between monitoring points, the time delay measurement accuracy is not limited by coarse synchronization accuracy; and the significance threshold effectively excludes monitoring point pairs without direct connection.

[0010] In some embodiments, the propagation delay features are converted into electrical distance using a response delay-electrical distance mapping model (i.e., a power distribution line signal propagation speed model, which includes the correspondence between line type parameters and signal propagation speed), and the propagation speed parameters of the model are self-calibrated using reference monitoring points with known electrical distances.

[0011] In this embodiment, the self-calibration mechanism enables the system to adapt to different line types and operating conditions, eliminating the impact of initial model bias on distance estimation accuracy.

[0012] In some embodiments, propagation characteristics are continuously monitored, and topology re-identification is triggered when the characteristic deviation exceeds the topology change detection threshold; data consistency is verified by statistical fusion of multiple fluctuation events and triangular inequality constraints.

[0013] In this embodiment, continuous monitoring enables dynamic tracking and updating of the topology, statistical fusion improves the effective delay resolution accuracy to one-tenth of the single sampling interval through the accumulation of a large number of independent events, and triangular inequality constraints eliminate abnormal data.

[0014] In some embodiments, natural fluctuation events are detected by the rate of change of voltage / current exceeding a preset fluctuation detection threshold, and topology inference is performed when the number of accumulated events reaches a sufficiency threshold.

[0015] In this embodiment, the data sufficiency judgment mechanism ensures that the subsequent statistical fusion has a sufficient sample size to achieve the required time delay resolution accuracy.

[0016] According to a second aspect of this application, a transformer substation electrical topology self-identification system is provided, including a data acquisition module, a fluctuation detection module, a time-series correlation analysis module, and a topology inference module.

[0017] According to a third aspect of this application, an electronic device is provided, including a memory and a processor.

[0018] According to a fourth aspect of this application, a smart fusion terminal for transformer substations is provided, comprising an electrical quantity acquisition unit and the aforementioned self-identification system.

[0019] According to a fifth aspect of this application, a computer-readable storage medium is provided.

[0020] In summary, the method and system for self-identification of electrical topology in transformer substations provided in this application have at least the following beneficial effects: 1. Passive Sensing Achieves Zero-Interference Topology Identification: By utilizing the naturally occurring voltage and current fluctuations during normal operation of the transformer substation as a topology detection signal source, automatic topology identification is achieved without the need for dedicated signal injection or power outages. Compared to existing technologies where signal injection methods require additional equipment and power outage confirmation methods affect reliability, this invention completes topology identification without affecting normal power supply. The inventive aspect of this invention lies in recognizing that topology information is implicit in the temporal correlation patterns of electrical quantities at each monitoring point, and designing a nontrivial extraction method based on cross-correlation analysis and subsampling interpolation, exceeding the conventional design choices of those skilled in the art.

[0021] 2. Accurate topology reconstruction based on dynamic time-domain features: By extracting and jointly analyzing the characteristics of wave propagation delay and attenuation, the branch connection relationship and electrical distance are determined by utilizing the time-domain differences in the propagation and response of transient disturbances in the network. Compared with the existing voltage-related clustering method, which can only determine the affiliation of transformers, this invention realizes the automatic reconstruction of the complete topology structure, and significantly improves the accuracy and information dimension of topology identification.

[0022] 3. Reduced implementation costs by eliminating the need for dedicated additional hardware: The voltage and current acquisition capabilities relied upon are inherent functions of the integrated terminal for the distribution area. Through cross-correlation subsampling interpolation, self-estimation and elimination of clock deviation between monitoring points, and multi-event statistical accumulation, it achieves effective time delay resolution accuracy far superior to the sampling interval and is not limited by coarse synchronization accuracy. There is no need to upgrade to MHz-level high-speed acquisition hardware, nor is it necessary to configure GPS or IEEE1588-level high-precision timing devices for each monitoring point, resulting in extremely low deployment marginal costs.

[0023] 4. Continuous dynamic tracking and adaptive updates: The topology is dynamically tracked by continuously monitoring changes in propagation characteristics. Multi-event statistical fusion and triangular inequality constraints ensure the reliability of identification. Compared with the poor timeliness of manual verification, it can automatically respond to dynamic changes in the transformer area topology. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the main process of the self-identification method for electrical topology of transformer substations provided in the embodiments of this application.

[0025] Figure 2 This is a schematic diagram of the fluctuation event detection subprocess provided in an embodiment of this application.

[0026] Figure 3 A schematic diagram illustrating the principle of temporal correlation analysis and subsampling interpolation provided in the embodiments of this application.

[0027] Figure 4 This is a schematic diagram illustrating the principle of the peak value of the cross-correlation function and the topological correlation provided in the embodiments of this application.

[0028] Figure 5 This is a schematic diagram of the sub-processes for topology reconstruction and dynamic updating provided in the embodiments of this application.

[0029] Figure 6 A schematic diagram of the functional modules of the transformer area electrical topology self-identification device provided in the embodiments of this application. Detailed Implementation

[0030] The technical solution of this application will now be clearly and completely described in conjunction with the accompanying drawings and embodiments.

[0031] The terms "first," "second," etc., are used for descriptive purposes only and should not be interpreted as indicating relative importance. "And / or" indicates that three relationships can exist.

[0032] In this application, "monitoring point" refers to a measurement node within a transformer substation that has the capability to sample voltage and / or current; "natural fluctuation event" refers to voltage and current disturbances not caused by human intervention caused by the normal operation of the transformer substation; and "electrical distance" refers to the equivalent line length along the electrical path between two monitoring points.

[0033] "Fluctuation propagation delay characteristic" is a core technical term in this application, defined as: the time offset extracted from the peak position of the cross-correlation function by performing cross-correlation analysis on the fluctuation time-series data of two monitoring points. This time offset reflects the combined response delay effect caused by the line distributed parameters (resistance, inductance, capacitance) during the process of the fluctuation disturbance traveling from one monitoring point to another along the network path, including the traveling wave propagation delay of high-frequency components and the response time difference caused by the mid- and low-frequency components passing through the line impedance. The magnitude of this characteristic quantity has a monotonic mapping relationship with the electrical distance between the two monitoring points.

[0034] The distribution transformer area is the basic management unit of the low-voltage distribution network. The intelligent integrated terminal of the distribution transformer area is deployed at the distribution transformer, integrating data acquisition, communication and computing functions.

[0035] To understand the propagation and response characteristics of electrical quantity fluctuations in low-voltage distribution networks, it is necessary to understand two physical mechanism levels.

[0036] The first level is the traveling wave propagation mechanism: when load switching generates a steep transient pulse (rise time on the order of microseconds, frequency components covering tens of kilohertz and above), this pulse propagates along the line in the form of a traveling wave. The propagation speed is determined by the line's distributed parameters, with typical values ​​ranging from 1.5 to 2.5 × 10⁻⁶. 8 Meters per second. For a branch length of 200 to 500 meters within the transformer area, the traveling wave propagation delay is approximately 1 to 3 microseconds. Direct measurement of this delay requires a sampling rate in the MHz range.

[0037] The second layer involves the impedance response mechanism: when transient disturbances caused by load changes encompass a wide frequency spectrum (from tens of hertz to thousands of hertz), the distributed impedance of the line causes frequency-dependent amplitude attenuation and phase delay for each frequency component. The series inductance of the line delays and attenuates high-frequency components, resulting in discernible time-domain differences in the transient waveforms received by monitoring points at different electrical distances. When performing cross-correlation analysis on these waveforms, the time shift of the cross-correlation function peak reflects the combined response delay between the two points due to impedance differences. Although this response delay differs physically from the pure traveling wave propagation delay, it also exhibits a monotonically increasing mapping relationship with electrical distance, constituting an effective source of information for topology inference. The aforementioned technical principle lies in the fact that cross-correlation analysis combined with sub-sampling period interpolation technology can achieve a delay resolution capability far superior to that of the sampling interval. When the signal bandwidth is B, the signal-to-noise ratio is SNR, and the observation window duration is T, the theoretical lower bound (Cramé-Rao bound) of the cross-correlation delay estimation is approximately:

[0038] σ τ ≈1 / (2π×B×sqrt(2×B×T×SNR)); For example, for a typical high-load switching event (signal bandwidth B approximately 5kHz, observation window T=5ms, signal-to-noise ratio SNR approximately 100, or 20dB), the standard deviation of the delay estimate for a single event is: σ τ ≈1 / (2π×5000×sqrt(2×5000×0.005×100))≈0.45μs; For events with a higher signal-to-noise ratio (e.g., 30dB), or after statistical averaging of N independent events, the random error component is further reduced to σ. τ / sqrt(N).

[0039] It should be noted that the Cramer-Rhodes boundary described above refers to random error components; clock deviations between monitoring points are systematic errors that cannot be eliminated by statistical averaging and must be deducted using the clock deviation self-estimation and elimination mechanism described in the specific implementation method. Under the premise that the clock deviation is eliminated through self-estimation, even with a sampling rate on the order of tens of kilohertz, the system can still achieve sub-microsecond effective time delay resolution accuracy through cross-correlation interpolation and multi-event statistical accumulation, which is sufficient to distinguish branches with different electrical distances within the transformer area.

[0040] Based on the above background information, the method for self-identification of electrical topology in the transformer substations in this application is as follows: like Figure 1 As shown in the figure, this application provides a method for self-identification of electrical topology in a transformer substation, including the following steps.

[0041] Step S101: The voltage and / or current of each monitoring point in the transformer substation are synchronously collected through the intelligent fusion terminal of the transformer substation to obtain the electrical quantity time sequence data of each monitoring point.

[0042] Each monitoring point samples data based on a unified time synchronization mechanism. In some embodiments, the sampling rate is set to 10kHz to 50kHz. Each monitoring point achieves coarse synchronization by receiving periodically broadcast synchronization messages from the smart fusion terminal of the distribution area via the distribution area carrier channel and / or wireless channel. The coarse synchronization accuracy is better than 100 microseconds. Coarse synchronization is only used to ensure that the waveform windows of the same fluctuation event captured by each monitoring point are aligned. The residual clock deviation between monitoring points does not directly determine the time delay measurement accuracy. This residual clock deviation is separated and subtracted by the clock deviation self-estimation and elimination mechanism described in step S103. Monitoring points include the low-voltage side of the transformer, each branch box, and the smart meter on the user side. For example, for a distribution area with 3 main lines and 60 users, 25 key monitoring points are set up to collect data at a sampling rate of 50kHz.

[0043] In some embodiments, a dedicated high-speed sampling module is provided with a sampling rate of 1MHz to 10MHz. Time synchronization adopts GPSPPS or IEEE1588 protocol to achieve sub-microsecond accuracy, which can directly capture the travel wave propagation delay.

[0044] In this embodiment, time-series data from each monitoring point are acquired synchronously to provide a consistent time reference for subsequent cross-correlation analysis, which is the data basis for extracting wave propagation delay characteristics.

[0045] Step S102: Detect natural fluctuation events generated by the normal operation of the transformer substation from the electrical quantity timing data.

[0046] like Figure 2 As shown, continuously collected data is analyzed in real time to detect fluctuation events caused by user start-up and shutdown of high-power loads and changes in distributed power supply output. The detection method is to calculate the voltage change rate.

[0047] dV / dt=(V(t)-V(t-Δt)) / Δt When |dV / dt| exceeds the preset detection threshold ε detect When the value is 2V / ms, a valid fluctuation event is determined, and the time intervals before and after that point are recorded. window Waveform data of each monitoring point within 10ms (e.g.).

[0048] In some embodiments, signal quality assessment is performed on the detected events. The effective bandwidth of the event is calculated:

[0049] BW event =sqrt(∑(f k 2 ×|X(fk )| 2 ) / ∑(|X(f k )| 2 )) Where f k For each frequency component, X(f) k ) represents the amplitude of the waveform spectrum.

[0050] The higher the effective bandwidth, the higher the accuracy of subsequent cross-correlation delay estimation. Select B. Wevent Events exceeding a bandwidth threshold (e.g., 1kHz) are designated as high-quality fluctuation events for subsequent delay extraction and statistical fusion. Within these high-quality fluctuation events, differentiated statistical fusion weights are assigned based on effective bandwidth and signal-to-noise ratio (SNR), with higher bandwidth and higher SNR events receiving greater weights. Typical high-bandwidth events include motor starting, capacitor switching, and inverter grid connection transients.

[0051] In some embodiments, data sufficiency is counted separately for each monitoring point pair: for any pair of monitoring points with a valid correlation, only high-quality events with effective bandwidth exceeding a bandwidth threshold are counted in the number of available events for that monitoring point pair, until the number of available events for each monitoring point pair reaches the data sufficiency threshold N. min Topology inference is only performed when (e.g., 200). Since high-bandwidth events typically account for only 5% to 15% of all detected events, and clock skew self-estimation requires each monitoring point to accumulate a sufficient number (e.g., no less than 30 times on each side) of high-bandwidth events on both sides, and high-power loads are spatially unevenly distributed within the distribution area, the number of high-bandwidth events located on the far side of a specific branch may only be in the single digits per day. Therefore, the calendar time required to achieve full network data sufficiency is significantly longer than the time estimated based on the total number of events.

[0052] N min The determination principle is to ensure that the time delay estimation accuracy after statistical fusion meets the target topology resolution requirements, and to ensure that the location of the fluctuation source covers different branches of the transformer area, so as to support clock deviation self-estimation and attenuation grouping analysis.

[0053] In this embodiment, passive event capture and quality classification are achieved through rate of change threshold and bandwidth screening, and the data sufficiency mechanism ensures that subsequent statistical fusion has sufficient samples to achieve the required subsampling period delay accuracy.

[0054] Step S103: Perform time-series correlation analysis on each monitoring point to determine the characteristics of wave propagation delay and wave attenuation.

[0055] like Figure 3 and Figure 4 As shown, for any two monitoring points i and j, the cross-correlation function is calculated using the fluctuation time series data in the same fluctuation event, and the fine time delay is extracted by subsampling interpolation.

[0056] The cross-correlation function is calculated as follows: R ij (τ)=∑(x i (t)×x j (t+τ)) / N s ;where R ij (τ) is the cross-correlation value at time offset τ, x i (t) and x j (t+τ) represents the fluctuation time series data of the two monitoring points, N s This represents the number of sampling points.

[0057] Since the actual time delay usually does not fall on the discrete sampling grid, sub-sampling periodic parabolic interpolation is used to achieve fine time delay estimation.

[0058] Let τ be the discrete maximum point of the cross-correlation function. max The value at point R and its adjacent points is R. -1 R0, R +1 The fine delay is then: τ ij =τ max +T s ×(R -1 -R +1 ) / (2×(R -1 -2×R0+R +1 )); Where T s The sampling period is denoted by . This interpolation allows the time delay resolution of a single event to exceed the sampling interval limitation. The theoretical accuracy is determined by the signal bandwidth and signal-to-noise ratio (see the Cramer-Rhodes bound formula in the background knowledge), and can reach T. s / 10 to T s / 50.

[0059] For example, at a sampling frequency of 50 kHz (T s Under the condition of 20μs, for a single motor starting event (BW approximately 3kHz, SNR approximately 25dB), the cross-correlation function between monitoring point A on the transformer side and monitoring point B at the end of a branch is calculated, and the fine time delay τ is obtained after interpolation. AB =1.8μs. The standard deviation of a single estimate is approximately 0.8μs.

[0060] Statistical fusion of delay values ​​extracted from N independent events: τ ij,fused =∑(w k ×τ ij,k ) / ∑(w k ) Where w k Let w be the weight of the k-th event. k =1 / σ k2 , σ k The standard deviation of the single-event delay estimate is determined by the effective bandwidth, signal-to-noise ratio, and effective observation duration of the event according to the aforementioned Cramer-Rao bound formula; for transient events where the effective observation duration is comparable to the fluctuation duration (i.e., the bandwidth product is approximately constant), w k Alternatively, it can be approximated by B of the event. Wevent 2 ×S NRk Proportional.

[0061] The effective accuracy after fusion is: σ fused =sqrt(1 / ∑(1 / σ k 2 )); For example, at a certain monitoring point, among 200 accumulated high-quality fluctuation events (all with effective bandwidth exceeding the bandwidth threshold), there are 50 events with high bandwidth and high signal-to-noise ratio (single event σ approximately 0.8 μs) and 150 events with bandwidth close to the threshold or low signal-to-noise ratio (single event σ approximately 2 μs). After weighted fusion, σ fused Approximately 0.1 μs is sufficient to distinguish time delay differences on the order of 1 μs (corresponding to electrical distance differences of approximately 150 to 250 meters).

[0062] In some embodiments, for any pair of monitoring points (i,j), before performing the above statistical fusion, the clock deviation between the monitoring points is first estimated and eliminated. The principle is as follows: the measured time delay τ between monitoring points i and j, obtained through cross-correlation... ij,meas The actual propagation delay τ ij,true Clock deviation δ between the two monitoring points ij It consists of two parts, namely:

[0063] τ ij,meas =τ ij,true +δ ij ; Where, δ ij Determined by the coarse synchronization residual error, it can reach the tens of microseconds level; due to the inherent repeatability of the processing link delay of the synchronization messages received by each monitoring point, and the continuous suppression of the frequency drift accumulation of the local clocks at each monitoring point by the periodically broadcast synchronization messages, δ ij Approximately constant over timescales ranging from minutes to hours, the random jitter of successive receptions of synchronization messages manifests as a superposition of δ. ij The zero-mean random component on the surface is weakened by average in subsequent multi-event statistical fusion.

[0064] And τ ij,true The sign of τ depends on the electrical position of the wave source relative to the monitoring point pair: when the wave source is located outside monitoring point i (the wave arrives at monitoring point i first and then at monitoring point j), τij,true =+|τ ij |;When the fluctuation source is located outside the monitoring point j, τ ij,true =-|τ ij |;When the fluctuation source is located on the branch between the two monitoring points, τ ij,true It lies somewhere in between.

[0065] Based on this, the measured time delay values ​​of the accumulated multiple events are clustered, and the upper and lower value clusters in the measured time delay distribution (corresponding to the event groups where the fluctuation source is located on both sides of the monitoring point) are selected. The weighted average of the two clusters is then calculated to obtain τ. A With τ B Then the clock deviation and the actual propagation delay are respectively: δ ij =(τ A +τ B ) / 2; |τ ij |=(τ A -τ B ) / 2; The criteria for determining which event group an event belongs to are: the order in which each monitoring point detects the initial edge of the fluctuation, and / or the spatial distribution pattern of the fluctuation amplitude at each monitoring point (the monitoring point with the largest fluctuation amplitude is adjacent to the fluctuation source). The measured time delay of each event belonging to the upper value cluster and the lower value cluster is subtracted by δ one by one. ij The absolute values ​​of the propagation delays for each event are then taken to obtain a sample of absolute propagation delays. The aforementioned statistical fusion is then performed on these samples, and the fusion result is used as the propagation delay feature value τ. ij,fused This is used for subsequent electrical distance calculations.

[0066] For events where the fluctuation source is located between two monitoring points, i.e., after deducting δ ij The absolute value of the delay after that is significantly less than (τ) A -τ B The event of ) / 2 only reflects the propagation delay of part of the path and is not included in the statistical fusion of the propagation delay of the monitoring point pair, so as to avoid underestimating the propagation delay of the whole path.

[0067] In some embodiments, δ is also used ij Perform sliding window tracking to compensate for clock drift: Divide the estimation window into fixed durations (e.g., 1 hour), and when two adjacent estimation windows yield δ... ij When the change exceeds the drift threshold (e.g., 0.5 microseconds), update δ with the estimate from the new window. ij And for the event latency data across windows before and after the update, the δ of each window is used respectively. ij Deduct it.

[0068] In this embodiment, clock deviation is explicitly separated and subtracted as a systematic error, so that the time delay measurement accuracy is no longer limited by the coarse synchronization accuracy, but is determined only by the cross-correlation interpolation accuracy and the statistical fusion sample size. This ensures that sub-microsecond effective propagation time delay characteristics can still be obtained under the conditions of conventional sampling rates of 10kHz to 50kHz and coarse synchronization on the order of 100 microseconds.

[0069] Fluctuation decay characteristic α ij Determined by the normalized cross-correlation peak value: α ij =R ij (τ ij ) / (σ i ×σ j ); Where σ i and σ j Let α be the standard deviation of the two-point fluctuation signal. ij It reflects the degree of amplitude attenuation between two points.

[0070] In some embodiments, the normalized cross-correlation peak value α is used. ij The peak significance of the cross-correlation function is characterized by a preset correlation threshold (i.e., significance threshold) ε. sig Determining a valid connection: α ij >ε sig If the value is 0.3, it is determined that there is an effective fluctuation propagation correlation path between the two monitoring points; otherwise, the possibility of a direct branch connection between the two monitoring points is excluded.

[0071] In this embodiment, cross-correlation analysis, combined with subsampling interpolation and multi-event statistical accumulation, achieves an effective delay resolution accuracy far superior to that of the sampling interval. This is the technical means of the present invention to extract topological information without requiring MHz-level high-speed sampling hardware, fully utilizing the statistical advantage of the frequent occurrence of natural fluctuation events in the transformer area.

[0072] Step S104: Determine the electrical topology information of the transformer area based on the wave propagation delay characteristics and wave attenuation characteristics.

[0073] like Figure 5 As shown, the topology of the transformer area is reconstructed based on propagation delay and attenuation characteristics.

[0074] Sub-step S104a: Determine electrical distance information.

[0075] The response delay-electrical distance mapping model (i.e., the signal propagation speed model for power distribution lines, both referring to the same model, which includes the correspondence between line type parameters and signal propagation speed, hereinafter referred to as the mapping model) is used to convert the delay characteristics into electrical distance. The mapping relationship is as follows: the combined delay of wave disturbance propagation and response in the line is proportional to the line length, with a proportionality coefficient (equivalent speed parameter, i.e., propagation speed parameter K)... cal The characteristics are determined by the line distribution parameters and the signal spectrum.

[0076] The mapping formula is: d ij =K cal ×τ ij,fused ; Where d ij For electrical distance, K cal τ is the calibrated time-delay-distance conversion coefficient. ij,fused This represents the propagation delay characteristic value after statistical fusion.

[0077] K cal The physical meaning of is the equivalent propagation / response speed of a signal in a specific line, and its value depends on the line type and the main frequency components of the transient signal.

[0078] For low-voltage overhead insulated lines, the typical range of kcal is 1.2 × 10⁻⁶. 8 Up to 2.2×10 8 m / s, for low-voltage cable lines it is 0.8×10 m / s 8 Up to 1.5×10 8 m / s. The lower bound of this range corresponds to a situation where the effective signal bandwidth is close to the bandwidth threshold (approximately 1 kHz) and the ratio of line unit resistance to inductive reactance R / ωL is relatively large, in which case the impedance response effect contributes significantly to the total time delay; the upper bound corresponds to a situation where the effective signal bandwidth is relatively high (above several kilohertz) and the propagation behavior approaches a traveling wave mechanism. All of the above numerical ranges are significantly lower than the pure traveling wave propagation speed of the corresponding line type (approximately 2.0 to 2.5 × 10⁻⁶ m / s for overhead insulated lines). 8 m / s, cable speed approximately 1.5 to 2.0 × 10 m / s 8 (m / s), reflecting the essence of Kcal as a comprehensive equivalent parameter that includes the contribution of impedance response. The actual value of Kcal should be based on the calibration results; the typical range mentioned above is only for reference in verifying the reasonableness of the calibration results.

[0079] In some embodiments, K is performed using a reference monitoring point with a known electrical distance. cal Self-calibration. For example, obtaining the actual line length d from the transformer to a branch box from the construction records. ref =180m, the system measured the fusion delay τ of this segment. reffused =1.1μs, then:

[0080] Kcal =d ref / τ reffused =180 / (1.1×10 -6 ) = 1.64 × 10 8 m / s By calculating the electrical distance between all valid monitoring point pairs, an electrical distance matrix D(i,j)=d is constructed. ij .

[0081] In some embodiments, for transformer substations where overhead insulated wires and cables are laid in a mixed manner, propagation velocity parameters are set and calibrated separately according to line type: the propagation velocity parameters corresponding to the overhead line are calibrated using known electrical distance reference monitoring points located on the overhead line segment, and the propagation velocity parameters corresponding to the cable line are calibrated using known electrical distance reference monitoring points located on the cable line segment; for monitoring point pairs where the path crosses two line types, the initial electrical distance is first calculated using the weighted equivalent value of the two types of propagation velocity parameters and the initial topology is reconstructed, and then, according to the path line type determined by the initial topology, the electrical distance and topology of each line segment contained in the path are iteratively updated using the propagation velocity parameters of the corresponding line type, until the change in electrical distance of each segment obtained from the two iterations is less than the set convergence limit (e.g., 1m).

[0082] Sub-step S104b: Determine the branch connection relationship information.

[0083] The attenuation characteristics reflect the energy loss and branching of the fluctuation along the path. The determination of branch connection relationships is performed according to the following sub-process.

[0084] 1. Grouping events with the same source: For each fluctuation event, construct an amplitude distribution vector obtained by normalizing the amplitude of each monitoring point after the maximum value. Calculate the cosine similarity between the amplitude distribution vectors of different events. Group events with similarity exceeding the same source determination threshold (e.g., 0.9) into the same event cluster. The same event cluster is regarded as originating from the same fluctuation source point s or fluctuation sources adjacent to the same electrical location.

[0085] 2. Constructing the attenuation vector and performing group analysis: For each event cluster, the monitoring point with the largest fluctuation amplitude in the event cluster (i.e., the monitoring point of the nearest fluctuation source s) is determined as the reference monitoring point of the event cluster; for each event within the cluster, the attenuation characteristic value between the reference monitoring point and each monitoring point m is calculated according to the definition of the normalized cross-correlation peak value described in step S103, and then the attenuation characteristic value of each event within the cluster is weighted and averaged according to the weights determined by the effective bandwidth and signal-to-noise ratio of each event to obtain the attenuation vector [α] corresponding to the fluctuation source. s1 ,α s2 ,…,α sM ], where M is the number of monitoring points, α smThis represents the attenuation characteristic from the fluctuation source s (with the reference monitoring point as the representative point of its electrical position) to the monitoring point m, where the attenuation characteristic corresponding to the reference monitoring point itself is set to 1.

[0086] After taking the logarithm of the attenuation vector, hierarchical clustering is performed. Monitoring points in the same level within the cluster are determined to be downstream of the same branch on the wave propagation path; the attenuation jump positions between different levels correspond to branch nodes.

[0087] The grouping determination is based on the following two criteria: Criterion 1 is the monotonicity of the same branch, that is, for the same fluctuation source s, if monitoring points i and j are located in the same branch and i is closer to the fluctuation source than j, then α should be satisfied. si ≥α sj Any grouping that violates the monotonicity rule will be rejected and the corresponding monitoring point will be re-filed.

[0088] Criterion 2 is cross-branch segmentation, that is, when the fluctuation passes through branch node k and splits into multiple downstream branches, the attenuation characteristics of the monitoring points at the beginning of each downstream branch show obvious segmentation relative to node k. The segmentation phenomenon can be explained by the following diversion approximation model: α sbranchm =α sk ×Z total / (Z total +Z branchm ); Where, α sk Let α be the attenuation characteristic value of the wave propagating from wave source s to branch node k. sbranchm Z represents the attenuation characteristic value of the fluctuation at the monitoring point at the beginning of the m-th downstream branch after it enters the downstream branch. branchm Z represents the input impedance of the m-th downstream branch as seen downstream from branch node k. total This is the parallel equivalent impedance of the input impedances of all downstream branches at branch node k, excluding the branch from which the wave originates. This shunt model is only used to explain the physical cause of the segmentation phenomenon. The actual determination process does not require prior knowledge of the specific values ​​of the impedances of each branch, but directly determines the segmentation structure obtained by hierarchical clustering of attenuation vectors.

[0089] 3. Combine the grouping results corresponding to multiple fluctuation sources at different locations: The attenuation classification of different fluctuation sources divides the monitoring point set from different directions. By taking the intersection operation of the grouping results of each fluctuation source, the belonging of the monitoring points on both sides of each branch node can be uniquely determined, and the branch connection relationship information between each monitoring point (i.e., the candidate parent-child connection relationship set) can be obtained for the topology reconstruction in sub-step S104c.

[0090] Sub-step S104c: Topology reconstruction.

[0091] The constrained minimum spanning tree algorithm is used to reconstruct the topology using electrical distance as the edge weight. The constraints are: the topology is a connected tree structure, the distance between each segment is within a reasonable range (no more than 500m per segment), and the branch structure is consistent with the attenuation grading and grouping results obtained in sub-step S104b.

[0092] Candidate topologies are generated as follows: Using the electrical distance matrix D(i,j) as edge weights, a minimum spanning tree is solved under the above constraints to serve as the initial candidate topology. For several edges in the initial candidate topology with the largest contribution to the fitting residual (e.g., the top three edges in terms of residual contribution), a local edge swap operation is performed one by one, i.e., the edge is disconnected and connected to an edge whose weight difference is less than the edge swap tolerance (e.g., 3 times σ). fused ×K cal The candidate adjacent edges of ) are used to generate several candidate topologies.

[0093] For each candidate topology, calculate the fitting residual: J=∑(w ij ×(τ ij,fused -τ ij,model ) 2 )+λ×∑(w ij ×(α ij -α ij,model ) 2 ) Where, τ ij,model The model time delay from monitoring point i to monitoring point j in the candidate topology is equal to the sum of the electrical distances of all line segments in the tree path from i to j in the candidate topology, divided by the calibrated conversion factor K. cal ;α ij,model The model attenuation under the candidate topology is equal to the segment attenuation factor β of each line segment contained in the path. seg The product of the series, the attenuation factor β of each line segment seg The attenuation ratio of adjacent monitoring point pairs on the same branch in sub-step S104b is estimated by least squares regression. ij To monitor the weights of point pairs (i,j), we take w. ij =1 / σ fused,ij 2 , where σ fusedij Let be the estimated standard deviation of the fusion delay at this monitoring point; λ is a dimensional balance factor used to ensure that the contributions of the delay residual term and the attenuation residual term to J are of the same order of magnitude. In some embodiments, λ is taken as median((τ ij,fused -τ ij,model ) 2 ) / median((α ij -α ij,model ) 2), where median represents the median of all valid monitoring point pairs; wherein the median is based on τ corresponding to the initial candidate topology. ij,model With α ij,model The calculation shows that the determined λ maintains the same value in the comparison of the fitting residuals of all candidate topologies, so as to ensure that the residuals of each candidate topology are comparable.

[0094] The candidate with the smallest J is selected as the final topology. The confidence level of the topology reconstruction result is determined by the bootstrap method: the cumulative event set is resampled with replacement (e.g., 100 times), and statistical fusion and topology reconstruction are repeatedly performed on the event set after each resampling. The proportion of the reconstruction result that is completely consistent with the final topology is used as the confidence level. For example, for a transformer area with 60 households, after the number of available events for each monitoring point pair reaches the data sufficiency threshold (200 times), a topology structure of 3 main lines and 7 branch nodes is reconstructed. In 100 resamplings, the reconstruction results are consistent with the final topology in 85 times, with a confidence level of 0.85.

[0095] In this embodiment, accurate topology identification is achieved through calibration mapping and global optimization reconstruction, self-calibration eliminates the influence of uncertainties in line parameters, and fitting residual optimization ensures the robustness of the results.

[0096] Furthermore, the method also includes step S105, dynamic topology update.

[0097] First, continuously monitor changes in transmission characteristics.

[0098] Calculate the relative deviation: Δ feature =|τ ijcurrent -τ ijstable | / τ ijstable ; Exceeding the change threshold ε change Re-identification is triggered when the value is 0.20 (e.g., 0.20), otherwise it remains unchanged.

[0099] Verifying data consistency using triangle inequality constraints: For any three points i, j, k, τ ij ≤τ ik +τ kj Data is valid if it meets the requirements; if it violates the requirements, its weight is reduced or the data is removed.

[0100] In this embodiment, dynamic updates enable the system to automatically track topology changes, and triangular inequality constraints eliminate erroneous data caused by multipath interference or measurement anomalies.

[0101] like Figure 6 As shown, this application provides a self-identification system for electrical topology of transformer substations, including: The data acquisition module is used to synchronously acquire the voltage and / or current of each monitoring point in the distribution area through the intelligent fusion terminal of the distribution area, and obtain the electrical quantity time sequence data of each monitoring point; The fluctuation detection module is used to detect natural fluctuation events generated by the normal operation of the transformer area from the electrical quantity time series data, evaluate the signal quality of each detected natural fluctuation event, filter high-quality fluctuation events with effective bandwidth exceeding a preset bandwidth threshold, and obtain the fluctuation time series data of each monitoring point corresponding to each high-quality fluctuation event. The time-series correlation analysis module is used to perform time-series correlation analysis between monitoring points based on the fluctuation time-series data of each monitoring point. It extracts the fluctuation propagation delay characteristics between any two monitoring points through subsampling interpolation of the cross-correlation function, and determines the fluctuation attenuation characteristics based on the cross-correlation peak amplitude. It performs self-estimation and elimination of the clock deviation between monitoring points on the fluctuation propagation delay characteristics extracted from multiple high-quality fluctuation events, and statistically fuses the multiple sets of propagation delay values ​​after eliminating the clock deviation to obtain the fused fluctuation propagation delay characteristics. The topology inference module is used to determine the electrical topology information of the transformer substation based on the fused wave propagation delay characteristics and wave attenuation characteristics.

[0102] The electronic device includes a memory, a processor, a communication interface, and a bus. The processor can be an ARM processor, a DSP, or an FPGA, configured to execute the methods described above. The communication interface includes an RS485 and / or carrier communication module for interacting with the smart meter, and a wireless module for uploading topology results to a higher-level system.

[0103] A smart integrated terminal for a transformer substation includes an electrical quantity acquisition unit for synchronously acquiring voltage and / or current at each monitoring point within the substation, and employs the transformer substation electrical topology self-identification system described above. The data acquisition module of the self-identification system is connected to the electrical quantity acquisition unit for automatically identifying the electrical topology of the substation based on the data acquired by the electrical quantity acquisition unit.

[0104] A computer-readable storage medium having a computer program stored thereon for running the above-described self-identification method.

[0105] The following is a specific application scenario using the method and system described in this invention: The intelligent fusion terminal for the distribution transformer area is installed at the distribution transformer. In some embodiments, it is applied to 10kV / 0.4kV distribution transformer areas. After deployment, it automatically accumulates data using existing acquisition capabilities, and completes the first topology identification after approximately 14 to 30 days of event accumulation. The accumulation period depends on the electricity activity of the distribution transformer area, the proportion of high-bandwidth events, and the spatial distribution of fluctuation sources: commercial areas have a high daily event density (500 to 1000 events, of which approximately 25 to 150 high-bandwidth events that can be used for latency statistics account for 5% to 15%, but high-bandwidth events located on the far side of individual branches may only occur a few times per day), and it takes approximately 7 to 14 days for each monitoring point to achieve sufficient data coverage of events on both sides; residential areas have a moderate event density (200 to 400 events per day, with approximately 10 to 60 high-bandwidth events per day), and it takes approximately 14 to 21 days; low-load distribution transformer areas may require 30 days.

[0106] In one verification embodiment, an actual transformer substation with complete as-built drawings was selected as the verification object. This substation included 3 main lines, 7 branch nodes, and 60 users, with a total line length of approximately 1.2 km. Overhead insulated wires and cables were mixed and laid. 25 monitoring points were deployed, with a sampling rate of 50 kHz and a coarse synchronization accuracy better than 100 microseconds. Data was automatically accumulated for 21 days using the aforementioned method. During this period, approximately 4300 valid natural fluctuation events were detected, including 612 high-bandwidth events. The number of usable events for each monitoring point pair reached the data sufficiency threshold N. min (200 times), and each monitoring point has no less than 30 high-bandwidth events on both sides, which meets the event coverage conditions on both sides required by the clock deviation self-estimation.

[0107] The reconstruction results were compared with the as-built drawings: the connection relationships of all 24 branch lines were correctly determined; the average absolute error of the electrical distance from each monitoring point to the low-voltage side monitoring point of the transformer was 8.5m, and the average relative error was 4.7% compared with the corresponding path length recorded in the as-built drawings (the average path length of the 24 monitoring points is about 180m); the topological confidence level of the evaluation by resampling 100 times using the bootstrap method was 0.85.

[0108] In contrast, voltage correlation clustering was used to identify the same transformer substation area. This method could only provide the transformer ownership relationship (with an ownership determination accuracy of 98%), but could not provide branch connection relationships, branch node locations, or electrical distances. The above verification results show that the technical solution of this application achieves automatic reconstruction of the complete branch-level topology without upgrading the high-speed acquisition hardware.

[0109] The embodiments of this application can be freely combined. The computer program product includes a non-transitory computer-readable storage medium storing program code. The sequence numbers of the method steps are for descriptive convenience only.

[0110] The above description is merely a specific embodiment, and all equivalent substitutions based on this technical solution should fall within the scope of protection. For example, replacing parabolic interpolation with Gaussian interpolation or phase correlation analysis, replacing minimum spanning tree with probabilistic graphical model inference, and replacing GPS synchronization with BeiDou timing or IEEE 1588 protocol are all equivalent substitutions that do not depart from the core concept.

Claims

1. A method for self-identification of electrical topology in a transformer substation, characterized in that, include: The voltage and / or current of each monitoring point in the transformer substation are collected synchronously through the intelligent integrated terminal of the transformer substation to obtain the electrical quantity time sequence data of each monitoring point. Detect natural fluctuation events generated by the normal operation of the transformer substation from the electrical quantity time series data, evaluate the signal quality of each detected natural fluctuation event, filter out high-quality fluctuation events with effective bandwidth exceeding a preset bandwidth threshold, and obtain the fluctuation time series data of each monitoring point corresponding to each high-quality fluctuation event. Based on the fluctuation time series data of each monitoring point, a time series correlation analysis is performed between the monitoring points. The fluctuation propagation delay characteristics between any two monitoring points are extracted by subsampling interpolation of the cross-correlation function, and the fluctuation attenuation characteristics are determined based on the cross-correlation peak amplitude. The wave propagation delay features extracted from multiple high-quality wave events are self-estimated and eliminated by clock deviation between monitoring points, and the multiple propagation delay values ​​after eliminating clock deviation are statistically fused to obtain the fused wave propagation delay features. Based on the fused wave propagation delay characteristics and wave attenuation characteristics, the electrical topology information of the transformer substation is determined; The natural fluctuation event refers to voltage and current disturbances caused by changes in user load and / or distributed power output during the normal operation of the transformer substation. The electrical topology information includes the branch connection relationships and electrical distances between each monitoring point. Determining the electrical topology information of the transformer substation based on the fused fluctuation propagation delay characteristics and fluctuation attenuation characteristics includes: using propagation feature-topology association mapping information, determining the electrical distance information between each monitoring point based on the fused fluctuation propagation delay characteristics, determining the branch connection relationship information between each monitoring point based on the fluctuation attenuation characteristics, and reconstructing the electrical topology of the transformer substation using a topology inference algorithm based on the electrical distance information and the branch connection relationship information. The propagation feature-topology association mapping information includes: the mapping relationship between the fluctuation propagation delay characteristics and the electrical distance information, and the mapping relationship between the fluctuation attenuation characteristics and the branch connection relationship information.

2. The method as described in claim 1, characterized in that, The process of performing time-series correlation analysis between monitoring points, extracting the fluctuation propagation delay characteristics between any two monitoring points through subsampling interpolation of the cross-correlation function, and determining the fluctuation attenuation characteristics based on the cross-correlation peak amplitude includes: Calculate the cross-correlation function between the fluctuation time series data of any two monitoring points; The time offset corresponding to the peak value of the cross-correlation function after subsampling interpolation is determined as the fluctuation propagation delay characteristic between the two monitoring points; The fluctuation attenuation characteristics between the two monitoring points are determined based on the ratio between the peak amplitude of the cross-correlation function and the amplitude of their respective fluctuation signals. If the peak significance of the cross-correlation function exceeds a preset correlation threshold, an effective fluctuation propagation correlation path is determined to exist between the two monitoring points; if the peak significance of the cross-correlation function does not exceed the preset correlation threshold, the possibility of a direct branch connection between the two monitoring points is excluded.

3. The method as described in claim 1, characterized in that, The step of determining the electrical distance information between each monitoring point based on the fused wave propagation delay characteristics includes: Using the power distribution line signal propagation speed model, the electrical distance information between each monitoring point is determined based on the fused wave propagation delay characteristics and the propagation speed parameters corresponding to the power distribution line signal propagation speed model. The signal propagation speed model for power distribution lines includes the correspondence between line type parameters and signal propagation speed. The method further includes: The propagation speed parameter is self-calibrated using the measured propagation delay of a reference monitoring point pair with known electrical distance, and the calibrated propagation speed parameter is used for subsequent electrical distance calculation.

4. The method as described in claim 1, characterized in that, Also includes: The wave propagation delay characteristics and wave attenuation characteristics are continuously monitored to determine the characteristic deviation between the current propagation characteristic value and the historical stable propagation characteristic value. If the characteristic deviation exceeds the topology change detection threshold, a topology change alarm is triggered and the re-identification of the electrical topology of the transformer area is initiated. If the characteristic deviation does not exceed the topology change detection threshold, it is determined that the current electrical topology information should be maintained unchanged. The method further includes: performing statistical fusion processing on multiple sets of propagation delay values ​​and attenuation values ​​extracted from multiple natural fluctuation events, and using the triangle inequality constraint to verify the consistency of the multiple sets of propagation delay values. If the delay relationship satisfies the triangle inequality constraint, the data set is determined to be valid. If the delay relationship violates the triangle inequality constraint, the weight of the data set is reduced or it is removed.

5. The method as described in claim 1, characterized in that, The detection of natural fluctuation events caused by the normal operation of the transformer substation from the electrical quantity time series data includes: Calculate the voltage rate of change and / or current rate of change of the electrical quantity time-series data; If the rate of change of voltage and / or the rate of change of current exceeds a preset fluctuation detection threshold, the natural fluctuation event is determined to have occurred at the current moment, and the event time and fluctuation amplitude characteristics are recorded. If the number of accumulated valid natural fluctuation events reaches the data sufficiency threshold, it is determined to perform the time-series correlation analysis and determine the electrical topology information; if the number of accumulated valid natural fluctuation events does not reach the data sufficiency threshold, it is determined to continue data accumulation and wait.

6. A self-identification system for electrical topology of a transformer substation, characterized in that, For performing the method as described in any one of claims 1-5, comprising: The data acquisition module is used to synchronously acquire the voltage and / or current of each monitoring point in the distribution area through the intelligent fusion terminal of the distribution area, and obtain the electrical quantity time sequence data of each monitoring point; The fluctuation detection module is used to detect natural fluctuation events generated by the normal operation of the transformer area from the electrical quantity time series data, evaluate the signal quality of each detected natural fluctuation event, filter high-quality fluctuation events with effective bandwidth exceeding a preset bandwidth threshold, and obtain the fluctuation time series data of each monitoring point corresponding to each high-quality fluctuation event. The time-series correlation analysis module is used to perform time-series correlation analysis between monitoring points based on the fluctuation time-series data of each monitoring point. It extracts the fluctuation propagation delay characteristics between any two monitoring points through subsampling interpolation of the cross-correlation function, and determines the fluctuation attenuation characteristics based on the cross-correlation peak amplitude. It performs self-estimation and elimination of the clock deviation between monitoring points on the fluctuation propagation delay characteristics extracted from multiple high-quality fluctuation events, and statistically fuses the multiple sets of propagation delay values ​​after eliminating the clock deviation to obtain the fused fluctuation propagation delay characteristics. The topology inference module is used to determine the electrical topology information of the transformer substation based on the fused wave propagation delay characteristics and wave attenuation characteristics.

7. The system as described in claim 6, characterized in that, The time series correlation analysis module is specifically used to: calculate the cross-correlation function between the time series data of the fluctuations at any two monitoring points, determine the time offset corresponding to the peak value of the cross-correlation function after subsampling interpolation as the fluctuation propagation delay feature, and determine the fluctuation attenuation feature based on the peak amplitude of the cross-correlation function; The system also includes: The topology update module is used to continuously monitor the wave propagation delay characteristics and the wave attenuation characteristics. When the characteristic deviation between the current propagation characteristic value and the historical stable propagation characteristic value exceeds the topology change detection threshold, the topology inference module is triggered to re-identify the electrical topology of the transformer area.

8. An electronic device, characterized in that, include: Memory; And a processor coupled to the memory, the processor being configured to perform the method as described in any one of claims 1 to 5 based on instructions stored in the memory.

9. A smart converged terminal for a distribution area, characterized in that, include: The electrical quantity acquisition unit is used to synchronously acquire the voltage and / or current of each monitoring point within the transformer substation area; The transformer substation electrical topology self-identification system as described in claim 6 or 7, wherein the data acquisition module of the self-identification system is connected to the electrical quantity acquisition unit, and is used to automatically identify the electrical topology of the transformer substation based on the data acquired by the electrical quantity acquisition unit.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 5.

Images