A method and system for intelligent monitoring of the operating status of low-voltage distribution areas

Abstract

The application discloses a kind of low-voltage area operating state intelligent monitoring method and system, it is related to electric power system monitoring technical field, the method includes the following steps: step S1, through the acquisition of multi-source heterogeneous operation data of deployment in low-voltage area sensing equipment;Step S2, to the multi-source heterogeneous operation data is carried out data cleaning and time alignment, and normalization is obtained to the regular time series data;Step S3, based on the regular time series data, constructs area health degree comprehensive calculation model, and the health degree index of each area is calculated;Step S4, according to the numerical range of area health degree index, generates graded early warning information;Step S5, based on historical voltage data and area topology structure information, the low voltage trend of each node of area in future period is predicted;Step S6, acquisition and prediction data are stored and visualized display.The application realizes the depth perception and fine management to area operating state by data analysis and prediction.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring technology, specifically to an intelligent monitoring method and system for the operating status of low-voltage distribution areas. Background Technology

[0002] A distribution area refers to a region in a power system supplied by a single distribution transformer. It is a term used in power economic operation and management, and in certain contexts, it can also be used as a place name. Its physical scope encompasses the lines, equipment, and coverage area from the low-voltage side of the transformer to the user's electricity meter. It is the basic unit for power supply companies to conduct distribution management, line loss accounting, and user services.

[0003] As the final stage of power transmission in the power grid, the operating status of low-voltage distribution substations directly determines the power quality and reliability for users. With socio-economic development, users' demands for power quality and power supply continuity are increasing, making real-time, accurate, and intelligent monitoring of low-voltage distribution substations crucial. Effective monitoring can promptly detect potential problems such as low voltage, equipment overload, and three-phase imbalance, forming the foundation for rapid fault response, improved operation and maintenance efficiency, and the safe and stable operation of the power grid.

[0004] For example, Chinese patent CN116566035A discloses a method for monitoring the status of low-voltage distribution areas, including: S1, a mobile terminal detects and acquires the grid structure data of the low-voltage distribution area and uploads it to the master station; S2, the master station retrieves low-voltage measurement data and establishes a low-voltage power grid quality analysis model based on the grid structure data of the low-voltage distribution area; S3, the low-voltage power grid quality analysis model is used to monitor the status of the low-voltage distribution area and resolve abnormal states. This invention aims to solve the power grid quality problems in low-voltage distribution areas by establishing a low-voltage power grid analysis model based on the grid structure of the low-voltage distribution area and measurement data. This model calculates and analyzes low-voltage weak points, guides the optimization of the low-voltage grid structure, solves problems such as low voltage, high loss, and frequent faults, and provides support for subsequent distribution area load, photovoltaic grid connection, and project planning.

[0005] For example, Chinese patent CN111026927B discloses an intelligent monitoring system for the operating status of low-voltage distribution areas. This system includes a data acquisition module for collecting operating parameter data; a topology identification data analysis module that uses a topology identification calculation model for statistical analysis, identifying or verifying relationships between distribution boxes and meters, and between customers and transformers, to achieve topology identification and mapping; and a topology identification imaging module for visualizing the data analysis results, dynamically generating topology maps and tracking and verifying them in real time, supporting dynamic and conditional queries of distribution area management, distribution area topology, anomalies, and analysis logs. This invention can effectively realize a safe electricity use management and safe operation and maintenance monitoring mode for the power grid, providing a reliable basis for overall power grid planning and unified management.

[0006] The above patents all suffer from the problems mentioned in the background technology: the evaluation methods are mostly based on single indicator threshold judgment or simple static weighted scoring, which cannot comprehensively and dynamically reflect the overall health level of the transformer area; and most of them are post-event alarms, which cannot achieve proactive early warning of faults. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide an intelligent monitoring method and system for the operating status of low-voltage distribution areas, addressing the shortcomings of the existing technology.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0009] A method for intelligent monitoring of the operating status of low-voltage distribution areas includes the following steps:

[0010] Step S1: Collect multi-source heterogeneous operating data through sensing devices deployed in the low-voltage distribution area, wherein the multi-source heterogeneous operating data includes: electrical data, event data, status data and environmental data;

[0011] Step S2: Perform data cleaning and time alignment on the multi-source heterogeneous operating data, and normalize it to obtain regular time-series data;

[0012] Step S3: Based on the regularized time-series data, construct a comprehensive calculation model for the health of each transformer substation and calculate the health index of each substation.

[0013] Step S4: Generate graded early warning information based on the numerical range of the health index of the transformer area;

[0014] Step S5: Based on historical voltage data and transformer area topology information, predict the low voltage trend of each node in the transformer area in the future period.

[0015] Step S6: Store and visualize the warning information, multi-source heterogeneous operation data, health index and low voltage trend prediction results.

[0016] Furthermore, in step S1, the electrical data includes three-phase voltage, three-phase current, and zero-sequence current; the event data includes switch trip signals and protection action events; the status data includes terminal online status; and the environmental data includes transformer oil temperature.

[0017] Furthermore, in step S2, data cleaning specifically includes: using an adaptive weighted sliding window filtering algorithm, wherein each data point within the sliding window is assigned an adaptive weight, which is determined by the time decay factor and the degree of mutation of the data point relative to the neighboring data, and the final cleaned data value is the weighted average of all data within the window.

[0018] Furthermore, in step S2, time alignment specifically includes: using a dynamic time warping algorithm to align the time of multi-source data.

[0019] Furthermore, in step S3, the calculation method of the comprehensive health calculation model of the transformer area is as follows: construct a basic health index and a penalty factor respectively, and multiply the basic health index and the penalty factor of each transformer area to obtain the final health index of each transformer area.

[0020] The basic health index includes: voltage quality index, communication status index, device event index, and load balancing index.

[0021] The penalty factors include: temperature-related penalty terms and voltage-related penalty terms. The temperature-related penalty terms consist of a temperature penalty coefficient and a transformer oil temperature over-limit indication function, while the voltage-related penalty terms consist of a voltage penalty coefficient and a voltage quality indication function.

[0022] Furthermore, in step S4, the specific process of generating graded early warning information is as follows: four health index intervals are preset. When the health index is greater than or equal to 0.8 and less than or equal to 1.0, it corresponds to a level one normal warning; when the health index is greater than or equal to 0.6 and less than 0.8, it corresponds to a level two attention warning; when the health index is greater than or equal to 0.4 and less than 0.6, it corresponds to a level three abnormal warning; and when the health index is greater than or equal to 0 and less than 0.4, it corresponds to a level four severe warning. The system calculates the health index in real time and determines its interval, and automatically generates the corresponding level of warning signal.

[0023] Furthermore, in step S5, predicting the low voltage trend of each node in the transformer area in the future specifically includes: using an undirected graph structure constructed based on the physical topology of the transformer area, and the historical voltage time series data corresponding to each node in the graph as model input; using a spatiotemporal graph convolutional network model for prediction, wherein the spatiotemporal graph convolutional network model contains three cascaded spatiotemporal convolutional blocks; each spatiotemporal convolutional block is composed of a spatial dimension graph convolutional layer and a time dimension one-dimensional convolutional layer connected in series; after the three spatiotemporal convolutional blocks are stacked, they are finally passed through a fully connected output layer to obtain the voltage prediction value sequence corresponding to each node in a specific future time period.

[0024] Furthermore, in the spatiotemporal convolution block, the graph convolution layer adopts a spectral graph convolution method based on Chebyshev polynomial approximation, and the order K of the Chebyshev graph polynomial is set to 3; the one-dimensional convolution layer adopts an dilated causal convolution structure, and the kernel size of the dilated causal convolution layer is set to 3, and its dilation coefficients are set to 1, 2 and 4 in the three blocks respectively.

[0025] A low-voltage distribution area operation status intelligent monitoring system, comprising:

[0026] The data acquisition module is used to collect multi-source heterogeneous operating data from various intelligent terminal devices in the low-voltage distribution area;

[0027] The data preprocessing module is used to execute the adaptive weighted sliding window filtering algorithm to clean and align the data;

[0028] The health calculation module is used to calculate the dynamically weighted health index of the transformer area;

[0029] The intelligent early warning module is used to generate tiered early warning information based on the health index;

[0030] The spatiotemporal prediction module is used to predict low voltage trends based on a spatiotemporal graph convolutional network model.

[0031] The data storage and visualization module is used to store all data and analysis results and provides a graphical human-computer interaction interface.

[0032] Furthermore, the health calculation module includes:

[0033] The entropy weight calculation unit is used to dynamically calculate the weighting coefficients of voltage quality, communication status, equipment events, and load balancing indicators based on the historical data volatility of each indicator.

[0034] The nonlinear transformation unit, with a built-in Sigmoid function, is used to perform nonlinear mapping processing on the voltage quality index and the device event index.

[0035] The penalty factor calculation unit is used to calculate the multiplicative penalty factor based on the transformer oil temperature exceeding the limit and the voltage quality index exceeding the limit.

[0036] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0037] 1. This invention overcomes the contradiction between effectively removing noise and preserving the true abrupt changes in data by employing an adaptive weighted sliding window filtering algorithm. This algorithm can automatically identify and smooth random noise while accurately preserving true fault characteristics such as voltage drops and current surges, providing a high-quality, highly reliable data foundation for subsequent advanced analysis and greatly improving the accuracy of the monitoring system.

[0038] 2. The dynamic weighted health index model proposed in this invention innovatively integrates multi-dimensional indicators such as voltage quality, communication status, equipment events, and load balancing, and introduces the entropy weight method to dynamically adjust the weights, enabling the evaluation results to objectively reflect the actual changes in the importance of each indicator. Furthermore, the introduction of nonlinear transformation and penalty factors makes the evaluation results not only comprehensive but also more sensitive to severe anomalies.

[0039] 3. This invention utilizes spatiotemporal graph convolutional networks for low-voltage trend prediction, overcoming the limitations of traditional time series prediction methods that only consider the time dimension while ignoring spatial topological relationships. This model can simultaneously capture the temporal evolution of voltage at each monitoring point within the transformer area and its spatial correlation and propagation effects in the electrical topology, thereby achieving earlier, more accurate, and spatially more identifiable predictions of low-voltage risks. This provides crucial technical support for the early deployment of voltage regulation measures and the prevention of low-voltage occurrences.

[0040] 4. This invention organically integrates data acquisition, cleaning, evaluation, early warning, and prediction functional modules into a collaborative overall system. This system not only provides full-cycle insight from real-time monitoring to long-term prediction, but also displays it centrally through a visual interface, greatly improving the overall perception of the status of low-voltage distribution areas and the efficiency of operation and maintenance management for operation and maintenance personnel, and realizing intelligent and lean operation and maintenance of low-voltage distribution networks. Attached Figure Description

[0041] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0042] Figure 1 This is a flowchart illustrating an embodiment of the present invention;

[0043] Figure 2 This is a schematic diagram of the system modules according to an embodiment of the present invention;

[0044] Figure 3 This is a schematic diagram of a spatiotemporal graph convolutional network according to an embodiment of the present invention;

[0045] Figure 4 This is a schematic diagram of a visualization system according to an embodiment of the present invention. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0047] like Figure 1 As shown, a method for intelligent monitoring of the operating status of low-voltage distribution areas includes the following steps:

[0048] Step S1: Collect multi-source heterogeneous operating data through sensing devices deployed in the low-voltage distribution area, wherein the multi-source heterogeneous operating data includes: electrical data, event data, status data and environmental data;

[0049] Step S2: Perform data cleaning and time alignment on the multi-source heterogeneous operating data, and normalize it to obtain regular time-series data;

[0050] Step S3: Based on the regularized time-series data, construct a comprehensive calculation model for the health of each transformer substation and calculate the health index of each substation.

[0051] Step S4: Generate graded early warning information based on the numerical range of the health index of the transformer area;

[0052] Step S5: Based on historical voltage data and transformer area topology information, predict the low voltage trend of each node in the transformer area in the future period.

[0053] Step S6: Store and visualize the warning information, multi-source heterogeneous operation data, health index and low voltage trend prediction results.

[0054] In step S1, the electrical data includes three-phase voltage, three-phase current, and zero-sequence current; the event data includes switch trip signals and protection action events; the status data includes terminal online status; and the environmental data includes transformer oil temperature.

[0055] In step S2, data cleaning specifically includes: using an adaptive weighted sliding window filtering algorithm, wherein each data point in the sliding window is assigned an adaptive weight, which is determined by the time decay factor and the degree of mutation of the data point relative to the neighboring data. The final cleaned data value is the weighted average of all data in the window.

[0056] The calculation formula for the adaptive weighted sliding window filtering algorithm is as follows:

[0057]

[0058] in, This represents the data value after cleaning at time t. This represents the original data value at time t+i. Indicates the adaptive weighting coefficient. This represents the relative position index within the sliding window. Indicates the size of the sliding window;

[0059] The formula for calculating the adaptive weight coefficient is as follows:

[0060]

[0061] in, Represents the time decay constant. This represents the difference between data points at adjacent time points. Indicates taking the absolute value;

[0062] Matching the window size of 2N+1, typically The setting causes the weight at the window edges to decay to a very small value, which can be set... .

[0063] In step S2, time alignment specifically includes: using a dynamic time warping algorithm to align the time of multi-source data.

[0064] Dynamic Time Warping (DTW) is a technique for measuring the similarity between two time series, particularly suitable for those time series that may have different velocities at different time points. DTW finds the optimal matching path by non-linearly warping the time series to minimize the distance between them.

[0065] The specific steps include: defining the distance matrix: given two time series X and Y, containing n and m data points respectively, first construct an n×m distance matrix D, where the element D[i][j] represents the Euclidean distance or other suitable distance metric between the i-th point of X and the j-th point of Y.

[0066] Cumulative Distance Matrix: Next, we define a cumulative distance matrix γ, where γ[i][j] represents the total distance of the shortest path from the starting point (0,0) to point (i,j). This shortest path needs to satisfy the following constraints:

[0067] Monotonicity: The path can only move to the right or up.

[0068] Boundary condition: The path must start at (0,0) and end at (n,m).

[0069] Continuity: Every step on the path must be continuous, meaning no points can be skipped.

[0070] Recursive calculation: Use dynamic programming to fill the cumulative distance matrix γ.

[0071] Finding the optimal path: Once the cumulative distance matrix is ​​filled, the optimal path can be found step by step by backtracking from γ[n][m] according to the principle of minimum cumulative distance.

[0072] Normalized distance: In order to compare time series of different lengths, the final minimum cumulative distance is usually normalized by dividing the path length by some function.

[0073] In step S3, the calculation method of the comprehensive health calculation model of the transformer area is as follows: construct the basic health index and the penalty factor respectively, and multiply the basic health index and the penalty factor of each transformer area to obtain the final health index of each transformer area.

[0074] The basic health index includes: voltage quality index, communication status index, device event index, and load balancing index.

[0075] The penalty factors include: temperature-related penalty terms and voltage-related penalty terms. The temperature-related penalty terms consist of a temperature penalty coefficient and a transformer oil temperature over-limit indication function, while the voltage-related penalty terms consist of a voltage penalty coefficient and a voltage quality indication function.

[0076] The calculation formula for the comprehensive health assessment model of the power grid area is as follows:

[0077]

[0078] in, This indicates the health index of the Taiwan area. This represents the basic health index. Indicates the penalty factor;

[0079] The formula for calculating the basic health index is as follows:

[0080]

[0081] in, Indicates the voltage quality index. Indicates the communication status index. Indicates the equipment event index, This represents the load balancing index. , , and These represent the weights of the voltage quality index, communication status index, equipment event index, and load balancing index, respectively. These weights are dynamically calculated using the entropy weight method based on the volatility of historical data for each indicator. The greater the volatility of an indicator, the higher its weight, indicating that the indicator is currently better able to distinguish differences in the health of the platform area. and This represents a nonlinear transformation function used to map the original index value to a contribution level and suppress extreme values.

[0082] in, For a standard Sigmoid function, the voltage quality index The higher the better, so we directly use the Sigmoid function to map it to contribution. To Inverting the value and then inputting it into the standard Sigmoid function ensures that the more events there are, the lower the contribution.

[0083] The formula for calculating the voltage quality index is as follows:

[0084]

[0085] in, This indicates the total number of voltage monitoring points within the transformer area. This represents the j-th voltage monitoring point. This represents the voltage value measured at the j-th monitoring point at time t. Indicates the rated voltage of the power grid;

[0086] The communication status index is the ratio of the number of online terminal devices to the total number of terminal devices configured in the area at time t.

[0087] The formula for calculating the equipment event index is:

[0088]

[0089] in, This represents the decay coefficient, used to control the rate at which the effect of an event on the exponential effect decays. This indicates the most recent time window, typically set to 24 hours. This indicates the total number of protection device tripping events in the past 24 hours. This represents the scaling factor, used to convert the magnitude of the zero-sequence current into an equivalent event intensity. This represents the average value of the zero-sequence current over the past 24 hours.

[0090] This index is an exponentially decaying function; the more severe the tripping event and ground fault characteristics, the larger the value within the parentheses, and the smaller the equipment event index. When there are no events, the equipment event index is 1.

[0091] The formula for calculating the load balancing index is:

[0092]

[0093] in, This represents the three-phase current at the transformer outlet side at time t. This represents the average value function; the fractional part in the formula calculates the three-phase unbalance, which is 1 when the three phases are perfectly balanced. The higher the unbalance, the lower the load balancing index.

[0094] The formula for calculating the penalty factor is:

[0095]

[0096] in, This represents the temperature penalty coefficient, typically set to 0.3. This indicates an indicator function; the function value is 1 when the condition inside the parentheses is true, and 0 when the condition is false. Indicates transformer oil temperature. This indicates the safe temperature threshold for a transformer, typically ranging from 85°C to 95°C (for oil-immersed distribution transformers). This represents the voltage penalty factor, typically ranging from 0.2 to 0.5. The critical threshold representing the voltage quality index is typically set to 0.7.

[0097] In step S4, the specific process of generating graded early warning information is as follows: four health index ranges are preset. When the health index is greater than or equal to 0.8 and less than or equal to 1.0, it corresponds to a level one normal warning; when the health index is greater than or equal to 0.6 and less than 0.8, it corresponds to a level two attention warning; when the health index is greater than or equal to 0.4 and less than 0.6, it corresponds to a level three abnormal warning; and when the health index is greater than or equal to 0 and less than 0.4, it corresponds to a level four severe warning. The system calculates the health index in real time and determines its range, and automatically generates the corresponding level of warning signal.

[0098] Based on the numerical range, different levels of warning information are generated:

[0099] Level 1 (Normal): 0.8 ≤ ≤ 1.0 (green)

[0100] Level 2 (Note): 0.6 ≤ < 0.8 (yellow)

[0101] Level 3 (Abnormal): 0.4≤ < 0.6 (orange)

[0102] Level 4 (Severe): 0 ≤ < 0.4 (red)

[0103] In step S5, predicting the low voltage trend of each node in the transformer substation for a future period specifically includes: using an undirected graph structure constructed based on the physical topology of the transformer substation, and the historical voltage time series data corresponding to each node in the graph as model input; using a spatiotemporal graph convolutional network model for prediction, wherein the spatiotemporal graph convolutional network model contains three cascaded spatiotemporal convolutional blocks; each spatiotemporal convolutional block is composed of a spatial dimension graph convolutional layer and a time dimension one-dimensional convolutional layer connected in series; after the three spatiotemporal convolutional blocks are stacked, they are finally passed through a fully connected output layer to obtain the voltage prediction value sequence corresponding to each node in a specific future period.

[0104] Model input:

[0105] Node characteristics: Historical voltage data of each node (transformer, meter) in the distribution area topology network.

[0106] Graph structure: Adjacency matrix A is constructed based on the physical wiring diagram of the transformer substation.

[0107] In the spatiotemporal convolution block, the graph convolution layer adopts the spectral graph convolution method based on Chebyshev polynomial approximation, and the order K of the Chebyshev graph polynomial is set to 3; the one-dimensional convolution layer adopts the dilated causal convolution structure, and the kernel size of the dilated causal convolution layer is set to 3, and its dilation coefficients are set to 1, 2 and 4 in the three blocks respectively.

[0108] Setting K to 3 means that the receptive field of each node covers its third-order neighbors (i.e., itself, its direct neighbors, and its neighbors' neighbors), which is usually sufficient to capture most of the spatial influence in the area.

[0109] With expansion coefficients of 1, 2, and 4, the temporal receptive field at the top layer is thus 1 + (3-1). 1 + (3-1) 2 + (3-1) 4 = 1+2+4+8=15 time steps, which can effectively cover a sufficiently long historical period.

[0110] like Figure 2 As shown, a low-voltage distribution area operation status intelligent monitoring system includes:

[0111] The data acquisition module is used to collect multi-source heterogeneous operating data from various intelligent terminal devices in the low-voltage distribution area;

[0112] The data preprocessing module is used to execute the adaptive weighted sliding window filtering algorithm to clean and align the data;

[0113] The health calculation module is used to calculate the dynamically weighted health index of the transformer area;

[0114] The intelligent early warning module is used to generate tiered early warning information based on the health index;

[0115] The spatiotemporal prediction module is used to predict low voltage trends based on a spatiotemporal graph convolutional network model.

[0116] The data storage and visualization module is used to store all data and analysis results and provides a graphical human-computer interaction interface.

[0117] The health calculation module includes:

[0118] The entropy weight calculation unit is used to dynamically calculate the weighting coefficients of voltage quality, communication status, equipment events, and load balancing indicators based on the historical data volatility of each indicator.

[0119] The nonlinear transformation unit, with a built-in Sigmoid function, is used to perform nonlinear mapping processing on the voltage quality index and the device event index.

[0120] The penalty factor calculation unit is used to calculate the multiplicative penalty factor based on the transformer oil temperature exceeding the limit and the voltage quality index exceeding the limit.

[0121] like Figure 3 The image shows the specific structure of the spatiotemporal graph convolutional network model:

[0122] The input layer takes historical voltage sequence data and transformer area topology adjacency matrix A as input.

[0123] Three convolutional blocks, each containing two sub-layers; the arrows indicate the data flow.

[0124] Graph convolutional layers process input data and output spatial features.

[0125] The kernel size of the one-dimensional convolutional layer is 3, and the dilation coefficients are d=1, d=2, and d=4 respectively. This layer receives spatial features and outputs spatiotemporal features.

[0126] The output of the third convolutional block is connected to a fully connected layer.

[0127] The final output layer outputs a future voltage prediction sequence.

[0128] like Figure 4 As shown, the low voltage monitoring and display platform for the transformer substation shows the low voltage records and trend charts for the transformer substation.

[0129] The examples described herein are merely preferred embodiments of the invention and are not intended to limit the concept and scope of the invention. Any modifications and improvements made by those skilled in the art to the technical solutions of the invention without departing from the design concept of the invention should fall within the protection scope of the invention.

Claims

1. A low-voltage transformer area operation state intelligent monitoring method, characterized in that, The method comprises the following steps: Step S1, collecting multi-source heterogeneous operation data through a sensing device deployed in a low-voltage transformer area, wherein the multi-source heterogeneous operation data comprises electrical data, event data, state data and environmental data; Step S2, performing data cleaning and time alignment on the multi-source heterogeneous operation data, and normalizing to obtain regular time series data; Step S3, constructing a transformer area health degree comprehensive calculation model based on the regular time series data, and calculating a health degree index of each transformer area; Step S4, generating graded early warning information according to the numerical range of the transformer area health degree index; Step S5, predicting low-voltage trends of nodes in the transformer area in a future period based on historical voltage data and transformer area topology structure information; Step S6, storing and visually displaying the early warning information, multi-source heterogeneous operation data, health degree index and low-voltage trend prediction results; In step S2, data cleaning specifically comprises: adopting an adaptive weighted sliding window filtering algorithm, wherein each data point in the sliding window is given an adaptive weight, and the adaptive weight is determined by a time decay factor and the mutation degree of the point data relative to the neighborhood data. The final cleaned data value is the weighted average value of all data in the window. The calculation formula of the adaptive weighted sliding window filtering algorithm is: ； wherein, represents a data value after cleaning at time t, represents an original data value at time t+i, represents an adaptive weight coefficient, represents a relative position index within a sliding window, represents a sliding window size; The calculation formula of the adaptive weight coefficient is: ； wherein, represents a time decay constant, represents a difference of adjacent time point data, represents taking an absolute value; In step S3, the calculation method of the transformer area health degree comprehensive calculation model is: a health degree basic index and a penalty factor are constructed respectively, and the health degree basic index and the penalty factor of each transformer area are multiplied to obtain the final health degree index of each transformer area; The health degree basic index comprises: a voltage quality index term, a communication state index term, a device event index term and a load balancing index term; The penalty factor comprises: a temperature related penalty term and a voltage related penalty term, wherein the temperature related penalty term is composed of a temperature penalty coefficient and a transformer oil temperature out-of-limit indicator function, and the voltage related penalty term is composed of a voltage penalty coefficient and a voltage quality indicator function; The calculation formula of the transformer area health degree comprehensive calculation model is: ； wherein, represents a health index of a transformer station, represents a health base index, represents a penalty factor; The calculation formula of the health degree basic index is: ； wherein denotes a voltage quality index, denotes a communication status index, denotes a device event index, denotes a load balancing index, , , and denote weights of the voltage quality index term, the communication status index term, the device event index term and the load balancing index term, respectively, and denote a non-linear transformation function; The calculation formula of the voltage quality index is: ； wherein, denotes the total number of voltage monitoring points within the substation, denotes the jth voltage monitoring point, denotes the voltage value measured at the jth monitoring point at time t, denotes the rated voltage of the power grid; The communication state index is the ratio of the number of terminal devices in an online state at time t to the total number of terminal devices configured in the transformer area; The calculation formula of the device event index is: ； wherein, represents a decay coefficient, used to control the decay speed of the influence of the event on the index, represents the recent time window, usually set to 24 hours, represents the total number of protection device tripping events in the last 24 hours, represents a proportionality coefficient, used to convert the amplitude of the zero sequence current into an equivalent event intensity, represents the average value of the zero sequence current in the last 24 hours; The calculation formula of the load balancing index is: ； wherein, denotes the three-phase current at the transformer outlet side at time t, denotes the averaging function; The calculation formula of the penalty factor is: ； wherein represents a temperature penalty coefficient, represents an indicator function, which has a value of 1 when the condition within the brackets is true and a value of 0 when the condition is false, represents the transformer oil temperature, represents a safety threshold for the transformer temperature, represents a voltage penalty coefficient, represents a critical threshold for the voltage quality index.

2. The method of claim 1, wherein, In step S1, the electrical data comprises three-phase voltage, three-phase current and zero-sequence current; the event data comprises switch tripping signals and protection action events; the state data comprises terminal online state; and the environmental data comprises transformer oil temperature.

3. The method of claim 2, wherein, In step S2, time alignment specifically comprises: using a dynamic time warping algorithm to perform time alignment on multi-source data.

4. The method of claim 3, wherein, In the step S4, the specific process of generating the hierarchical early warning information is: four health degree index intervals are set in advance, when the health degree index is greater than or equal to 0.8 and less than or equal to 1.0, it corresponds to a first level normal early warning, when the health degree index is greater than or equal to 0.6 and less than 0.8, it corresponds to a second level attention early warning, when the health degree index is greater than or equal to 0.4 and less than 0.6, it corresponds to a third level abnormal early warning, and when the health degree index is greater than or equal to 0 and less than 0.4, it corresponds to a fourth level serious early warning; the system calculates the health degree index in real time and judges the interval it belongs to, and automatically generates the early warning signal of the corresponding level.

5. The method of claim 4, wherein, In the step S5, the low voltage trend of each node in the future period of the substation area is predicted, specifically including: a undirected graph structure constructed by the physical topology connection relationship of the substation area, and historical voltage time series data corresponding to each node in the graph as model input; a spatiotemporal graph convolution network model is used for prediction, the spatiotemporal graph convolution network model contains 3 cascaded spatiotemporal convolution blocks; each spatiotemporal convolution block is composed of a graph convolution layer in spatial dimension and a one-dimensional convolution layer in time dimension in series; after the 3 spatiotemporal convolution blocks are stacked, the final voltage prediction value sequence corresponding to each node in a specific period in the future is obtained through a fully connected output layer.

6. The method of claim 5, wherein, In the spatiotemporal convolution block, the graph convolution layer adopts a spectral graph convolution method based on Chebyshev polynomial approximation, and the order K of the Chebyshev graph polynomial is set to 3; the one-dimensional convolution layer adopts a dilated causal convolution structure, and the convolution kernel size of the dilated causal convolution layer is set to 3, and its dilated coefficients are set to 1, 2 and 4 in the 3 blocks.

7. A low-voltage transformer area operation state intelligent monitoring system based on the low-voltage transformer area operation state intelligent monitoring method of any one of claims 1-6, characterized in that, It comprises: a data acquisition module for acquiring multi-source heterogeneous operation data from various intelligent terminal devices in the low-voltage substation area; a data preprocessing module for executing an adaptive weighted sliding window filtering algorithm to clean and align the data; a health degree calculation module for calculating a dynamically weighted substation health degree index; an intelligent early warning module for generating hierarchical early warning information according to the health degree index; a spatiotemporal prediction module for predicting the low voltage trend based on a spatiotemporal graph convolution network model; a data storage and visualization module for storing all data and analysis results and providing a graphical human-computer interaction interface.

8. The system of claim 7, wherein, The health degree calculation module comprises: an entropy weight method calculation unit for dynamically calculating the weight coefficients of the voltage quality, communication status, device event and load balancing indexes according to the historical data volatility of each index; a nonlinear transformation unit with a built-in Sigmoid function for nonlinear mapping processing of the voltage quality index and the device event index; a penalty factor calculation unit for calculating a multiplicative penalty factor according to the transformer oil temperature overrun and the voltage quality index overrun.

Images