A power transmission and transformation equipment health state prediction method and system

Abstract

The application provides a power transmission and transformation equipment health state prediction method and system, and the method comprises the following steps: obtaining equipment state monitoring data and switch state data of power transmission and transformation equipment, and constructing power transmission and transformation equipment state data; obtaining the switch state of all equipment and corresponding equipment structure information, and constructing the operation correlation weight of any two equipment at the same time; generating a health state characterization vector of the corresponding equipment under the constraint of the state bearing operation correlation graph; based on the health state characterization vector generated at the current moment, the health state characterization vector at the last moment and the health state prediction value at the last moment, calculating the health state prediction value at the current moment, and mapping the health state prediction value to a discrete output label according to the historical prediction baseline of the equipment itself. The application improves the accuracy, stability and scene explanation ability of the prediction result, and better supports state maintenance, risk early warning and operation and maintenance decision-making.

Description

Technical Field

[0001] This invention belongs to the field of health status prediction of power transmission and transformation equipment, and particularly relates to a method and system for predicting the health status of power transmission and transformation equipment. Background Technology

[0002] Transmission and transformation equipment is the core basic unit in a power system responsible for power transformation, transmission, interruption, isolation, and protection. It mainly includes transformers, circuit breakers, disconnectors, busbars, and transmission lines. This type of equipment operates under the combined effects of high voltage, heavy loads, frequent switching of operating conditions, and complex environments. Its health status is not determined by a single monitoring quantity, but is influenced by the equipment's own operating status, switch connection status, and the status of adjacent equipment in the same electrical corridor. It continuously evolves with load transfers, operating mode adjustments, busbar switching, bay connection / deactivation, and the accumulation of local defects. In current engineering practice, most health assessment methods for transmission and transformation equipment still rely on single-equipment monitoring quantities for threshold judgment, trend analysis, or scoring calculations. While these methods can reflect the operational anomalies of individual equipment at a certain moment, they do not adequately consider the effective operational correlations between equipment caused by switch opening and closing and changes in wiring structure. They fail to depict the true process of coordinated changes and local deviations in the status of multiple devices within the same operating corridor. Some methods introduce topological relationships for joint analysis, but they usually directly use fixed wiring diagrams or static adjacency relationships, lacking a process to unify the organization of equipment monitoring status and switch status, and to identify the current effective operational associations accordingly. Therefore, they are difficult to adapt to scenarios such as bus switching, main transformer switching, line switching, and bay mode adjustment that frequently occur in the actual operation of power transmission and transformation systems. In the process of multi-equipment state modeling, existing methods also use the interaction of adjacent equipment states in a relatively crude way, often lacking a constraint mechanism consistent with the current operating channel. This is not conducive to extracting the collaborative state characteristics within the same operating domain, and it is also difficult to identify the degree of local imbalance of equipment relative to the neighboring state, resulting in a lack of stable correspondence between the health status representation and the actual operating conditions. Furthermore, in the prediction output stage, existing technologies typically focus on providing a health score or simple level at a certain moment, rarely combining the current characterization intensity, the rate of change at adjacent moments, and the cumulative trend of continuous deterioration for comprehensive judgment. They also lack a processing method that combines the equipment's own historical operating baseline to form an individualized output label. Therefore, when facing scenarios such as continuous high load operation of the main transformer, state transition after circuit breaker operation, and local abnormal rise caused by line operation mode switching, the prediction results are prone to problems such as insufficient sensitivity, insufficient stability, or weak interpretability.

[0003] Because the difficulty in predicting the health status of power transmission and transformation equipment lies in the continuous processing chain of data organization, operation correlation identification, health characterization generation, and prediction output, it is necessary to construct a technical solution that can unify equipment status, switch status, operation correlation, and time evolution process to improve the consistency and engineering applicability between health status prediction results and actual operating status. Summary of the Invention

[0004] The purpose of this invention is to propose a method and system for predicting the health status of power transmission and transformation equipment, thereby solving the above-mentioned problems.

[0005] To achieve the above objectives, a method for predicting the health status of power transmission and transformation equipment is provided in a first aspect of the present invention, the method comprising the following steps: S1. Obtain equipment status monitoring data and switch status data of power transmission and transformation equipment, and construct power transmission and transformation equipment status data; wherein, the power transmission and transformation equipment status data includes a status vector of all equipment at the same point in time, indexed and organized according to equipment identifier; the status vector includes equipment status monitoring data and the switch status of the corresponding equipment; S2. Obtain the switch status of all devices and the corresponding device structure information, combine the corresponding device status monitoring data, construct the operation association weight of any two devices at the same time; and based on the operation association weight of all device pairs, compare it with a predefined judgment threshold to determine whether any two devices establish an operation association at the same time. If the weight of the current device pair's operational association is greater than the predefined judgment threshold, then an operational association is established; otherwise, it is not established. Based on all operational associations, each device pair forms a device node and its corresponding associated edge, which are organized into a state-carrying operational association graph. S3. Under the constraints of the state-bearing operation association diagram, for each device, the corresponding device state vector, the cooperative state of neighboring devices in the operation channel, and the local imbalance characteristics relative to the neighboring devices are fused to generate the health status representation vector of the corresponding device. S4. Based on the health status representation vector generated at the current moment, the health status representation vector at the previous moment, and the health status prediction value at the previous moment, calculate the health status prediction value at the current moment, and map the health status prediction value into discrete output labels according to the device's own historical prediction baseline.

[0006] Preferably, the power transmission and transformation equipment includes: transformers, transmission lines, and circuit breakers; The equipment status monitoring data always includes two components: the first component represents the current flow or load status of the equipment at the current moment, and the second component represents the stress status of the equipment at the current moment.

[0007] Preferably, the equipment structure information includes the station, voltage level, and bus segment or bay information carried in the equipment number.

[0008] Preferably, in step S2, the specific method for identifying valid operational associations and generating a dynamic operational association graph is as follows: First, using the equipment structure information, all equipment is grouped into a candidate operation set; Then, within each candidate run set, the run association weight, which measures the association strength, is calculated based on the on / off status of any two devices, the degree of difference between their status monitoring data vectors, and the structure matching coefficient predefined by the device structure relationship. Finally, the calculated operational association weights of all device pairs are compared with the judgment thresholds corresponding to the device type combinations. Device pairs whose weight values ​​reach or exceed the thresholds are judged to have operational associations, and the operational association graph is constructed with devices as nodes and association relationships as edges.

[0009] Preferably, the collaborative state of neighboring devices within the operating channel is generated based on the operating association weights of adjacent devices in the state-carrying operating association graph; The local imbalance feature relative to the neighboring device is generated based on the local imbalance coefficient and the state vector of the corresponding device.

[0010] Preferably, when the power transmission and transformation equipment is a transformer, its condition monitoring data includes load current and oil temperature; when the power transmission and transformation equipment is a circuit breaker, its condition monitoring data includes current carrying capacity and operation count increment; when the power transmission and transformation equipment is a transmission line, its condition monitoring data includes line current and bus side voltage deviation or line temperature rise.

[0011] Preferably, the local imbalance coefficient is used to reflect the degree of deviation of the equipment from its operating channel.

[0012] Preferably, in step S4, the method for calculating the continuous predicted value of the health status is as follows: The continuous predicted value of health status is obtained by weighted summing of the current health level term based on the average intensity of the current health status representation vector, the short-term change term based on the rate of change of the health status representation vector between the current and previous moments, and the continuous degradation memory term based on the historical predicted value of the current influence.

[0013] Preferably, in step S4, the historical prediction baseline is determined by the median reference value of the historical prediction value sequence of the corresponding device within the rolling time window, and the historical fluctuation reference value is obtained by statistically analyzing the dispersion of the prediction value sequence within the same rolling window. The current continuous forecast value is then compared with a combination of the historical forecast baseline and historical fluctuation reference value, and mapped to discrete output labels including normal, attention and warning.

[0014] A second aspect of the present invention provides a health status prediction system for power transmission and transformation equipment, the system comprising: The data construction module is used to acquire equipment status monitoring data and switch status data of power transmission and transformation equipment, and construct power transmission and transformation equipment status data; wherein, the power transmission and transformation equipment status data includes a status vector of all equipment at the same point in time, indexed and organized according to equipment identifier; the status vector includes equipment status monitoring data and the switch status of the corresponding equipment; The association identification module is used to obtain the switch status of all devices and the corresponding device structure information, combine the corresponding device status monitoring data, construct the operation association weight of any two devices at the same time, and compare the operation association weight of all device pairs with a predefined judgment threshold to determine whether any two devices establish an operation association at the same time. If the weight of the current device pair's operational association is greater than the predefined judgment threshold, then an operational association is established; otherwise, it is not established. Based on all operational associations, each device pair forms a device node and its corresponding associated edge, which are organized into a state-carrying operational association graph. The characterization generation module is used to, under the constraints of the state-bearing operation association graph, fuse the corresponding device state vector, the cooperative state of neighboring devices in the operation channel, and the local imbalance features relative to the neighboring devices for each device to generate the health status characterization vector of the corresponding device. The prediction output module is used to calculate the predicted health status value at the current moment based on the health status representation vector generated at the current moment, the health status representation vector at the previous moment, and the predicted health status value at the previous moment, and to map the predicted health status value to discrete output labels according to the device's own historical prediction baseline.

[0015] The beneficial technical effects of the present invention are at least as follows: This invention focuses on the goal of predicting the health status of power transmission and transformation equipment, and integrates the input organization method, equipment association expression method, health status representation method and result output method in the prediction process.

[0016] Specifically, this invention first organizes equipment status monitoring information and switch status information into equipment-level status data under a unified time reference, and maps key status quantities of different devices to a unified expression structure, so that subsequent cross-device comparisons, difference calculations, and weighted fusion have a consistent data foundation; then, combining structural information such as the site, voltage level, bus section or bay, and switch connection status of the equipment, it identifies the effective operational associations between the devices under the current operating conditions, and forms an operational association diagram that simultaneously carries the equipment status and association relationships, so that the actual interaction relationships between devices can be updated synchronously with changes in operating mode; on this basis, the equipment's own status is... The health status is integrated into the health status representation, which includes the coordinated status of neighboring devices within the same operating channel and the local imbalance characteristics of the device relative to its neighboring status. This allows the representation results to reflect both the overall characteristics of the operating domain in which the device is located and the degree of deviation of locally abnormal devices. Furthermore, in the prediction output stage, the current health status representation intensity, the representation changes at adjacent times, and the continuous degradation memory are all included in the calculation. Combined with the device's own historical prediction baseline, continuous prediction values ​​and output markers are formed, enabling different types of equipment such as main transformers, circuit breakers, and lines to obtain more realistic prediction results under conditions of slow degradation, sudden rise, and operating condition switching.

[0017] Through the above processing, this invention elevates the prediction of the health status of power transmission and transformation equipment from single-point monitoring analysis to a continuous prediction method oriented towards real operational relationships and time evolution processes. This enables the equipment health status to form a complete closed technical chain from the construction of original monitoring data, identification of operational correlations, generation of health characteristics to the output of prediction results, thereby improving the accuracy, stability and scenario interpretation capabilities of the prediction results, and better supporting condition-based maintenance, risk warning and operation and maintenance decisions. Attached Figure Description

[0018] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0019] Figure 1 This is a flowchart of a method for predicting the health status of power transmission and transformation equipment according to the present invention. Detailed Implementation

[0020] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0021] like Figure 1As shown in the figure, an embodiment of the present invention provides a method for predicting the health status of power transmission and transformation equipment, the method comprising: S1. Obtain equipment status monitoring data and switch status data of power transmission and transformation equipment, and construct power transmission and transformation equipment status data; wherein, the power transmission and transformation equipment status data includes a status vector of all equipment at the same point in time, indexed and organized according to equipment identifier; the status vector includes equipment status monitoring data and the switch status of the corresponding equipment.

[0022] Specifically, this step revolves around constructing the input for predicting the health status of power transmission and transformation equipment. It organizes the equipment status monitoring data and switch status data, scattered across the station control system, bay control devices, and dispatch automation system, into a unified power transmission and transformation equipment status data set, D. In actual systems, transformer oil temperature is typically collected by a body temperature sensor and sent to the station control layer via an online monitoring device; line current is generally collected by a current transformer and uploaded to the SCADA system via a control device; bus voltage is collected by a voltage transformer and entered into the real-time database on the dispatch side; and the open / close position signals of circuit breakers and disconnectors are collected by protection devices or control units and uploaded as remote signaling quantities. These data naturally differ in their acquisition rhythms. Status monitoring data is mostly refreshed at fixed intervals, while switch status data is updated when the status changes. Therefore, the two types of data are first organized synchronously using a unified time scale: valid sampled values ​​near the target time are selected for status monitoring data, and the latest status value corresponding to that time is used for switch status data, thus forming a consistent status description for each piece of equipment at the same point in time. Taking a certain line bay as an example, if its circuit breaker changes from open to closed at a certain moment, then from that moment until the next state change, the corresponding switch state of the line will remain closed; the monitoring values ​​such as line current and bus voltage at the same moment will be mapped to the line equipment according to a unified time scale, forming a synchronous state record under the same equipment object.

[0023] Furthermore, after time alignment, a device state vector is constructed for each transmission and transformation device. To ensure that subsequent difference calculations, weighted summations, and norm operations between different devices can be directly performed, the original monitoring channels of each device are first mapped to state monitoring data vectors of the same dimension. In this embodiment, The fixed component comprises two parts: the first part represents the current-carrying or load state of the equipment at the current moment, and the second part represents the stress state of the equipment at the current moment. For transformers, the first part is mapped from the load current, and the second part is mapped from the oil temperature. For transmission lines, the first part is mapped from the line current, and the second part is mapped from the bus voltage deviation or line temperature rise monitoring value. For circuit breakers, the first part is mapped from the current-carrying current, and the second part is mapped from the increment of the action count or the mechanism state quantity per unit time. The resulting... Maintain the same dimension across all devices, and then correlate it with the on / off state. Combined to form a device state vector: ; in, This represents the state vector of the i-th power transmission and transformation equipment at time t; This represents a fixed two-dimensional state monitoring data vector of the device at time t, which is derived from the original monitoring channels in the online monitoring device, measurement and control device, or SCADA real-time library and obtained after unified mapping. This indicates the switching state of the device at time t, derived from the position signal of the circuit breaker or disconnector, and obtained after being uploaded by the protection device or monitoring and control device. After this construction, the device state vector simultaneously represents the current numerical state and the connection status of the device. For example, the load current mapping value of a transformer at time t is... Oil temperature mapping value The high-voltage side switch status value Then it can be written as The line current mapping value of a line that is out of service at the same time is... The bus-side voltage offset mapping value is The status value of the circuit breaker is Then it can be written as The raw sampled values ​​of each monitoring channel are mapped to a unified numerical range based on the equipment's rated parameters or historical operating range. The parameters required for mapping come from the rated configuration field in the equipment ledger system and long-term operating records in the historical database. After completing the construction of the single-device state vector, the state vectors of all devices at the same point in time are... Organize the data by equipment identifier to form the status data of power transmission and transformation equipment. In its implementation, it can be stored as a mapping structure of "device number - status vector" and updated over time by the scheduling master server or edge computing nodes.

[0024] S2. Obtain the switch status of all devices and the corresponding device structure information, combine the corresponding device status monitoring data, construct the operation association weight of any two devices at the same time; and based on the operation association weight of all device pairs, compare it with a predefined judgment threshold to determine whether any two devices establish an operation association at the same time. If the operational association weight of the current device pair is greater than the predefined judgment threshold, an operational association is established; otherwise, it is not established. Based on all operational associations, each device pair forms a device node and its corresponding association edge, which is organized into a state-bearing operational association graph.

[0025] Specifically, this step uses the power transmission and transformation equipment status data generated in S1. Using this as input, the system identifies effective operational relationships between various power transmission and transformation equipment at the same time scale and generates a state-bearing operational relationship diagram. In the actual system, step one has already uniformly expressed the state of each device at time t as a state vector. This includes equipment status monitoring data. and switch status Therefore, this step directly uses these state vectors as the basis for association identification processing. For each device record, the scheduling master station or edge computing node can read it from the real-time database using the device number. The equipment number also carries equipment structure information such as site code, voltage level, bus section number, or bay number. This information comes from the power grid model file or equipment ledger system and is bound to the monitoring points one by one during data acquisition. Using these structure fields, all equipment can be grouped at the current point in time, so that equipment at the same site, voltage level, and located in the same bus section or bay is grouped into the same candidate set. For example, incoming circuit breakers, bus tie circuit breakers, and main transformer high-voltage side monitoring points on a certain 220kV bus section will enter the same set, while equipment on another bus section or at a different voltage level will be grouped into other sets. The resulting candidate sets already reflect the potential relationships at the electrical wiring level and are directly derived from... Information accompanying the device number.

[0026] Furthermore, within each candidate set, runtime association weights are constructed for any two devices i and j. This weighting utilizes switch status, device structural relationships, and the consistency of status monitoring data to describe the correlation strength between devices at the current moment. The specific calculation is as follows: ; in, This represents the operational association weight between device i and device j at time t; and These represent the on / off states of the two devices at time t, directly derived from the state vector in step one. and Discrete state quantities in; and Step 1 is uniformly constructed into a fixed two-dimensional state monitoring data vector, where the first dimension represents the flow or load state and the second dimension represents the stress state. Both of these are derived from the monitoring values ​​that have been aligned and mapped in Step 1. This represents the sum of the absolute values ​​of the vector components, used to measure the degree of difference between the states of two devices; The small positive number parameters that are pre-set in the system are stored in the scheduling master station configuration table to ensure calculation stability; This represents the equipment structure matching coefficient, whose value is obtained through equipment number parsing. For example, equipment pairs within the same busbar segment have a value of 1, equipment pairs in different busbar segments but with the same voltage level have a smaller value, and equipment pairs between different voltage levels have a value of 0. This coefficient is determined by the wiring relationship field in the power grid model or equipment ledger and loaded during system initialization. Through this expression, equipment will only receive a higher association weight if it structurally belongs to the same operating domain, is simultaneously in the connected state in terms of switching status, and exhibits a consistent trend in status monitoring. Taking a main transformer operation scenario as an example, if the oil temperature ratio is 0.62 and the current ratio is 0.71 in the main transformer's state vector, while the current ratio of its high-voltage side circuit breaker is 0.65 and its state is closed, then the difference between the two is small and their switching states are consistent, thus resulting in a higher association weight. Even if a current measurement exists on the other disconnected line, its... Ultimately, the weights will be compressed, thus preventing the formation of effective associations.

[0027] Furthermore, after obtaining the operational association weights for all device pairs, corresponding decision thresholds are set based on the device type combinations to transform continuous weights into discrete association relationships. Specifically, this is expressed as: ; in, This indicates whether device i and device j have established an operational association at time t; This represents the threshold value corresponding to the combination of device types; its value is provided by the system configuration table. and Equipment type codes are derived from the type definition of each piece of equipment in the equipment ledger system, such as "main transformer body," "line circuit breaker," and "busbar bay." Different type combinations can have different thresholds. For example, the threshold between the main transformer body and its high-voltage side circuit breaker can be set to a higher value to ensure that the association is established only when the two have strong consistency in status; the threshold for adjacent switching equipment within the same busbar segment can be set to a medium value to reflect their shared participation in busbar operation. In actual calculations, if a certain equipment pair... And if the corresponding threshold is 0.70, then a decision is made. This indicates that a connection has been established between the two devices; if the other device... If the threshold is 0.60, the result is 0, and no association is established. Through this process, the relationships between all device pairs are transformed into discrete association edges.

[0028] Based on all The result is that each device node and its corresponding associated edges are organized into a state-carrying operation association graph. In this graph structure, each node corresponds to a power transmission and transformation device, and its state vector is preserved. Each edge represents a valid operational association between devices at the current time point. In implementation, an adjacency list structure can be used for storage, that is, recording the list of associated devices and corresponding associations for each device, and updating this structure at each time scale. This results in... It retains the device status data from step one and introduces the operational relationships between devices, enabling subsequent processing to directly perform status interaction and health status modeling on this graph structure.

[0029] S3. Under the constraints of the state-bearing operation association diagram, for each device, the corresponding device state vector, the cooperative state of neighboring devices in the operation channel, and the local imbalance characteristics relative to the neighboring devices are fused to generate the health status representation vector of the corresponding device.

[0030] Specifically, this step uses the state-carrying operation association diagram formed in step two. As input, each node in this graph already carries the state vector of the corresponding device. Each edge has been... Indicates whether there is a valid operational association between the devices at the current moment, and by This reflects the strength of the association. Therefore, this step is not to rebuild the device relationships, but rather... Under the given operational constraints, the state information of a single device is expanded to a health status representation composed of "the device's own state, the state of its neighboring operating channels, and local imbalance characteristics". The reason for this approach is that health changes in power transmission and transformation equipment are usually not isolated events. Equipment on the same busbar segment, the same line corridor, and the same main transformer side often exhibit group changes during load transfer, temperature rise accumulation, and operation mode switching. Simultaneously, if the state difference between a particular piece of equipment and its neighboring equipment within the same operating corridor continues to amplify, it often indicates that the equipment has deviated from its normal health deterioration. Therefore, this step, based on the operational correlation diagram in step two, simultaneously extracts two types of information: "coordinated state" and "imbalanced state," and merges them into a health state representation that can be directly used for the next step of prediction.

[0031] Furthermore, in practical implementation, the scheduling master station or edge computing node first... Read each device node and its adjacency relationship one by one. For any device i, obtain its neighbor set from the adjacency list. ,in By all satisfying The device j is composed of nodes; then the device's own state vector is read from the graph node attributes. and the state vectors of all neighboring devices Simultaneously, the operational association weights between device i and its neighboring device j are read from the edge attributes. In project deployment, this step can be accomplished through graph structure traversal, for example, by storing the graph in memory. The data is stored in a structure of "device number - node status - associated edge list". For each device, the neighbor list is extracted sequentially, and then the corresponding weight is read. Because... Since it comes from step two, the node states and edge weights used in this step are naturally on the same time scale, and the interaction between devices is completely constrained by the current running association.

[0032] For each device, a cooperative state component is first formed using the states of neighboring devices, and then an imbalance component is formed using the local differences between the device and its neighbors. The cooperative state component reflects the overall operating characteristics of the operating channel in which the device resides, while the imbalance component reflects the degree of deviation of the device from its operating channel. For device i, its local imbalance coefficient is first calculated. : ; in, This represents the local imbalance coefficient of device i at time t; This represents the set of neighboring devices of device i at time t, derived from the runtime association graph output in step two. All devices connected to device i in the middle; The weight representing the operational association between device i and device i is directly taken from the weight value saved in the graph edge in step two. and These represent the state vectors of device i and device j at time t, respectively, derived from step one and stored as node attributes. middle; This represents the sum of the absolute values ​​of the vector components, used to characterize the degree of difference between the states of two devices; This is a small positive parameter in the system configuration used to ensure the stability of the denominator. The significance of this coefficient lies in ensuring that if device i maintains a high degree of consistency with all its neighboring devices, then... Smaller; if device i, although in the same operating channel, has a state vector that continuously deviates from its neighboring devices, then It will increase. Taking the high-voltage side operation scenario of the main transformer as an example, if the state vector of the main transformer node is... The state vectors of its adjacent high-voltage side circuit breakers and bus bay nodes are respectively and Furthermore, if the corresponding edge weight is high, the difference between the primary variable and its neighboring nodes is small, resulting in... It will remain at a low level; if the temperature rise of the main transformer increases abnormally while neighboring equipment does not show synchronous changes, the state difference will increase, and correspondingly... This will also improve, thereby explicitly introducing this localized abnormality into the health profile.

[0033] Furthermore, after obtaining the local imbalance coefficient, a health status characterization of the equipment is constructed. This approach employs a fusion of three components: "self-state component," "neighborhood cooperative component," and "imbalance enhancement component," to generate the representation. The neighborhood cooperative component is obtained through weighted aggregation using associated weights, while the imbalance enhancement component applies a local imbalance coefficient to the device's own state, making devices exhibiting deviations more identifiable in the representation. Specifically: ; in, This represents the health status representation vector of device i at time t; This represents the fusion coefficient of the device's own state components. Its value is stored in the system parameter table and can be set according to the device type. For example, a larger value can be used for main transformer equipment and a medium value can be used for line equipment. This represents the imbalance enhancement coefficient, the value of which is also given in the parameter table. It is used to adjust the degree to which local imbalance enhances the health status representation; the remaining variables have the same meaning as in the previous formula. In this expression, the first term retains the equipment's own state, the second term introduces the cooperative state of neighboring equipment within the same operating channel, and the third term applies the local imbalance information to the equipment's own state again, making the equipment's health status representation more sensitive to anomalies when local anomalies occur. Taking a circuit breaker on a 220kV busbar section as an example, if its own state vector is... The weighted sum of adjacent line intervals and bus tie equipment after weight normalization is: The local imbalance coefficient is ,Pick , The resulting health status representation will retain the circuit breaker's own status while incorporating the operational influence of equipment in the same channel, and will enhance the representation of its local deviation. This approach allows the health representation to reflect both the coordinated load status of equipment in the current electrical channel and the deviation characteristics of locally abnormal equipment in the channel, making it particularly suitable for generating health representations of power transmission and transformation equipment under conditions of operating mode switching, load transfer, and accumulation of local defects.

[0034] Furthermore, after completing the above calculations, the health status representation vectors of all devices will be generated. Organized by equipment number, forming a health status representation set for power transmission and transformation equipment. In engineering implementation, The method can be the same as in step one. The same data organization method is used for storage, namely, using a mapping structure of "device number - health status representation vector", which is stored in the real-time database of the scheduling master station or the memory of the edge computing node and refreshed once at each unified time scale. At this time, the representation vector corresponding to each device is no longer just a collection of single monitoring values, but a comprehensive expression that integrates the device's own status, the collaborative status of the operating neighboring domain, and local imbalance characteristics.

[0035] S4. Based on the health status representation vector generated at the current moment, the health status representation vector at the previous moment, and the health status prediction value at the previous moment, calculate the health status prediction value at the current moment, and map the health status prediction value into discrete output labels according to the device's own historical prediction baseline.

[0036] Specifically, this step directly receives the set of health status representations of power transmission and transformation equipment output from step three. The health status representation vector of each device at time t is denoted as... Since step one has already mapped the equipment status monitoring data to a unified numerical range, and steps two and three only perform weighted combination, difference measurement, and proportional fusion, therefore... All components are dimensionless values, making them suitable for further cross-time comparisons and predictive calculations. This step generates the final health status prediction result set based on three aspects: "current health level," "rate of change at adjacent times," and "cumulative trend of continuous deterioration." Each device corresponds to one result record, which also contains continuous predicted values. and predicted output tags This design is because power transmission and transformation equipment experiences both slow, prolonged degradation and sudden, rapid increases during actual operation. For example, a main transformer may exhibit a gradually increasing health risk when operating under continuous high load, while a circuit breaker may experience a sudden change in contact state after a single opening or closing. If only the current intensity is used, the trend information may be weakened; if only differential changes are used, instantaneous fluctuations may be amplified. Therefore, this step incorporates the intensity, rate of change, and historical predicted values ​​into the same prediction formula, enabling the prediction output to simultaneously reflect "how high it is now" and "whether it is continuously deteriorating."

[0037] Furthermore, the prediction formula in this step is derived from a combination of two types of basic formulas. The first type comes from the vector norm formula in mathematics, used to compress a multidimensional representation vector into a single intensity value; the second type comes from the first-order difference and recursive smoothing concepts in time series analysis, used to characterize changes between adjacent time points and historical state memory. Based on this, this step extends the formula to the power transmission and transformation scenario: the overall intensity of the representation vector is taken as the current health level, the difference between representations between adjacent time points is taken as a short-term change term, and the predicted value obtained at the previous time point is taken as an accumulated memory term, which, together with the current intensity, affects the current result. The resulting prediction formula is not a simple superposition, but rather a representation of the health state formed by the same equipment under operational constraints. It is transmitted continuously along the time direction. The specific calculation is as follows: ; in, This represents the predicted health status of device i at time t; This represents the health status representation vector of device i at time t, which comes from the output of step three; This represents the health status representation vector of device i at the previous time scale, which comes from the previous time representation result cached by the main station or edge node; This represents the predicted health status value of device i at the previous time scale, which comes from the calculation result of the previous round in this step and is stored in the real-time cache. This represents the sum of the absolute values ​​of the vector components. This is a mathematical method for calculating the first norm, which is used to compress multidimensional representations into single-valued strengths. This represents the dimension of the health status representation vector for device i, and this value is determined during system initialization. The number of components is automatically written into the device parameter table; , , These represent the weights for current health level, short-term changes, and continuous degradation memory, respectively. These three weights are provided by the equipment category parameter table, and the main transformer can be configured with higher weights. Circuit breakers can be configured with higher performance. The line equipment can be of medium configuration; A small positive number is preset to ensure the stability of the denominator. In this formula, the first term is directly obtained from the vector norm formula, representing the average intensity of the current health status representation; the second term is derived from the first-order difference formula, representing the rate of change between adjacent time points; the third term is rewritten using the recursive smoothing idea, which multiplies the predicted value at the previous time point by the current representation intensity and then divides it by the sum of the current and previous time point representation intensities, ensuring that devices with persistently high values ​​continue to maintain a strong abnormal memory at the current time point. Since all norms, proportional terms, and historical results in the formula are dimensionless, the weighted sum of the three terms yields... It remains a dimensionless quantity, suitable for direct use as an indicator for predicting equipment health. This deductive relationship can be understood as follows: first, the vector representation obtained in step three is compressed into a comparable numerical basis using the norm; then, the difference between adjacent time points is used to form a trend enhancement term; finally, historical prediction values ​​are used to introduce continuity constraints, thereby forming a comprehensive prediction value for the current time point.

[0038] Furthermore, to ensure that the prediction results can be directly applied in engineering, this step further incorporates continuous values. This mapping represents the device-level output status. This mapping originates from threshold discrimination methods in statistics and is modified to incorporate the individual historical operating baselines of the power transmission and transformation equipment. In the actual system, each device maintains its own historical prediction sequence. The master station extracts the median reference value from the rolling window as the baseline, and then extracts the dispersion as the fluctuation reference value. This avoids deviations when different devices use a uniform threshold, ensuring that the main transformer, circuit breaker, and line each obtain their corresponding output markers based on their own historical operating background. Specifically, this is expressed as follows: ; in, This represents the predicted output flag for device i at time t, with values ​​of 0, 1, or 2, corresponding to normal, watch out, and warning, respectively. The predicted health status value obtained from the previous formula; The historical prediction baseline value of device i is obtained from the median reference value of the historical prediction result sequence of the device within the scrolling window; The historical fluctuation reference value of device i is obtained by statistical analysis of the discrete range of the prediction result sequence within the same rolling window; This represents the warning amplification factor, given in the equipment category parameter table, used to control the warning sensitivity. The logical relationship between this formula and the previous formula is very direct: the previous formula first obtains the continuous prediction value. The formula then maps the continuous prediction values ​​into a displayable, pushable, and searchable output state based on the device's historical baseline. Therefore, the two formulas constitute a "continuous prediction - discrete output" relationship.

[0039] According to the actual calculation process, in the main station program, the current representation vector of each device is first read. The representation vector of the previous time step and the predicted value at the previous time step Then read the equipment parameter table. , , , and Substituting into the previous equation, we obtain Then, read the device information based on the historical scrolling window. , and the corresponding device type Substituting into the formula, we get Finally, the "Equipment Number - Predicted Value - Output Flag" is written back to the real-time database and pushed to the monitoring interface. Taking a 220kV main transformer as an example, let its current health status representation vector be... The representation vector at the previous time step is The predicted value at the previous time step was Then there is , , If the device represents a dimension The parameter table provides , , , Then the first term is approximately The second term is approximately The third item is approximately To obtain the current predicted value Let's further assume that the historical forecast baseline of this main transformer within the most recent rolling window... Fluctuation reference value Equipment category parameters Then there is ,because Therefore, we get This indicates that the system has entered an early warning state. The calculation result can be directly displayed on the monitoring platform as a current health prediction value of 0.391 for the main transformer, corresponding to a high warning level. Maintenance personnel can further trace back to the health status representation components in step three when viewing equipment details, determining whether the current anomaly originates from a coordinated load increase or a localized imbalance. Through the above processing, the health status representation set obtained in step three... This is transformed into the final set of health status prediction results for power transmission and transformation equipment. Each result record includes the continuous predicted values ​​for the corresponding device. With predicted output tags This result preserves the current health strength while also incorporating the rate of change and continuous degradation memory at adjacent moments. This allows the health evolution of the main transformer, circuit breaker, and line under different operating conditions to be expressed continuously, stably, and with scenario-specific characteristics. Furthermore, it generates engineering results that can be directly used for monitoring display and maintenance prompts through discrete output tags.

[0040] This invention also provides a health status prediction system for power transmission and transformation equipment, the system comprising: The data construction module is used to acquire equipment status monitoring data and switch status data of power transmission and transformation equipment, and construct power transmission and transformation equipment status data; wherein, the power transmission and transformation equipment status data includes a status vector of all equipment at the same point in time, indexed and organized according to equipment identifier; the status vector includes equipment status monitoring data and the switch status of the corresponding equipment; The association identification module is used to obtain the switch status of all devices and the corresponding device structure information, combine the corresponding device status monitoring data, construct the operation association weight of any two devices at the same time, and compare the operation association weight of all device pairs with a predefined judgment threshold to determine whether any two devices establish an operation association at the same time. If the weight of the current device pair's operational association is greater than the predefined judgment threshold, then an operational association is established; otherwise, it is not established. Based on all operational associations, each device pair forms a device node and its corresponding associated edge, which are organized into a state-carrying operational association graph. The characterization generation module is used to, under the constraints of the state-bearing operation association graph, fuse the corresponding device state vector, the cooperative state of neighboring devices in the operation channel, and the local imbalance features relative to the neighboring devices for each device to generate the health status characterization vector of the corresponding device. The prediction output module is used to calculate the predicted health status value at the current moment based on the health status representation vector generated at the current moment, the health status representation vector at the previous moment, and the predicted health status value at the previous moment, and to map the predicted health status value to discrete output labels according to the device's own historical prediction baseline.

[0041] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0042] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of apparatuses or units may be electrical, mechanical, or other forms.

[0043] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0044] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for predicting the health status of power transmission and transformation equipment, characterized in that, The method includes: S1. Obtain equipment status monitoring data and switch status data of power transmission and transformation equipment, and construct power transmission and transformation equipment status data; wherein, the power transmission and transformation equipment status data includes a status vector of all equipment at the same point in time, indexed and organized according to equipment identifier; the status vector includes equipment status monitoring data and the switch status of the corresponding equipment; S2. Obtain the switch status of all devices and the corresponding device structure information, combine the corresponding device status monitoring data, construct the operation association weight of any two devices at the same time; and based on the operation association weight of all device pairs, compare it with a predefined judgment threshold to determine whether any two devices establish an operation association at the same time. If the weight of the current device pair's operational association is greater than the predefined judgment threshold, then an operational association is established; otherwise, it is not established. Based on all operational associations, each device pair forms a device node and its corresponding associated edge, which are organized into a state-carrying operational association graph. S3. Under the constraints of the state-bearing operation association diagram, for each device, the corresponding device state vector, the cooperative state of neighboring devices in the operation channel, and the local imbalance characteristics relative to the neighboring devices are fused to generate the health status representation vector of the corresponding device. S4. Based on the health status representation vector generated at the current moment, the health status representation vector at the previous moment, and the health status prediction value at the previous moment, calculate the health status prediction value at the current moment, and map the health status prediction value into discrete output labels according to the device's own historical prediction baseline.

2. The method for predicting the health status of power transmission and transformation equipment according to claim 1, characterized in that, The power transmission and transformation equipment includes: transformers, transmission lines, and circuit breakers; The equipment status monitoring data always includes two components: the first component represents the current flow or load status of the equipment at the current moment, and the second component represents the stress status of the equipment at the current moment.

3. The method for predicting the health status of power transmission and transformation equipment according to claim 1, characterized in that, The equipment structure information includes the station, voltage level, and bus section or bay information carried in the equipment number.

4. The method for predicting the health status of power transmission and transformation equipment according to claim 3, characterized in that, In step S2, the specific method for identifying valid operational associations and generating a dynamic operational association graph is as follows: First, using the equipment structure information, all equipment is grouped into a candidate operation set; Then, within each candidate run set, the run association weight, which measures the association strength, is calculated based on the on / off status of any two devices, the degree of difference between their status monitoring data vectors, and the structure matching coefficient predefined by the device structure relationship. Finally, the calculated operational association weights of all device pairs are compared with the judgment thresholds corresponding to the device type combinations. Device pairs whose weight values ​​reach or exceed the thresholds are judged to have operational associations, and the operational association graph is constructed with devices as nodes and association relationships as edges.

5. The method for predicting the health status of power transmission and transformation equipment according to claim 1, characterized in that, The collaborative state of neighboring devices within the operating channel is generated based on the operating association weights of adjacent devices in the state-carrying operating association diagram. The local imbalance feature relative to the neighboring device is generated based on the local imbalance coefficient and the state vector of the corresponding device.

6. The method for predicting the health status of power transmission and transformation equipment according to claim 2, characterized in that, When the power transmission and transformation equipment is a transformer, its condition monitoring data includes load current and oil temperature; when the power transmission and transformation equipment is a circuit breaker, its condition monitoring data includes current carrying capacity and operation count increment; when the power transmission and transformation equipment is a transmission line, its condition monitoring data includes line current and bus side voltage deviation or line temperature rise.

7. The method for predicting the health status of power transmission and transformation equipment according to claim 5, characterized in that, The local imbalance coefficient is used to reflect the degree of deviation of the equipment from its operating channel.

8. The method for predicting the health status of power transmission and transformation equipment according to claim 1, characterized in that, In step S4, the method for calculating the continuous predicted value of the health status is as follows: The continuous predicted value of health status is obtained by weighted summing of the current health level term based on the average intensity of the current health status representation vector, the short-term change term based on the rate of change of the health status representation vector between the current and previous moments, and the continuous degradation memory term based on the historical predicted value of the current influence.

9. The method for predicting the health status of power transmission and transformation equipment according to claim 8, characterized in that, In step S4, the historical prediction baseline is determined by the median reference value of the historical prediction value sequence of the corresponding device within the rolling time window, and the historical fluctuation reference value is obtained by statistically analyzing the dispersion of the prediction value sequence within the same rolling window. The current continuous forecast value is then compared with a combination of the historical forecast baseline and historical fluctuation reference value, and mapped to discrete output labels including normal, attention and warning.

10. A method for predicting the health status of power transmission and transformation equipment, characterized in that, The system includes: The data construction module is used to acquire equipment status monitoring data and switch status data of power transmission and transformation equipment, and construct power transmission and transformation equipment status data; wherein, the power transmission and transformation equipment status data includes a status vector of all equipment at the same point in time, indexed and organized according to equipment identifier; the status vector includes equipment status monitoring data and the switch status of the corresponding equipment; The association identification module is used to obtain the switch status of all devices and the corresponding device structure information, combine the corresponding device status monitoring data, construct the operation association weight of any two devices at the same time, and compare the operation association weight of all device pairs with a predefined judgment threshold to determine whether any two devices establish an operation association at the same time. If the weight of the current device pair's operational association is greater than the predefined judgment threshold, then an operational association is established; otherwise, it is not established. Based on all operational associations, each device pair forms a device node and its corresponding associated edge, which are organized into a state-carrying operational association graph. The characterization generation module is used to, under the constraints of the state-bearing operation association graph, fuse the corresponding device state vector, the cooperative state of neighboring devices in the operation channel, and the local imbalance features relative to the neighboring devices for each device to generate the health status characterization vector of the corresponding device. The prediction output module is used to calculate the predicted health status value at the current moment based on the health status representation vector generated at the current moment, the health status representation vector at the previous moment, and the predicted health status value at the previous moment, and to map the predicted health status value to discrete output labels according to the device's own historical prediction baseline.

Images