Emergency assessment method and device for distribution transformer ultimate survival life

Abstract

The application relates to a power distribution transformer limit survival life emergency evaluation method and device. The method comprises the following steps: for each disaster scene, extracting the equipment equivalent life loss rate of the power distribution transformer under each disaster scene from a knowledge base, wherein the knowledge base contains a mapping relationship between real-time disaster intensity and the equipment equivalent life loss rate of the power distribution transformer. In the case that the power distribution transformer is in at least two target disaster scenes at the same time, based on the nonlinear acceleration degradation effect in the case of concurrent disasters, the equipment equivalent life loss rate corresponding to each target disaster scene is converted into a predicted damage trajectory of the power distribution transformer and a dynamic failure threshold configured for each target disaster scene. Based on the time point corresponding to the intersection of the predicted damage trajectory and the dynamic failure threshold, the limit survival life of the power distribution transformer is determined. The method makes the prediction result be doubly constrained by the damage evolution process and the dynamic failure condition, and improves the accuracy of determining the limit survival life.

Description

Technical Field

[0001] This application relates to the field of transformer life assessment technology, and in particular to an emergency assessment method and apparatus for the ultimate survival life of distribution transformers. Background Technology

[0002] In recent years, extreme weather events have become more frequent, severely impacting the safe operation and survival limits of distribution transformers. Transformers face extremely complex physical damage processes in harsh environments, and the damage mechanisms caused by different types of disasters differ fundamentally. Against this backdrop, it is necessary to construct a mapping relationship between disaster conditions and the equivalent lifespan reduction of distribution transformers.

[0003] Traditional solutions primarily address the impact of high temperature and high humidity environments on transformer lifespan. Furthermore, current assessment systems rely on static empirical formulas to determine the remaining lifespan of distribution transformers, which still suffers from inaccurate predictions of remaining lifespan. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for emergency assessment of the ultimate survival life of distribution transformers, which can solve the above-mentioned technical problems.

[0005] Firstly, this application provides an emergency assessment method for the ultimate survival life of a distribution transformer. The method includes: The critical disaster intensity of distribution transformers under multiple disaster scenarios was derived by analyzing the evolution mechanism of multiple disasters. Where the real-time disaster intensity experienced by the distribution transformer is greater than the critical disaster intensity, the distribution transformer experiences a shortened lifespan. For each disaster scenario, the equivalent life loss rate of the distribution transformer under each disaster scenario is extracted from the knowledge base. The knowledge base contains the mapping relationship between the real-time disaster intensity of the distribution transformer and the equivalent life loss rate of the equipment. When the distribution transformer is simultaneously subjected to at least two target disaster scenarios, based on the nonlinear accelerated degradation effect during multiple disasters, the equipment equivalent life loss rate corresponding to each of the target disaster scenarios is transformed into the predicted damage trajectory of the distribution transformer and the dynamic failure threshold configured for each of the target disaster scenarios. The ultimate lifespan of the distribution transformer is determined based on the time corresponding to the intersection of the predicted damage trajectory and the dynamic failure threshold.

[0006] Secondly, this application also provides an emergency assessment device for the ultimate survival life of a distribution transformer. The device includes: The analysis module is used to deduce the critical disaster intensity of the distribution transformer under multiple disaster scenarios through multi-hazard evolution mechanism analysis; wherein, when the real-time disaster intensity experienced by the distribution transformer is greater than the critical disaster intensity, the distribution transformer exhibits a shortened lifespan. An extraction module is used to extract, for each disaster scenario, the equivalent life loss rate of the distribution transformer under each disaster scenario from a knowledge base. The knowledge base contains the mapping relationship between the real-time disaster intensity of the distribution transformer and the equivalent life loss rate of the equipment. The conversion module is used to convert the equipment equivalent life loss rate corresponding to each of the target disaster scenarios into the predicted damage trajectory of the distribution transformer and the dynamic failure threshold configured for each of the target disaster scenarios, based on the nonlinear accelerated degradation effect when multiple disasters occur simultaneously, when the distribution transformer is simultaneously under at least two target disaster scenarios. The processing module is used to determine the ultimate lifespan of the distribution transformer based on the time corresponding to the intersection of the predicted damage trajectory and the dynamic failure threshold.

[0007] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the above method steps.

[0008] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the above method steps.

[0009] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the above method steps.

[0010] The aforementioned emergency assessment methods, devices, computer equipment, storage media, and computer programs for the ultimate survival life of distribution transformers derive the critical disaster intensities of distribution transformers under various disaster scenarios through multi-hazard evolution mechanism analysis. This overcomes the shortcomings of traditional single fixed thresholds that cannot distinguish disaster types, enabling the critical disaster intensity to be scenario-adaptive. This establishes differentiated safety boundaries for different disasters, avoiding premature or delayed warnings due to threshold mismatch.

[0011] In cases where the real-time disaster intensity experienced by the distribution transformer exceeds the critical disaster intensity, the transformer's lifespan is shortened. Isolating the damaged and non-damaged operating conditions helps filter out invalid disturbance data below the critical value and improves the accuracy of subsequent data processing.

[0012] For each disaster scenario, the equivalent life loss rate of the distribution transformer under each disaster scenario is extracted from the knowledge base. The knowledge base contains the mapping relationship between the real-time disaster intensity of the distribution transformer and the equivalent life loss rate. By utilizing the real-time input and output mapping structure of the knowledge base, dynamic differential calculation of the loss rate is realized, eliminating the need for complex online calculations. This ensures the scenario accuracy of the loss rate value and meets the rapid response requirements of the engineering site.

[0013] When a distribution transformer is simultaneously subjected to at least two target disaster scenarios, based on the nonlinear accelerated degradation effect under multiple concurrent disasters, the equivalent life loss rate of the equipment corresponding to each target disaster scenario is transformed into the predicted damage trajectory of the distribution transformer and a dynamic failure threshold configured for each target disaster scenario. Firstly, the introduction of the nonlinear accelerated degradation effect overcomes the deficiency of the traditional linear superposition model, which severely underestimates the synergistic effect of multiple disasters. Secondly, the dynamic failure threshold replaces the fixed threshold, reflecting the real-time changes in equipment failure conditions under different disaster combinations, making the damage trajectory closer to the real physical process.

[0014] The ultimate lifespan of a distribution transformer is determined based on the time corresponding to the intersection of the predicted damage trajectory and the dynamic failure threshold. This method transforms the determination of the lifespan end into a trajectory-threshold geometric intersection problem, ensuring that the prediction results are simultaneously constrained by both the damage evolution process and the dynamic failure conditions, rather than relying on a pre-set lifespan end, thus improving the accuracy of determining the ultimate lifespan. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating an emergency assessment method for the ultimate survival life of a distribution transformer in one embodiment. Figure 2 This is a flowchart illustrating a method for determining the critical disaster intensity in one embodiment; Figure 3 This is a flowchart illustrating a method for determining a prediction model for the equivalent life depreciation time of a device in one embodiment. Figure 4 This is a flowchart illustrating a method for determining a predicted damage trajectory and a dynamic failure threshold in one embodiment. Figure 5 In one embodiment, the first A schematic diagram of the critical intensity surface of a disaster type; Figure 6 This is a structural block diagram of an emergency assessment device for the ultimate survival life of a distribution transformer in one embodiment; Figure 7 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0017] In one embodiment, such as Figure 1 As shown, an emergency assessment method for the ultimate survival life of a distribution transformer is provided, including: S101, through multi-hazard evolution mechanism analysis, derives the critical disaster intensity of distribution transformers under multiple disaster scenarios.

[0018] Among them, when the real-time disaster intensity of the distribution transformer is greater than the critical disaster intensity, the distribution transformer will experience a shortened lifespan.

[0019] Among them, the multi-hazard evolution mechanism refers to the inherent law that when two or more different types of disasters (such as earthquakes, rainstorms, lightning, wind disasters, high temperatures, etc.) occur concurrently in time or superimpose in space, the disasters drive the nonlinear changes in the equipment damage state through physical field coupling (such as stress field, thermal field, electromagnetic field) or failure chain reaction.

[0020] A distribution transformer is a static electromagnetic induction device that steps down the voltage of a power distribution network to the voltage on the user side of a low-voltage distribution network. Distribution transformers may experience a shortened lifespan due to disasters.

[0021] The phenomenon of lifespan reduction refers to the physical process in which, after a distribution transformer is subjected to an actual disaster exceeding the critical disaster intensity, its internal insulation materials (cellulose paper, insulating oil) or mechanical structure (winding clamping force, core clamping parts) undergo cumulative and irreversible degradation, resulting in an accelerated reduction of the remaining technical life of the equipment relative to the normal aging curve.

[0022] A disaster scenario refers to the types of extreme external environmental events and their combinations of conditions that threaten the lifespan of distribution transformers. It is an independent disaster category with a clear physical mechanism, quantifiable intensity indicators, and specific damage patterns. Disaster scenarios can include: freezing rain scenarios, icing scenarios, strong wind scenarios, sandstorm scenarios, and wildfire scenarios, etc.

[0023] Critical disaster intensity can be represented as the minimum amount of hazard that, under a specific disaster scenario and given the material, structure, service life, and current operating conditions of a distribution transformer, can cause irreversible degradation of the equipment's insulation or mechanical properties. For example, in a freezing rain scenario, critical disaster intensity can represent the critical intensity of the freezing rain disaster (freezing rain intensity). Similarly, in an icing scenario, critical disaster intensity can represent the critical intensity of the icing disaster (ice thickness).

[0024] Real-time disaster intensity can be represented by the instantaneous or short-term effective value of the disaster's impact, obtained at the current monitoring moment by sensors (such as accelerometers, voltage transformers, temperature sensors, and humidity meters) deployed at or around the distribution transformer. Specifically, real-time disaster intensity can be represented by the wind speed, freezing rain intensity, dust concentration, ice thickness, and wildfire heat radiation intensity experienced by the distribution transformer at the current monitoring moment, obtained directly from the actual meteorological station measurement system.

[0025] The critical disaster intensity can be determined by calculation models. Specifically, S101 above includes: for typical disasters such as freezing rain, icing, strong wind, sandstorm and wildfire in high-altitude and cold regions, based on the dominant action path of each disaster on the additional life loss of transformers, dynamic disaster critical intensity calculation models for freezing rain, icing, strong wind, sandstorm and wildfire are constructed respectively. The calculation models characterize the minimum disaster intensity threshold (critical disaster intensity) that causes additional life loss of distribution transformers.

[0026] There may be various disaster scenarios, and the corresponding critical disaster intensity varies depending on the scenario. Specifically, S101 above includes: for each disaster scenario, obtaining the duration of the disaster and environmental variables of the distribution transformer under the disaster scenario through multi-hazard evolution mechanisms. The disaster scenarios include freezing rain, icing, strong wind, sandstorm, and wildfire scenarios, and the environmental variables include ambient temperature, ambient wind speed, air density, relative humidity, and distance from the fire source. Based on the duration of the disaster and the environmental variables, the critical disaster intensity of the distribution transformer under the disaster scenario is determined.

[0027] It overcomes the shortcomings of traditional single fixed thresholds that cannot distinguish disaster types, and enables the critical disaster intensity to be scene-adaptive, thereby establishing differentiated safety boundaries for different disasters and avoiding premature or late warnings due to threshold mismatch.

[0028] S102, for each disaster scenario, extracts the equivalent life loss rate of the distribution transformer under each disaster scenario from the knowledge base.

[0029] The knowledge base refers to a pre-built, structured data storage and reasoning system. It contains the mapping relationship between the real-time disaster intensity of distribution transformers and the equipment's equivalent lifespan loss rate.

[0030] Specifically, the knowledge base stores a multi-dimensional mapping relationship between the real-time disaster intensity and the equipment's equivalent life loss rate under different disaster scenarios for distribution transformers.

[0031] The equivalent life loss rate of equipment represents the ratio of the life loss of a distribution transformer caused by a specific disaster intensity per unit time (or a single disaster event) to the normal aging loss of the transformer under standard rated operating conditions.

[0032] There are several ways to extract equipment life loss rates from a knowledge base. Specifically, the difference between real-time disaster intensity and critical disaster intensity, as well as the disaster event indicator variable and sampling time interval corresponding to the real-time disaster intensity, can be input into the knowledge base to output the equivalent equipment life loss rate of the distribution transformer under each disaster scenario.

[0033] The knowledge base can be an equivalent lifespan depreciation time prediction model. Specifically, the equipment lifespan depreciation rate is extracted from the knowledge base, including: For each disaster scenario, the disaster event indicator variable corresponding to the real-time disaster intensity and the sampling time interval are input into the equivalent life loss time prediction model, which outputs the cumulative equivalent life loss time under the disaster. Based on the cumulative equivalent life loss time under the disaster, the equipment equivalent life loss rate of the distribution transformer under the disaster scenario is obtained.

[0034] By leveraging the real-time mapping structure of the knowledge base's input and output, dynamic differential calculation of the loss rate is achieved without the need for complex online calculations. This ensures the accuracy of the loss rate values ​​in different scenarios while meeting the rapid response requirements of engineering sites.

[0035] S103, when the distribution transformer is simultaneously under at least two target disaster scenarios, based on the nonlinear accelerated degradation effect when multiple disasters occur concurrently, the equivalent life loss rate of the equipment corresponding to each target disaster scenario is transformed into the predicted damage trajectory of the distribution transformer and the dynamic failure threshold configured for each target disaster scenario.

[0036] Among them, the target disaster scenario refers to those disaster scenarios that are currently being monitored and confirmed in real time by on-site sensors, and whose real-time disaster intensity has exceeded their respective critical disaster intensity—that is, the set of disaster types that are actually occurring and have triggered lifespan reduction. Specifically, the target disaster scenario can be obtained from the meteorological station measurement system, which can be used to determine the target disaster scenario where the distribution transformer is located at that moment.

[0037] The nonlinear accelerated degradation effect under multiple concurrent disasters indicates that when two or more target disaster scenarios act on the same distribution transformer at the same time, the physical damage (mechanical stress, electrical stress, thermal stress) caused by each disaster is coupled and mutually promoted within the equipment, resulting in the total equivalent damage rate being greater than the sum of the damage rates when each disaster acts alone.

[0038] The predicted damage trajectory is a continuous function curve of the equivalent lifetime loss as a function of the predicted time, generated by integrating the combined loss rate of each target disaster scenario over time or by sequentially accumulating disaster events, based on the nonlinear accelerated degradation effect.

[0039] The dynamic failure threshold represents the maximum cumulative equivalent life loss that a distribution transformer can withstand, recalculated in real time for a specific combination of concurrent disaster scenarios. The dynamic failure threshold is not a fixed constant, but rather a threshold that is dynamically updated based on the type of disaster combination, the current health status of the equipment, and the passage of time.

[0040] Specifically, S103 includes: based on the nonlinear accelerated degradation effect under multiple concurrent disasters, converting the target real-time disaster intensity of the distribution transformer in at least two target disaster scenarios into a coupled degradation amplification coefficient for multiple concurrent disasters; determining the target comprehensive life loss rate under multiple extreme disaster coupling based on the equivalent life loss rate of the equipment corresponding to each target disaster scenario and the coupled degradation amplification coefficient; converting the target comprehensive life loss rate into the predicted damage trajectory of the distribution transformer based on the neural operator of the trained transformer health status prediction model; and converting the target comprehensive life loss rate into a dynamic failure threshold configured for each target disaster scenario based on the threshold-related parameters of the trained transformer health status prediction model.

[0041] Specifically, S103 includes: when the distribution transformer is simultaneously under at least two target disaster scenarios, based on the nonlinear accelerated degradation effect when multiple disasters occur concurrently, inputting the equivalent life loss rate of the equipment corresponding to each target disaster scenario into the transformer health status prediction model, and outputting the predicted damage trajectory of the distribution transformer and the dynamic failure threshold configured for each target disaster scenario.

[0042] First, the introduction of nonlinear accelerated degradation effect overcomes the deficiency of traditional linear superposition model in severely underestimating the synergistic effect of multiple disasters. Second, the dynamic failure threshold replaces the fixed threshold, which can reflect the real-time changes in equipment failure conditions under different disaster combinations, making the damage trajectory closer to the real physical process.

[0043] S104. Based on the time corresponding to the intersection of the predicted damage trajectory and the dynamic failure threshold, determine the ultimate lifespan of the distribution transformer.

[0044] The time corresponding to the intersection of the predicted damage trajectory and the dynamic failure threshold represents the moment when the predicted damage trajectory curve and the dynamic failure threshold curve first intersect (or become tangent) in the same time coordinate system. This time signifies that the cumulative damage to the equipment has just reached the maximum allowable tolerance limit under the current disaster combination.

[0045] Ultimate survival life (ESW) is the longest predicted duration a distribution transformer can sustain under the combined effects of multiple disasters, starting from the current monitoring time, until its cumulative equivalent life damage first reaches the maximum allowable value of the dynamic failure threshold. For example, ESW can provide maintenance personnel with a clear time window to schedule emergency repairs, load transfers, or emergency evacuations.

[0046] Specifically, when the predicted damage trajectory reaches the dynamic failure threshold, the ultimate survival life of the distribution transformer under multiple concurrent disasters is output. The ultimate survival life can be used to arrange emergency maintenance, load transfer, or emergency evacuation.

[0047] The aforementioned emergency assessment method for the ultimate survival life of distribution transformers derives the critical disaster intensity of distribution transformers under multiple disaster scenarios through multi-hazard evolution mechanism analysis. This overcomes the shortcomings of traditional single fixed thresholds that cannot distinguish disaster types, enabling the critical disaster intensity to be scenario-adaptive, thereby establishing differentiated safety boundaries for different disasters and avoiding premature or late warnings due to threshold mismatch.

[0048] In cases where the real-time disaster intensity experienced by the distribution transformer exceeds the critical disaster intensity, the transformer's lifespan is shortened. Isolating the damaged and non-damaged operating conditions helps filter out invalid disturbance data below the critical value and improves the accuracy of subsequent data processing.

[0049] For each disaster scenario, the equivalent life loss rate of the distribution transformer under each disaster scenario is extracted from the knowledge base. The knowledge base contains the mapping relationship between the real-time disaster intensity of the distribution transformer and the equivalent life loss rate. By utilizing the real-time input and output mapping structure of the knowledge base, dynamic differential calculation of the loss rate is realized, eliminating the need for complex online calculations. This ensures the scenario accuracy of the loss rate value and meets the rapid response requirements of the engineering site.

[0050] When a distribution transformer is simultaneously subjected to at least two target disaster scenarios, based on the nonlinear accelerated degradation effect under multiple concurrent disasters, the equivalent life loss rate of the equipment corresponding to each target disaster scenario is transformed into the predicted damage trajectory of the distribution transformer and a dynamic failure threshold configured for each target disaster scenario. Firstly, the introduction of the nonlinear accelerated degradation effect overcomes the deficiency of the traditional linear superposition model, which severely underestimates the synergistic effect of multiple disasters. Secondly, the dynamic failure threshold replaces the fixed threshold, reflecting the real-time changes in equipment failure conditions under different disaster combinations, making the damage trajectory closer to the real physical process.

[0051] The ultimate lifespan of a distribution transformer is determined based on the time corresponding to the intersection of the predicted damage trajectory and the dynamic failure threshold. This method transforms the determination of the lifespan end into a trajectory-threshold geometric intersection problem, ensuring that the prediction results are simultaneously constrained by both the damage evolution process and the dynamic failure conditions, rather than relying on a pre-set lifespan end, thus improving the accuracy of determining the ultimate lifespan.

[0052] In one embodiment, such as Figure 2 The flowchart shown illustrates the method for determining the critical disaster intensity. Through multi-hazard evolution mechanism analysis, the critical disaster intensity of the distribution transformer under various disaster scenarios is derived, including: S201, for each disaster scenario, obtains the duration of the disaster and the environmental variables of the disaster for the distribution transformer under the disaster scenario through the multi-hazard evolution mechanism.

[0053] The disaster scenarios include freezing rain, icing, strong winds, sandstorms, and wildfires. The environmental variables include ambient temperature, ambient wind speed, air density, relative humidity, and distance from the fire source.

[0054] Among them, the duration of disaster refers to the duration for which a specific disaster affects the distribution transformer.

[0055] Disaster environmental variables can represent key environmental variables in a disaster scenario.

[0056] For example, in the freezing rain scenario, the duration of the disaster is the duration of the freezing rain, and the environmental variable is the ambient temperature. In the icing scenario, the duration of the disaster is the duration of the icing, and the environmental variable is the ambient wind speed. In the strong wind scenario, the duration of the disaster is the duration of the strong wind, and the environmental variable is the air density. In the sandstorm scenario, the duration of the disaster is the duration of the sandstorm, and the environmental variable is the relative humidity.

[0057] For example, in the wildfire scenario, the duration of the disaster is the duration of the wildfire, and the environmental variable of the disaster is the distance from the fire source.

[0058] S202, based on the duration of the disaster and environmental variables of the disaster, determines the critical disaster intensity of the distribution transformer under the disaster scenario.

[0059] Specifically, based on the duration of the disaster and environmental variables related to the disaster, a critical intensity calculation model for the disaster is constructed, and the critical disaster intensity corresponding to the critical intensity calculation model is obtained.

[0060] For example, freezing rain weakens the external insulation performance through low-temperature wet deposition and thawing, leading to electrical aging and reduced lifespan of transformers. The critical intensity calculation model for this disaster can be a freezing rain disaster critical intensity calculation model, and the expression for the freezing rain disaster critical intensity calculation model is as follows: ; in, Indicates the duration of freezing rain Ambient temperature The critical freezing rain intensity (critical disaster intensity) below; The leakage current threshold is the threshold value for electrical aging caused by moisture in the external insulation. The operating voltage that the external insulation withstands; The coefficient by which freezing rain intensity is converted into the conductivity of a conductive water film; This is the freezing rain temperature amplification factor; The characteristic temperature at which a conductive water film forms or a freeze-thaw state occurs due to freezing rain; Width of the temperature-sensitive range for freezing rain; This is the baseline moisture duration corresponding to the leakage current threshold.

[0061] For example, icing increases thermal resistance and raises the temperature of hot spots, leading to reduced insulation lifespan due to thermal aging. The critical intensity calculation model for icing disasters can be a critical intensity calculation model for icing disasters, and the expression for this model is as follows: ; in, Indicates the actual duration of icing. Wind speed The critical icing thickness (critical disaster intensity) below. This is the coefficient for the increase in thermal resistance due to icing. A This represents the effective heat dissipation area of ​​the transformer. Equivalent heat transfer coefficient under non-icing conditions; Wind speed heat transfer index; The equivalent load loss coefficient under rated load; Indicates the duration of icing. The permissible hot spot temperature critical value; This refers to the winding hot spot temperature under normal operating conditions. This is a function of the thermal aging rate; This represents the inverse function of the thermal aging rate function; The threshold temperature for icing hotspots. for The reference duration at the specified temperature.

[0062] The duration of icing is as follows: Permissible hot spot temperature critical value The expression is as follows: ; For example, strong winds cause mechanical fatigue by impacting the external connection components of a transformer through wind pressure and continuous vibration. The critical intensity calculation model for a disaster can be a critical intensity calculation model for strong wind disasters. The expression for the critical intensity calculation model for strong wind disasters is as follows: ; in, Indicates the duration of strong winds air density Critical wind speed (critical disaster intensity) below; The allowable wind-induced stress amplitude is determined by structural strength verification, wind-resistant design, or vibration monitoring thresholds; This is the coefficient for the conversion of wind load into structural stress amplitude; This refers to the drag coefficient; The windward area of ​​the transformer or its auxiliary structures; To allow for the reference strong wind duration corresponding to the stress amplitude; This is the fatigue damage index.

[0063] For example, sandstorms reduce the external insulation performance of transformers through pollution deposition. The critical intensity calculation model for the disaster can be a sandstorm disaster critical intensity calculation model, and the expression for the sandstorm disaster critical intensity calculation model is as follows: ; in, Indicates the duration of the sandstorm relative humidity H The critical dust concentration below; The equivalent salt density threshold represents the critical pollution level at which external insulation contamination reaches an electrical risk state. This is the coefficient for the conversion of dust deposition into equivalent salinity; This represents the dust deposition coefficient. Humidity amplification factor; Humidity sensitivity coefficient; The characteristic of a dirt layer is that its moisture absorption and conductivity are significantly enhanced by relative humidity.

[0064] For example, wildfires increase local thermal stress in transformers through thermal radiation, triggering thermal aging. The critical intensity calculation model for a disaster can be a wildfire disaster critical intensity calculation model, and the expression for the wildfire disaster critical intensity calculation model is as follows: ; in, Indicates the duration of the wildfire Distance between the fire source and the transformer The critical wildfire thermal radiation intensity below; The equivalent heat transfer coefficient of the transformer surface; This is the coefficient of thermal radiation attenuation with distance; The absorption rate of thermal radiation on the outer surface of the transformer; The duration of the wildfire was The critical value of hotspot temperature at that time; This refers to the winding hot spot temperature under normal operating conditions. This refers to the temperature threshold for hot spots where wildfires cause thermal aging. Its corresponding baseline duration.

[0065] The duration of the wildfire was as follows: Hot spot temperature critical value The expression is as follows: ; Based on this, the duration of the disaster and environmental variables can be used as input variables for the surface, and the corresponding critical disaster intensity can be used as the output variable for the surface, thus constructing the first... The critical intensity surface of a disaster type can be represented as: ; in, Indicates the first Critical disaster intensity; For the first Disaster-like critical intensity model; Duration of the disaster; For disaster environmental variables; This represents the model parameter set consisting of equipment structural parameters, operating status parameters, and threshold parameters. These represent the upper and lower bounds of the effective value range of the two input variables, determined based on the disaster monitoring scope and historical disaster samples, respectively.

[0066] In this embodiment, firstly, the general disaster intensity is decomposed into a time dimension (duration) and a physical dimension (five environmental variables). This makes the derivation of critical intensity no longer dependent on a single transient extreme value, but comprehensively considers the cumulative effect of the disaster and the influence of the environmental medium, significantly improving the ability of the critical value to represent actual complex working conditions. Secondly, for typical extreme disasters such as freezing rain, icing, strong winds, sandstorms, and wildfires in high-altitude and cold regions, the critical intensity of distribution transformer life loss under different disasters is derived through multi-hazard evolution mechanism analysis. A dynamic threshold surface of disaster critical intensity covering multi-dimensional environmental conditions is constructed, which facilitates the subsequent screening of extreme weather data that cause damage to the life of distribution transformers, thereby improving the prediction accuracy of the ultimate survival life of distribution transformers.

[0067] In one embodiment, the equivalent life loss rate of the distribution transformer under various disaster scenarios is extracted from the knowledge base, including: The difference between real-time disaster intensity and critical disaster intensity, as well as the disaster event indicator variable and sampling time interval corresponding to the real-time disaster intensity, are input into the knowledge base to output the equipment equivalent life loss rate of the distribution transformer under each disaster scenario.

[0068] The difference between the real-time disaster intensity and the critical disaster intensity represents the numerical difference or relative ratio between the two. Based on this difference, the disaster event indicator variable can be determined to be 1 or 0.

[0069] Disaster event indicator variables indicate whether a specific type of disaster is ongoing or newly occurring at a certain sampling time or sampling interval, serving as a boundary marker distinguishing between periods during and between disasters.

[0070] The sampling time interval represents the time step between two adjacent sensor data acquisitions during continuous monitoring. The sampling time interval determines the discretization granularity of the loss rate output on the time axis.

[0071] Specifically, based on real-time disaster intensity With critical disaster intensity The difference between the values ​​is used to determine the moments in real-time disaster intensity that are greater than or equal to the critical disaster intensity. The disaster event indicator variable corresponding to this moment is set to 1, and vice versa. This determines the disaster event indicator variable corresponding to the real-time disaster intensity. Furthermore, based on the real-time disaster intensity Disaster event indicator variables and sampling time interval The input is the knowledge base, and the output is the equivalent lifespan depreciation time of the device. The specific mapping relationship expression of the knowledge base is as follows: ; in, Indicates the real-time disaster intensity. This indicates the equipment's equivalent life loss time, representing the equipment's equivalent life loss rate for distribution transformers in strong wind, freezing rain, sandstorm, icing, and wildfire scenarios, respectively.

[0072] It should be noted that the construction of the knowledge base involves the calculation of low-rank ordinary differential equations. The Mamba model with low-rank ordinary differential equations is introduced for knowledge base construction. Compared with the conventional method of using pure data-driven recurrent neural networks for lifetime extrapolation, the former has a smaller error in predicting the ultimate survival time.

[0073] In this embodiment, firstly, by introducing multi-dimensional input combinations, the retrieval criteria of the knowledge base are expanded from a single intensity value to a triple criterion of intensity, state, and duration, significantly improving the conditional matching accuracy of the damage rate value and avoiding fuzzy matching errors caused by relying solely on intensity values ​​for table lookup. Secondly, the knowledge base can output damage rates in segments according to event boundaries, avoiding the miscalculation of non-disaster periods into additional damage, thereby eliminating the cumulative bias effect in cumulative lifetime prediction. Thirdly, by introducing a sampling time interval, the damage rate output can adapt to the sampling frequency differences of different sensors, ensuring that the damage rate output by the knowledge base is integral and summable in the time dimension, providing a time alignment benchmark for the continuous integration of subsequent damage trajectory prediction.

[0074] In one embodiment, such as Figure 3 The flowchart shown illustrates the method for determining the equipment's equivalent life depreciation time prediction model. The knowledge base for the equivalent life depreciation rate is the equipment's equivalent life depreciation time prediction model. The emergency assessment method for the ultimate survival life of distribution transformers also includes: S301, obtain the full life cycle historical disaster records of retired distribution transformers.

[0075] The full life-cycle historical disaster record includes at least one disaster data point, which includes the disaster types corresponding to the decommissioned distribution transformers. Duration of the disaster The intensity of the disaster during its duration Critical intensity during the duration of the disaster .

[0076] S302, based on the full life cycle historical disaster records, determines the time sequence and disaster type indicator variables of the damage-causing disasters corresponding to decommissioned distribution transformers.

[0077] Specifically, determine disaster-type indicator variables. The process includes: disaster type indication can be obtained through the first Taiwan distribution transformer At any moment suffer the first State variables of typical disasters The decision is expressed as follows: ; in, Indicates the first Taiwan's distribution transformer suffered the first A fragment of an event typical of a disaster. Only state variables... If the value is greater than 0, the disaster type indicator variable The value is 1, as long as the state variable is 1. The sum is 0, indicating the disaster type. The value is 0. State variable. The intensity of the disaster during its duration can be used as a measure. Critical intensity during the duration of the disaster The difference between them is determined by the following expression: ; Among them, as long as in If the moment Any disaster intensity of Taiwan power distribution transformer Greater than or equal to critical strength Then state variables It is 1 if it is not 1, otherwise it is 0.

[0078] Disaster type indicator variable , can represent the first Taiwan's distribution transformer suffered the first The first typical disaster Disaster type indicator variable, if The typical disaster type in this case belongs to kind.

[0079] also, Indicates the first The number of damaging disasters suffered by a distribution transformer throughout its entire life cycle can be measured using state variables. The decision is expressed as follows: ; , They represent the first The set of start and end times of typical damaging disasters suffered sequentially throughout the entire life cycle of a distribution transformer can also be determined by state variables, as shown in the following expression: ; in, , They represent the first Taiwan's distribution transformer suffered the first The start and end times of a typical disaster.

[0080] ; in, Indicates the first The equivalent life loss time of a distribution transformer throughout its entire life cycle has the physical meaning of the life loss of the equipment under typical disaster intensity per hour during its entire life cycle. , They represent the first The ideal service life of a Taiwanese distribution transformer at the time of manufacture and its actual service life upon retirement.

[0081] Specifically, the time series of the intensity of the damaging disaster It can be determined by the duration of the disaster and The intensity of the disaster during its duration get.

[0082] S303 inputs the temporal sequence of the damage-causing disaster intensity and the disaster type indicator variables into the convolution module, and outputs the fragment impact characteristics of the local disaster event of the equipment.

[0083] Specifically, the time series of the intensity of the damage disaster and disaster type indicator variables Input convolution module, output fragment impact features of local disaster events on the device. Fragment impact characteristics The expression is as follows: ; in, This represents causal convolution, used to extract the first... The first power distribution transformer experienced the following event. Secondary damage typical disaster, the first Impact characteristics of localized event segments in disaster-like events; Represents a nonlinear activation function; , The first The input projection matrix and bias terms of the linear embedding layer of the disaster-like Mamba module.

[0084] It should be noted that the time series of the damage-causing disaster intensity It can be a logarithmically compressed sequence. Specifically, to reduce the impact of typical extreme disaster intensities on model training, the disaster intensity during the disaster duration can be pre-compressed. Logarithmic compression is performed to obtain The expression is as follows: ; in, Indicates the first Taiwan distribution transformer Time of the first Log-standardized values ​​of disaster intensity can effectively mitigate the long-tailed distribution of disaster intensity. , They represent the first Logarithmic mean and standard deviation of the intensity of each type of disaster; It is a relatively small constant.

[0085] Log-standardized value of disaster intensity Based on this, consider the time distribution, i.e. the duration of the disaster. and The time series of the damage-causing disaster intensity was obtained. .

[0086] S304, the fragment impact characteristics and sampling interval input parameter generation module, obtains the life loss output matrix, and maps the life loss output matrix to the disaster cumulative equivalent life loss time under the target disaster scenario.

[0087] Specifically, the module for generating parameters for the equipment equivalent life depreciation time prediction model.

[0088] Specifically, the first The first power distribution transformer experienced the following event. Secondary damage typical disaster, the first Fragment impact characteristics of disasters With sampling interval The common input is fed into the parameter generation layer, which adaptively generates the low-rank matrix of the transformer's low-rank frequent differential equation, which contains the hidden damage state transition matrix. , Disaster Intensity Input Driving Matrix and lifetime loss output matrix The specific expression is as follows: ; in, The first output of the parameter generation layer front end The first power distribution transformer experienced the following event. Secondary damage typical disaster, the first Intermediate characterization of disaster intensity; , These are the front-end projection matrix and bias term of the parameter generation layer, respectively; This represents the parameter generation layer, used for adaptively generating the low-rank matrix of transformer hidden damage state transitions. , Disaster Intensity Input Driving Matrix Lifetime loss output matrix And all of the aforementioned matrices belong to ; , , , Mapped to respectively U, V, B, C The linear projection matrix of the parameter layer, , , , These are the corresponding bias terms; For the first The first power distribution transformer experienced the following event. Secondary damage typical disaster, the first The disaster-like state transition matrix, due to its high dimension, is solved indirectly through a low-rank matrix.

[0089] Specifically, after obtaining the lifetime loss output matrix, in order to fully characterize the impact on the first generation throughout the entire life cycle... The first time that a distribution transformer has suffered a shortened lifespan. The continuous accumulation process of latent damage state under typical disaster effects is investigated. A low-rank ordinary differential equation is used to integrate the latent damage state of the transformer in each damage-causing event segment, and the output matrix is ​​obtained through lifetime loss. Map it to the first Accumulated equivalent life loss time for typical disasters (target disaster scenarios) .

[0090] Specifically, the expression for the cumulative equivalent life loss time under the target disaster scenario is as follows: ; in, Indicates the first The first power distribution transformer experienced the following event. Secondary damage typical disaster, the first The final state of hidden damage in disaster-like events. The expression is as follows: ; in, Indicates the first The first power distribution transformer experienced the following event. Secondary damage typical disaster, the first The initial state of hidden damage in disaster-like events. Indicates the first The first power distribution transformer experienced the following event. Secondary damage typical disaster, the first The initial state of hidden damage in a disaster is the equipment's first... -1st typical disaster causing damage The final state of hidden damage in disaster-like events; the transformer's entire life cycle. N Following the first typical disaster, its first The cumulative equivalent life loss time for typical disasters is ,Right now , , , , .

[0091] S305, when the loss function, which includes the cumulative equivalent life loss time due to disasters and the physical consistency constraint, satisfies the model iteration termination condition, update the model parameters of the convolution module and the parameter generation module respectively to obtain the trained equipment equivalent life loss time prediction model.

[0092] For decommissioned distribution transformers, the actual observable value is only the equivalent lifespan depreciation time over their entire life cycle. The middle one The cumulative equivalent life loss time caused by secondary damage Since direct observation is not possible, to ensure the robustness of model training, weakly supervised training is performed using the equivalent lifetime depreciation time of the transformer's entire lifespan, with physical consistency constraints added. The loss function is as follows: ; In the formula, the first term is The prediction error of the life-cycle depreciation time of a distribution transformer; the second item is the monotonicity constraint of a typical single-incident damage event, ensuring that the transformer suffers damage from the first... The non-negativity of the additional lifespan loss time after a typical disaster; the third term is a disaster type constraint, ensuring that the transformer suffers the first typical disaster. The typical disaster does not include the first In the event of a disaster of this type, the disaster does not result in additional lifespan loss time; the fourth item is a smoothing constraint for the transformer's hidden damage state. It is a hyperparameter that controls the strength of regularization.

[0093] Specifically, when the loss function satisfies the model iteration termination condition, the model parameters of the convolutional module and the parameter generation module are updated to obtain the trained equipment equivalent life depreciation time prediction model. This model comprises a disaster-specific Mamba encoder, a low-rank matrix generation layer for neural network differential equations, and a disaster-specific low-rank ODE integral module, collectively forming a knowledge base for the mapping between typical disaster intensity and equivalent life depreciation of distribution transformers. For any distribution transformer... By inputting the difference between the real-time disaster intensity and the critical disaster intensity, the disaster event indicator variable corresponding to the real-time disaster intensity, and the sampling time interval into the Mamba-ODE (MODE) model, the equivalent life loss time (equipment equivalent life loss rate) of equipment under typical disasters such as strong winds, sandstorms, freezing rain, icing, and wildfires can be generated.

[0094] Disaster-specific Mamba encoder: Leveraging the linear complexity and long-sequence modeling capabilities of the Mamba architecture, it achieves efficient parallel encoding of disaster time-series data. It also branches processing by disaster type to ensure that features from different scenarios do not interfere with each other. Neural ODE low-rank matrix generation layer: Models the continuous-time evolution of loss rates as solutions to differential equations, enabling the knowledge base to have continuous temporal reasoning capabilities (not limited to discrete sampling points) and adapt to inputs of arbitrary time steps. Disaster-specific low-rank ODE integration module: Significantly reduces the computational cost of ODE integration through low-rank decomposition, and integrates separately by disaster type, achieving efficient, interpretable, and scenario-independent cumulative loss calculation. The mapping knowledge base, composed of these three components, possesses four advantages compared to conventional lookup table-based knowledge bases: continuous reasoning, lightweight computation, physical consistency, and scenario scalability, providing a deployable, lightweight core engine for online real-time prediction.

[0095] In this embodiment, firstly, real-world data from the complete service life of decommissioned equipment is used for training, enabling the model to learn the entire chain of damage evolution from initial commissioning to the end of its life. This avoids the operational deviations caused by relying solely on accelerated aging data from laboratories, ensuring the authenticity and completeness of the model training from the data source. Secondly, a physical consistency constraint is introduced into the loss function, ensuring that the model training process not only fits the data but is also rigidly constrained by physical laws, thereby significantly suppressing overfitting and enhancing the model's generalization reasoning ability under unseen disaster combinations. Thirdly, a Mamba model based on low-rank divine frequent differential equations is used to mine the nonlinear mapping relationship between disaster intensity and transformer equivalent life loss, constructing a knowledge base of distribution transformer equivalent life loss under typical extreme weather conditions.

[0096] In one embodiment, such as Figure 4 The flowchart illustrating the method for determining the predicted damage trajectory and dynamic failure threshold shows that, in step S103 above, the following are specifically included: S401, based on the nonlinear accelerated degradation effect under multiple disasters, transforms the target real-time disaster intensity of the distribution transformer in at least two target disaster scenarios into the coupling degradation amplification factor of multiple disasters.

[0097] When two or more target disaster scenarios act on the same distribution transformer at the same time, the physical damage effects of each disaster are coupled and mutually reinforcing within the equipment, resulting in a total equivalent damage rate that is greater than the sum of the damage rates when each disaster acts alone. The coupling degradation amplification factor of multiple disasters is the synergistic amplification factor of this mutual coupling and mutual reinforcement.

[0098] The target real-time disaster intensity represents a specific disaster that has been identified as a target disaster scenario at the current moment (i.e., the real-time intensity has exceeded its critical intensity), and its intensity value is the real-time sensor measured value.

[0099] Specifically, when a distribution transformer faces concurrent disasters such as strong winds and icing, the stress it experiences far exceeds the linear superposition of a single disaster. The PF-PINO dynamic solution method can be used to transform the real-time disaster intensities of multiple targets into a multi-disaster concurrent coupling degradation amplification factor. The expression is as follows: ; in, and They are respectively Time of the first Class and the Real-time disaster intensity of the type of disaster (target real-time disaster intensity); and The first Class and the The physical limit stress on a transformer under the independent action of a type of disaster; The coupled limit stress when two types of disasters occur simultaneously; This is the fatigue index of the material corresponding to the key load-bearing components of the transformer. The learnable parameter matrix and bias terms inside PF-PINO are automatically optimized during PF-PINO backpropagation; The physical confidence domain boundary threshold is used for dynamic compensation.

[0100] PF-PINO is a Physics-Informed Neural Operator framework used to accelerate and improve the accuracy of phase-field modeling. PF-PINO explicitly embeds the residuals of the phase-field control equations as physical constraints into its training loss function, ensuring that its predictions always follow physical laws.

[0101] for The multi-dimensional condition monitoring feature vector of the distribution transformer extracted at each moment is expressed as follows: ; in, These are, respectively, the characteristic quantities of dissolved gas in transformer oil, the characteristic quantities of hot spot temperature in transformer windings, the characteristic quantities of partial discharge amplitude, and the characteristic quantities of peak vibration energy spectral density.

[0102] S402, based on the equivalent life loss rate of the corresponding equipment in each target disaster scenario and the coupling degradation amplification coefficient, determine the comprehensive life loss rate of the target under multiple extreme disaster coupling.

[0103] Among them, the target comprehensive life loss rate under the coupling of multiple extreme disasters This represents the comprehensive lifespan reduction rate, derived by considering the coupled and mutually reinforcing effects of physical damage caused by disasters within the equipment. Specifically, the target comprehensive lifespan reduction rate... The expression is as follows: ; in, This refers to the inherent natural aging rate of the transformer under the factory design standard. D and G This indicates the total number of extreme weather types occurring at the current moment; and These are the equivalent lifespan loss rates of the equipment corresponding to each target disaster scenario; For the first Class and the Nonlinear coupling accelerates degradation coefficient when multiple disasters occur simultaneously.

[0104] S403, based on the neural operator of the trained transformer health status prediction model, transforms the target comprehensive life loss rate into the predicted damage trajectory of the distribution transformer.

[0105] Specifically, after obtaining the target comprehensive life loss rate, the target comprehensive life loss rate can be converted into the predicted damage trajectory of the distribution transformer based on the neural operator in the trained transformer health status prediction model.

[0106] Specifically, methods for training transformer health status prediction models include: To constrain the temporal extrapolation process of the neural operator, its evolution trajectory is restricted to a space that conforms to the physical aging laws of electrical equipment, and a physical evolution equation characterizing the overall health state degradation of the transformer is constructed: ; in, For transformers in Macroscopic health status parameters at any given time, among which This indicates that the transformer is brand new and in good condition. This indicates that the system is completely ineffective. The residuals of the physical control equations; This represents the time-series gradient of the transformer's health degradation. The above formula ensures that at any time step, the transformer's health degradation gradient strictly corresponds to the current target overall lifetime degradation rate.

[0107] Fourier neural operators are employed to learn hazards in the frequency domain and discover long-range temporal degradation mappings throughout the transformer's entire lifespan. The residuals of the physical control equations are used as a soft-constraint embedding loss function for network training. The single-step temporal evolution mapping of the operator is as follows: ; Specifically, for neural-to-neural operators All internal weight parameters (including) The methods for determining this include: By integrating historical monitoring sequences of operating equipment with transformer operation and maintenance records, a system is constructed that includes data calculated from meteorological sequences. Transformer health status at the current moment Temporal degradation trajectory training set; for neural operators All internal weight parameters (including) The parameter update process, which involves random initialization, is as follows: ; in, These represent the penalty weights for data fidelity and physical residuals, respectively; Y represents the total number of samples. These represent the operator-predicted transformer state and the actual transformer calibration state, respectively. and These are the exponential decay rate constants estimated by the first and second moments, respectively.

[0108] In each training iteration, the network performs forward propagation to output the predicted state. To obtain the joint total loss for the current iteration Then, automatic differentiation is applied to calculate the gradient of the joint loss with respect to each network parameter. The Adam optimization algorithm is used to calculate the partial first-order moment estimate of the gradient. With second-order moment estimation And perform bias correction, using backpropagation at the set learning rate. Iteratively update network parameters , to the joint loss function Convergence, extracting and solidifying the current network parameters, and outputting the best trained neural operator. .

[0109] Based on this, the neural operator of the trained transformer health state prediction model The target comprehensive life loss rate Predicted damage trajectory converted into distribution transformer .

[0110] S404, based on the threshold-related parameters of the trained transformer health status prediction model, converts the target comprehensive life loss rate into a dynamic failure threshold configured for each target disaster scenario.

[0111] Among them, threshold-related parameters may include disaster-sensitive trigger offset. Threshold steepness coefficient calibrated through clearance tests and threshold dynamic penalty sensitivity coefficient .

[0112] Specifically, if the distribution transformer is an unoperated transformer, set the actual initial health level. If the distribution transformer is in operation, its actual initial health level is calculated by integrating the current online monitoring status. (Obtained from this formula) Considering that the insulation material's resistance will decrease due to fatigue after being subjected to extreme weather and concurrent impacts, this embodiment proposes a dynamic lifetime failure threshold that considers the fatigue accumulation of real-time disaster intensity to overcome the risk of missed assessments caused by traditional static safety boundaries. This is a dynamic failure threshold configured for each target disaster scenario. The expression is as follows: ; in, This is the inherent static decommissioning threshold under normal operating conditions. The target is the overall life loss rate. Extreme disaster resistance physical limit benchmark value designed for transformers For disaster-sensitive trigger offset, To achieve the threshold steepness coefficient calibrated through the clearance test, The threshold dynamic penalty sensitivity coefficient.

[0113] Specifically, after obtaining the dynamic failure threshold configured for each target disaster scenario... And the predicted damage trajectory of distribution transformers. Afterwards, the predicted damage trajectory can be compared in real time. With dynamic failure threshold The ultimate lifespan of a distribution transformer under complex operating conditions is obtained using the following formula: ; in, To predict the failure time point at which the damage trajectory curve first reaches the dynamic failure threshold. Based on this failure time point. and the start time of the simulation The total duration of the ultimate survival time can be determined. The expression is as follows: ; In one embodiment, to verify the feasibility of the emergency assessment method for the ultimate survival life of distribution transformers under complex and harsh environments proposed in the above embodiments, this implementation case uses a transformer actually in service at a power transmission and distribution network hub substation in a high-altitude and cold region of my country, as well as a meteorological observation sequence including multiple extreme weather events such as strong winds, sandstorms, freezing rain, icing, and wildfires, as the research object. First, the duration of the disaster and the environmental variables of the disaster are used as input variables on the surface, and the corresponding critical disaster intensity is used as the output variable on the surface, constructing the first... The critical intensity surface of a disaster, such as Figure 5 As shown, a knowledge base for the equivalent life loss of distribution transformers under typical extreme weather conditions (equipment equivalent life loss time prediction model) is constructed. The comprehensive life loss rate of transformers under multiple coupled disasters is quantified by calling upon the knowledge base. Furthermore, the dynamic lifetime failure threshold of real-time disaster intensity should be considered. The ultimate survival life of distribution transformers under multiple concurrent disasters was evaluated. .

[0114] To verify the effectiveness and advancement of the method in this embodiment, four evaluation schemes were set up for comparative verification. The proposed method, under the condition of multiple disasters occurring simultaneously, comprehensively considers the coupled degradation effects of multiple disasters and relies on the PF-PINO model and dynamic lifetime failure threshold to conduct emergency assessment of the transformer's ultimate survival life. Scheme 1 uses a traditional transformer life assessment method, only considering the impact of conventional environmental factors such as high temperature or high humidity on lifespan, without considering the damage caused by extreme weather disasters. Scheme 2 uses a static lifetime loss assessment model driven by a single disaster, only targeting the dominant single disaster and ignoring the coupled superposition effect of concurrent disasters such as strong winds. Scheme 3 does not introduce low-rank ordinary differential equations and the PF-PINO model, but only uses conventional pure data-driven recurrent neural networks (such as LSTM) for lifetime time-series extrapolation, and uses a fixed static lifetime failure threshold. The evaluation results of different schemes are compared in Table 1. Table 1 Comparison of Emergency Insulation Life Assessment Indicators for Transformers under Different Schemes ; In this embodiment, the evaluation method effectively extracts the dynamic threshold of the critical intensity of disasters under concurrent multi-hazard scenarios. By combining the Mamba model of low-rank god frequent differential equations with the PF-PINO operator, it not only significantly reduces the quantification deviation of equivalent lifetime damage and the prediction error of ultimate lifetime, but also achieves a more efficient computational response rate while ensuring the error rate is maintained. Simulation results verify the significant advantages of the proposed method in emergency assessment of the ultimate survival life of distribution transformers in complex and harsh environments.

[0115] In this embodiment, firstly, by explicitly calculating the coupling degradation amplification factor, the qualitatively perceived synergistic damage is transformed into a quantitative and computable amplification factor. This ensures that all subsequent calculations reflect the physical coupling effects between multiple hazards, fundamentally eliminating the systematic underestimation of concurrent risks by the traditional linear superposition model. Secondly, leveraging the continuous function approximation capability and high-dimensional mapping advantage of neural operators, the discrete comprehensive loss rate input is transformed into a continuously time-varying damage evolution curve. This gives the predicted damage trajectory temporal continuity and input sensitivity, rather than discrete-point linear interpolation, significantly improving the tracking accuracy and response smoothness of the damage trajectory to dynamic changes in the comprehensive loss rate. Thirdly, the equivalent lifetime loss knowledge base is invoked, and based on the regular curves obtained from the knowledge base, the initial lifetime health state of the transformer is considered. With overall life damage rate As initial boundary conditions, a transformer health status prediction model based on PF-PINO multi-hazard concurrent conditions and an emergency assessment model for the ultimate survival life of the main transformer were constructed, and Fourier neural operators were introduced. The aging evolution of transformers under multiple concurrent disasters is studied in the frequency domain, and the damage trajectory of the transformer is dynamically extrapolated over time. A dynamic lifetime failure threshold considering the real-time disaster intensity is proposed. In the predicted damage time trajectory First time reaching the dynamic lifetime failure threshold Transformer's ultimate survival life under multiple disaster concurrency output .

[0116] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0117] Based on the same inventive concept, this application also provides an emergency assessment device for the ultimate survival life of distribution transformers to implement the above-mentioned emergency assessment method for the ultimate survival life of distribution transformers. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the emergency assessment device for the ultimate survival life of distribution transformers provided below can be found in the limitations of the emergency assessment method for the ultimate survival life of distribution transformers described above, and will not be repeated here.

[0118] In one embodiment, such as Figure 6 As shown, an emergency assessment device for the ultimate survival life of a distribution transformer is provided, wherein: Analysis module 601 is used to deduce the critical disaster intensity of the distribution transformer under multiple disaster scenarios through multi-hazard evolution mechanism analysis; among them, when the real-time disaster intensity of the distribution transformer is greater than the critical disaster intensity, the distribution transformer will experience a shortened lifespan. Extraction module 602 is used to extract the equivalent life loss rate of the distribution transformer under each disaster scenario from the knowledge base. The knowledge base contains the mapping relationship between the real-time disaster intensity of the distribution transformer and the equivalent life loss rate of the equipment. The conversion module 603 is used to convert the equivalent life loss rate of the equipment corresponding to each target disaster scenario into the predicted damage trajectory of the distribution transformer and the dynamic failure threshold configured for each target disaster scenario, based on the nonlinear accelerated degradation effect when multiple disasters occur simultaneously, when the distribution transformer is simultaneously under at least two target disaster scenarios. Processing module 604 is used to determine the ultimate lifespan of the distribution transformer based on the time corresponding to the intersection of the predicted damage trajectory and the dynamic failure threshold.

[0119] Each module in the aforementioned emergency assessment device for the ultimate survival life of distribution transformers can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0120] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 7As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores critical disaster intensity data. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements an emergency assessment method for the ultimate survival life of a distribution transformer.

[0121] Those skilled in the art will understand that Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0122] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method steps.

[0123] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the above-described method steps.

[0124] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the above-described method steps.

[0125] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0126] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0127] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. An emergency assessment method for the ultimate survival life of a distribution transformer, characterized in that, The method includes: The critical disaster intensity of distribution transformers under multiple disaster scenarios was derived by analyzing the evolution mechanism of multiple disasters. Where the real-time disaster intensity experienced by the distribution transformer is greater than the critical disaster intensity, the distribution transformer experiences a shortened lifespan. For each disaster scenario, the equivalent life loss rate of the distribution transformer under each disaster scenario is extracted from the knowledge base. The knowledge base contains the mapping relationship between the real-time disaster intensity of the distribution transformer and the equivalent life loss rate of the equipment. The knowledge base for the equivalent life depreciation rate is a prediction model for the equivalent life depreciation time of equipment. The process involves acquiring the full lifecycle historical disaster records of decommissioned distribution transformers. Each full lifecycle historical disaster record includes at least one disaster data point, comprising the disaster type, duration, intensity, and critical intensity corresponding to the decommissioned distribution transformer. Based on these full lifecycle historical disaster records, the process determines the damage-causing disaster intensity time series and disaster type indicator variables corresponding to the decommissioned distribution transformer. The damage-causing disaster intensity time series and disaster type indicator variables are input into a convolution module to output fragment impact features of local disaster events affecting the equipment. The fragment impact features and sampling interval are input into a parameter generation module to obtain a lifespan loss output matrix, which is then mapped to the cumulative equivalent lifespan loss time under the target disaster scenario. When the loss function, including the cumulative equivalent lifespan loss time and physical consistency constraints, satisfies the model iteration termination condition, the model parameters of the convolution module and the parameter generation module are updated to obtain a trained equipment equivalent lifespan loss time prediction model. When the distribution transformer is simultaneously subjected to at least two target disaster scenarios, based on the nonlinear accelerated degradation effect during multiple disasters, the equipment equivalent life loss rate corresponding to each of the target disaster scenarios is transformed into the predicted damage trajectory of the distribution transformer and the dynamic failure threshold configured for each of the target disaster scenarios. The ultimate lifespan of the distribution transformer is determined based on the time corresponding to the intersection of the predicted damage trajectory and the dynamic failure threshold.

2. The method according to claim 1, characterized in that, The critical disaster intensity of distribution transformers under various disaster scenarios is derived through multi-hazard evolution mechanism analysis, including: For each disaster scenario, the duration of the disaster and the environmental variables of the disaster are obtained by the multi-hazard evolution mechanism for the distribution transformer under the disaster scenario; The disaster scenarios include freezing rain, icing, strong wind, sandstorm, and wildfire scenarios, and the environmental variables of the disasters include ambient temperature, ambient wind speed, air density, relative humidity, and distance from the fire source. Based on the duration of the disaster and the environmental variables of the disaster, the critical disaster intensity of the distribution transformer under the disaster scenario is determined.

3. The method according to claim 1, characterized in that, The step of extracting the equivalent lifespan loss rate of the distribution transformer under each of the disaster scenarios from the knowledge base includes: The difference between the real-time disaster intensity and the critical disaster intensity, as well as the disaster event indicator variable and sampling time interval corresponding to the real-time disaster intensity, are input into the knowledge base, and the equivalent life loss rate of the distribution transformer under each of the disaster scenarios is output.

4. The method according to claim 1, characterized in that, When the distribution transformer is simultaneously subjected to at least two target disaster scenarios, based on the nonlinear accelerated degradation effect during multiple disasters, the equivalent life loss rate of the equipment corresponding to each of the target disaster scenarios is converted into the predicted damage trajectory of the distribution transformer and the dynamic failure threshold configured for each of the target disaster scenarios, including: Based on the nonlinear accelerated degradation effect under multiple disasters, the target real-time disaster intensity of the distribution transformer in at least two target disaster scenarios is transformed into the coupling degradation amplification factor of multiple disasters. Based on the equivalent life loss rate of the equipment corresponding to each of the target disaster scenarios and the coupling degradation amplification coefficient, the target comprehensive life loss rate under multiple extreme disaster coupling is determined; Based on the neural operator of the trained transformer health status prediction model, the target comprehensive life loss rate is converted into the predicted damage trajectory of the distribution transformer. Based on the threshold-related parameters of the trained transformer health status prediction model, the target comprehensive life loss rate is converted into a dynamic failure threshold configured for each target disaster scenario.

5. An emergency assessment device for the ultimate survival life of a distribution transformer, characterized in that, The device includes: The analysis module is used to deduce the critical disaster intensity of the distribution transformer under multiple disaster scenarios through multi-hazard evolution mechanism analysis; wherein, when the real-time disaster intensity experienced by the distribution transformer is greater than the critical disaster intensity, the distribution transformer exhibits a shortened lifespan. An extraction module is used to extract, for each disaster scenario, the equivalent life loss rate of the distribution transformer under each disaster scenario from a knowledge base. The knowledge base contains the mapping relationship between the real-time disaster intensity of the distribution transformer and the equivalent life loss rate of the equipment. The knowledge base for the equivalent life depreciation rate is a prediction model for the equivalent life depreciation time of equipment. The process involves acquiring the full lifecycle historical disaster records of decommissioned distribution transformers. Each full lifecycle historical disaster record includes at least one disaster data point, comprising the disaster type, duration, intensity, and critical intensity corresponding to the decommissioned distribution transformer. Based on these full lifecycle historical disaster records, the process determines the damage-causing disaster intensity time series and disaster type indicator variables corresponding to the decommissioned distribution transformer. The damage-causing disaster intensity time series and disaster type indicator variables are input into a convolution module to output fragment impact features of local disaster events affecting the equipment. The fragment impact features and sampling interval are input into a parameter generation module to obtain a lifespan loss output matrix, which is then mapped to the cumulative equivalent lifespan loss time under the target disaster scenario. When the loss function, including the cumulative equivalent lifespan loss time and physical consistency constraints, satisfies the model iteration termination condition, the model parameters of the convolution module and the parameter generation module are updated to obtain a trained equipment equivalent lifespan loss time prediction model. The conversion module is used to convert the equipment equivalent life loss rate corresponding to each of the target disaster scenarios into the predicted damage trajectory of the distribution transformer and the dynamic failure threshold configured for each of the target disaster scenarios, based on the nonlinear accelerated degradation effect when multiple disasters occur simultaneously, when the distribution transformer is simultaneously under at least two target disaster scenarios. The processing module is used to determine the ultimate lifespan of the distribution transformer based on the time corresponding to the intersection of the predicted damage trajectory and the dynamic failure threshold.

Images