Meteorological safety risk early warning method and system for heterogeneous power equipment

Abstract

This application discloses a method and system for meteorological safety risk early warning of heterogeneous power equipment, belonging to the field of power equipment early warning technology. The method includes: acquiring the inherent parameter set and risk information set of the heterogeneous power equipment; acquiring meteorological characteristics corresponding to the risk information set and performing correlation assessment to acquire a single meteorological correlation parameter set for the heterogeneous power equipment; randomly combining the meteorological characteristics to acquire comprehensive meteorological characteristics and performing correlation assessment to acquire a comprehensive meteorological correlation parameter set for the heterogeneous power equipment; combining the single meteorological correlation parameter set and the comprehensive meteorological correlation parameter set to acquire a meteorological risk parameter set, and training a meteorological early warning agent to perform meteorological safety risk early warning. This solves the technical problem that existing technologies fail to fully consider the individual differences of heterogeneous power equipment and the coupling effect of multiple meteorological factors, resulting in insufficient accuracy of meteorological safety risk early warning.

Description

Technical Field

[0001] This application relates to the field of power equipment early warning technology, specifically to a meteorological safety risk early warning method and system for heterogeneous power equipment. Background Technology

[0002] With the deepening of smart grid construction, the scale of the power system continues to expand, the power grid structure is becoming increasingly complex, and various power equipment exhibits significant heterogeneity in terms of type, specifications, and operating environment.

[0003] However, different types of electrical equipment differ fundamentally in their physical structure, insulation materials, heat dissipation methods, and anti-interference capabilities, resulting in varying sensitivities to meteorological factors and different fault response modes. Existing meteorological early warning methods for electrical equipment use a uniform meteorological threshold standard, failing to fully consider the individual differences in withstand voltage levels, insulation aging, and installation locations of heterogeneous equipment, leading to discrepancies between early warning results and actual risk levels.

[0004] Furthermore, existing technologies typically analyze the correlation between single meteorological factors and equipment failure in isolation, neglecting the complex risk effects of the coupled effects of multiple meteorological factors, which often leads to a non-linear surge in the probability of equipment failure. Summary of the Invention

[0005] This application provides a meteorological safety risk early warning method and system for heterogeneous power equipment, which solves the technical problem that the existing technology fails to fully consider the individual differences of heterogeneous power equipment and the coupling effect of multiple meteorological factors, resulting in insufficient accuracy of meteorological safety risk early warning.

[0006] The technical solution to the above-mentioned technical problems in this application is as follows:

[0007] In a first aspect, this application provides a meteorological safety risk early warning method for heterogeneous power equipment, the method comprising:

[0008] Obtain the inherent parameter set and risk information set of heterogeneous power equipment;

[0009] Obtain the meteorological characteristics corresponding to the risk information set, and conduct a correlation assessment in combination with the risk information set to obtain a single meteorological correlation parameter set for heterogeneous power equipment;

[0010] The meteorological features are randomly combined to obtain comprehensive meteorological features, and the correlation is evaluated in conjunction with the risk information set to obtain a comprehensive meteorological correlation parameter set for heterogeneous power equipment.

[0011] By combining the single meteorological correlation parameter set and the comprehensive meteorological correlation parameter set, a meteorological risk parameter set is obtained. Based on the inherent parameter set, risk information set and meteorological risk parameter set, a meteorological early warning agent is trained to conduct meteorological safety risk early warning.

[0012] Secondly, this application provides a meteorological safety risk early warning system for heterogeneous power equipment, including:

[0013] The information acquisition module is used to acquire the inherent parameter set and risk information set of heterogeneous power equipment;

[0014] The feature acquisition module is used to acquire meteorological features corresponding to the risk information set, and to perform correlation assessment in combination with the risk information set to acquire a single meteorological correlation parameter set for heterogeneous power equipment.

[0015] The feature combination module is used to randomly combine the meteorological features to obtain comprehensive meteorological features, and combine them with the risk information set to perform correlation assessment to obtain a comprehensive meteorological correlation parameter set for heterogeneous power equipment.

[0016] The risk warning module is used to obtain a meteorological risk parameter set by combining the single meteorological correlation parameter set and the comprehensive meteorological correlation parameter set, and to train a meteorological warning agent based on the inherent parameter set, risk information set and meteorological risk parameter set to carry out meteorological safety risk warning.

[0017] This application provides one or more technical solutions, which have at least the following technical effects or advantages:

[0018] This application provides a meteorological safety risk early warning method and system for heterogeneous power equipment. First, it acquires the inherent parameter set and risk information set of the heterogeneous power equipment, establishing a correlation between individual equipment differences and fault history. Second, it assesses the correlation between a single meteorological factor and the risk information set, obtaining a single meteorological correlation parameter set to quantify the sensitivity of different equipment types to specific meteorological conditions. Third, it obtains a comprehensive meteorological correlation parameter set by randomly combining meteorological features and screening for comprehensive meteorological combinations with high synchronization rates, revealing the composite risk characteristics under the coupled effects of multiple meteorological factors. Finally, it integrates the single and comprehensive meteorological correlation parameter sets to form a meteorological risk parameter set, and trains a meteorological early warning agent based on the inherent parameter set, risk information set, and meteorological risk parameter set to achieve early warning considering equipment heterogeneity and meteorological coupling effects.

[0019] Through the above technical solution, this application breaks through the limitations of the traditional unified threshold early warning mode. By hierarchical correlation assessment and agent training, it significantly improves the accuracy and adaptability of meteorological safety risk early warning for heterogeneous power equipment. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating the meteorological safety risk early warning method for heterogeneous power equipment provided in an embodiment of this application;

[0022] Figure 2 This is a schematic diagram of the structure of the meteorological safety risk early warning system for heterogeneous power equipment provided in the embodiments of this application.

[0023] The components represented by each number in the attached diagram are explained below:

[0024] Information collection module 11, feature acquisition module 12, feature combination module 13, risk warning module 14. Detailed Implementation

[0025] This application provides a meteorological safety risk early warning method and system for heterogeneous power equipment, which addresses the technical problem that existing technologies fail to fully consider the individual differences of heterogeneous power equipment and the coupling effect of multiple meteorological factors, resulting in insufficient accuracy of meteorological safety risk early warning.

[0026] Example 1, as Figure 1 As shown in the embodiments of this application, a meteorological safety risk early warning method for heterogeneous power equipment is provided, including:

[0027] S10: Obtain the inherent parameter set and risk information set of heterogeneous power equipment;

[0028] In this embodiment of the application, since the heterogeneous power devices differ in voltage level, insulation medium, cooling method, installation environment, etc., the inherent parameter set and risk information set of the heterogeneous power devices are collected first.

[0029] Specifically, step S10 in the method includes:

[0030] Obtain the inherent parameter set of heterogeneous power equipment, wherein the inherent parameter set includes equipment type and equipment parameter standards;

[0031] Obtain a risk information set for heterogeneous power equipment, wherein the risk information set includes fault type and fault period.

[0032] In this embodiment of the application, the inherent parameter set of heterogeneous power equipment includes equipment type and equipment parameter standard. The equipment type covers a variety of categories such as transformers, circuit breakers, disconnect switches, surge arresters, and cables. The equipment parameter standard includes technical indicators such as rated voltage, rated current, insulation class, protection class, installation altitude, heat dissipation method, insulation medium type, and service life. The inherent parameters determine the power equipment's tolerance to meteorological environment and its sensitivity characteristics.

[0033] The risk information set is constructed based on historical operation and maintenance data and real-time monitoring records. Fault types are classified into categories such as insulation breakdown, mechanical damage, thermal fault, discharge fault, and corrosion aging according to the equipment fault mechanism. The fault time period records the start time and duration of the fault.

[0034] Furthermore, by aligning the fault period with the corresponding meteorological observation data in a time series, a temporal correlation between equipment faults and meteorological conditions is established.

[0035] S20: Obtain the meteorological characteristics corresponding to the risk information set, and conduct a correlation assessment in combination with the risk information set to obtain a single meteorological correlation parameter set for heterogeneous power equipment;

[0036] In this embodiment, meteorological characteristic data during the fault occurrence period are extracted based on the temporal correlation between the fault period and meteorological observation data. For each individual meteorological type, its statistical characteristic value during the fault period is calculated to obtain a meteorological feature vector. Then, a correlation assessment is performed in conjunction with a risk information set to quantify the correlation strength between a single meteorological factor and equipment failure.

[0037] Specifically, step S20 in the method includes:

[0038] Obtain the meteorological characteristics of the risk information set corresponding to the fault period, wherein the meteorological characteristics include meteorological type and meteorological period;

[0039] Extract the meteorological type frequency corresponding to the same fault type in the risk information set to obtain single-type correlation parameters;

[0040] Extract the overlap duration of fault time periods and meteorological time periods corresponding to the same fault type in the risk information set, and obtain a single duration correlation parameter;

[0041] For each meteorological type, the single-type correlation parameter and the single-duration correlation parameter are calculated by weighting to obtain the single meteorological correlation parameter of the meteorological type to the fault type.

[0042] The single meteorological correlation parameter is integrated based on the inherent parameter set to obtain a single meteorological correlation parameter set.

[0043] In this embodiment of the application, firstly, for each weather type, the frequency of its occurrence during the various fault occurrence periods is statistically analyzed, and the co-occurrence frequency of the weather type and the specific fault type is calculated as a single-type correlation parameter.

[0044] Secondly, the degree of temporal overlap between the fault period and the meteorological period is analyzed, and the proportion of the overlap duration to the total fault duration is calculated as a single duration correlation parameter to reflect the depth of the impact of the continuous effect of meteorological conditions on the occurrence of the fault.

[0045] Then, weighting coefficients are set according to the importance of equipment type and the extremeness of meteorological conditions. The single-type correlation parameter and the single-duration correlation parameter are weighted and fused to obtain the single meteorological correlation parameter of the meteorological type to the fault type. Among them, the weights are dynamically adjusted according to the disaster intensity level of the meteorological type and the criticality level of the equipment. For example, higher weights are given to extreme meteorological events such as typhoons and ice storms, and higher weights are given to the core equipment of the hub substation.

[0046] Finally, the single meteorological correlation parameters are classified and integrated according to the dimensions such as equipment type, voltage level, and years of operation in the inherent parameter set to form a structured single meteorological correlation parameter set, which characterizes the differences in sensitivity of different types of heterogeneous power equipment to single meteorological factors.

[0047] For example, the calculation process of a single meteorological correlation parameter for an insulation breakdown fault occurring under thunderstorm conditions for a 110kV oil-immersed transformer is as follows: First, the number of times thunderstorms and insulation breakdown faults of this type of transformer are co-occurring in historical data is counted, and the single type correlation parameter is obtained as 0.78; second, the average overlap period between the thunderstorm period and the fault period is calculated as 0.65, and the single duration correlation parameter is obtained; then, the type weight is set to 0.6 and the duration weight is set to 0.4, and the single meteorological correlation parameter is obtained by weighted calculation as 0.728; finally, combined with the inherent parameters such as the rated voltage and service life of the transformer, it is classified into the single meteorological correlation parameter set of the high voltage level, long service life equipment category.

[0048] S30: Randomly combine the meteorological features to obtain comprehensive meteorological features, and combine them with the risk information set to conduct a correlation assessment to obtain a comprehensive meteorological correlation parameter set for heterogeneous power equipment;

[0049] In this embodiment of the application, considering that it is difficult to capture the compound risk effect under the coupling effect of multiple meteorological conditions by analyzing a single meteorological factor, this embodiment of the application introduces a comprehensive meteorological feature construction mechanism, generates a multi-meteorological factor superposition scenario by random combination, and performs correlation assessment by combining risk information set to quantify the impact of multi-meteorological coupling effect on the failure of heterogeneous power equipment.

[0050] Specifically, step S30 in the method includes:

[0051] The meteorological features are combined, and meteorological combinations with a synchronous occurrence rate greater than the occurrence rate threshold are selected to obtain comprehensive meteorological features, wherein the comprehensive meteorological features include comprehensive meteorological type and comprehensive meteorological time period;

[0052] Extract the frequency of the comprehensive meteorological type with the same fault type from the risk information set, and obtain the comprehensive type correlation parameter;

[0053] Extract the overlap duration between the fault time period corresponding to the same fault type in the risk information set and the comprehensive meteorological time period to obtain the comprehensive duration correlation parameter;

[0054] For each of the comprehensive meteorological types, the comprehensive type correlation parameter and comprehensive duration correlation parameter are calculated by weighting, and the comprehensive meteorological correlation parameter of the comprehensive meteorological type to the fault type is obtained;

[0055] The comprehensive meteorological correlation parameters are integrated based on the inherent parameter set to obtain the comprehensive meteorological correlation parameter set.

[0056] In this embodiment, meteorological features are first randomly combined to generate a candidate combination set containing two or more meteorological types. The synchronous occurrence rate of each candidate combination in historical meteorological data is calculated, and meteorological combinations with a synchronous occurrence rate greater than a preset occurrence rate threshold are selected as comprehensive meteorological features. The comprehensive meteorological features include comprehensive meteorological types and corresponding comprehensive meteorological time periods. The comprehensive meteorological type represents a complex meteorological scene with multiple meteorological factors superimposed, and the comprehensive meteorological time period records the complete time window in which multiple meteorological conditions occur simultaneously.

[0057] Secondly, for the selected comprehensive meteorological characteristics, the frequency of their occurrence during the various fault occurrence periods is statistically analyzed, and the co-occurrence frequency of the comprehensive meteorological type and the specific fault type is calculated as the comprehensive type correlation parameter, reflecting the close relationship between multiple meteorological coupling scenarios and equipment faults.

[0058] Furthermore, the temporal overlap characteristics between the fault period and the comprehensive meteorological period are analyzed, and the proportion of the overlap duration to the total fault duration is calculated as a comprehensive duration correlation parameter to assess the cumulative impact of the continuous effects of the combined meteorological conditions on the fault evolution.

[0059] Then, differentiated weights are set according to the extreme degree, superposition complexity, and criticality level of each meteorological factor in the comprehensive meteorological type, and the comprehensive meteorological type correlation parameters and comprehensive duration correlation parameters are weighted and fused to obtain the comprehensive meteorological correlation parameters of the comprehensive meteorological type for the fault type. Among them, the weights are dynamically adjusted according to the comprehensive meteorological coupling strength level and the importance level of the equipment. For example, high-disaster combinations such as high temperature and high humidity superposition and strong wind and rainfall coupling are given higher weights, and critical equipment such as ultra-high voltage substations and urban core ring network cabinets are given higher weights.

[0060] Finally, the comprehensive meteorological correlation parameters are classified and integrated according to the dimensions of equipment category, insulation medium type, and installation environment characteristics in the inherent parameter set to form a structured comprehensive meteorological correlation parameter set, which characterizes the differences in composite sensitivity of different types of heterogeneous power equipment under multiple meteorological coupling conditions.

[0061] For example, the calculation process of the comprehensive meteorological correlation parameter for a discharge fault occurring under the superimposed meteorological conditions of high temperature and high humidity for a 220kV gas-insulated metal-enclosed switchgear is as follows: First, candidate combinations of high temperature and high humidity are generated by random combination. The synchronous occurrence rate of this combination in historical data is calculated to be 0.32, which is greater than the preset threshold of 0.25, so it is determined as a comprehensive meteorological feature. Second, the co-occurrence frequency of the superimposed high temperature and high humidity scenario with the discharge fault of this type of switchgear is counted, and the comprehensive type correlation parameter is obtained as 0.85. Third, the average overlap time ratio between the comprehensive high temperature and high humidity period and the fault period is calculated as 0.72, and the comprehensive duration correlation parameter is obtained. Then, the type weight is set to 0.5 and the duration weight is set to 0.5, and the comprehensive meteorological correlation parameter is obtained by weighted calculation as 0.785. Finally, combined with the inherent parameters such as the insulation medium type and protection level of the equipment, it is classified into the comprehensive meteorological correlation parameter set of the gas-insulated, indoor-installed equipment category.

[0062] Specifically, the meteorological features are combined, and meteorological combinations with a synchronization rate greater than a threshold are selected to obtain comprehensive meteorological features, including:

[0063] All extracted single meteorological features are randomly combined in pairs or more to obtain several meteorological combinations. Each meteorological combination contains multiple meteorological types and the corresponding meteorological time periods for each meteorological type.

[0064] The total number of times all weather types occur simultaneously in each weather combination within a preset time range is calculated, along with the sum of the total number of times each weather type occurs individually within the preset time range.

[0065] The synchronous occurrence rate of each weather combination is obtained by dividing the total number of times each weather combination occurs synchronously by the sum of the total number of times the corresponding weather type occurs individually.

[0066] Meteorological combinations with a synchronous occurrence rate greater than the occurrence rate threshold are identified as comprehensive meteorological types. The meteorological time periods of each meteorological type in the effective meteorological combinations are extracted and combined as comprehensive meteorological time periods, and comprehensive meteorological characteristics are obtained by integration.

[0067] In this embodiment, firstly, all extracted single meteorological features are randomly combined in pairs or more to generate a candidate meteorological combination set. Specifically, let the set of single meteorological types be M={m1,m2,...,m n}, then the candidate combinations include all pairs (m i ,m j ), triplet (m) i ,m j ,m k Up to n tuples, where i≠j≠k. Considering the computational complexity and the actual physical meaning of meteorological coupling, the embodiments of this application preferably limit the combination dimension to no more than three dimensions, that is, to consider the superposition effect of a maximum of three meteorological factors.

[0068] Secondly, for each candidate meteorological combination, the total number of times it occurs synchronously within a preset time range T is counted. The criterion for synchronous occurrence is that all meteorological types within the combination simultaneously meet their respective meteorological threshold conditions at the same time or in adjacent observation times. For example, the synchronous occurrence of the high temperature and high humidity combination requires that both the temperature threshold and the humidity threshold be met simultaneously, and the time difference between their occurrence times does not exceed the preset tolerance Δt.

[0069] Next, calculate the synchronous occurrence rate of each meteorological combination, i.e., "Synchronous occurrence rate = total number of synchronous occurrences of each meteorological combination ÷ sum of the total number of occurrences of the corresponding meteorological type alone". This indicator reflects the degree of synergy of multiple meteorological factors. The higher the synchronous occurrence rate, the more significant the meteorological coupling characteristics of the combination, and the more attention should be paid to its potential impact on power equipment failure.

[0070] Then, the calculated synchronization rate is compared and filtered with a preset synchronization rate threshold. The synchronization rate threshold is dynamically set based on historical meteorological statistical characteristics and early warning accuracy requirements, with a typical value range of 0.15 to 0.35. Meteorological combinations with a synchronization rate greater than the preset synchronization rate threshold are identified as effective comprehensive meteorological types, and the union of the meteorological time periods of each meteorological type in the combination is extracted as the comprehensive meteorological time period. The start and end times of the comprehensive meteorological time period are the start time of the earliest starting meteorological type and the end time of the latest ending meteorological type within the combination. This time period fully covers the complete time window of the superposition of multiple meteorological factors.

[0071] Finally, the selected comprehensive meteorological types are integrated with the corresponding comprehensive meteorological time periods to form structured comprehensive meteorological features. These comprehensive meteorological features not only record the type information of multiple meteorological factors superimposed, but also retain the time boundary of the continuous effect of complex meteorological conditions.

[0072] S40: By combining the single meteorological correlation parameter set and the comprehensive meteorological correlation parameter set, a meteorological risk parameter set is obtained, and a meteorological early warning agent is trained based on the inherent parameter set, risk information set and meteorological risk parameter set to conduct meteorological safety risk early warning.

[0073] In this embodiment, based on obtaining a single meteorological correlation parameter set and a comprehensive meteorological correlation parameter set, the two types of parameters are further integrated to construct a complete meteorological risk parameter set, and a meteorological early warning intelligent agent is trained based on multi-source data to realize intelligent early warning of meteorological safety risks of heterogeneous power equipment.

[0074] Specifically, by combining the single meteorological correlation parameter set and the comprehensive meteorological correlation parameter set, a meteorological risk parameter set is obtained, including:

[0075] Extract all single meteorological correlation parameters from the single meteorological correlation parameter set, and extract all comprehensive meteorological correlation parameters from the comprehensive meteorological correlation parameter set;

[0076] The single meteorological correlation parameters and the comprehensive meteorological correlation parameters were normalized to obtain the normalized correlation parameters for each single meteorological type and the normalized correlation parameters for each comprehensive meteorological type.

[0077] The normalized correlation parameters of each individual meteorological type and the normalized correlation parameters of each comprehensive meteorological type are integrated to form a meteorological risk parameter set.

[0078] In this embodiment, firstly, all correlation parameters are extracted from the single meteorological correlation parameter set and the comprehensive meteorological correlation parameter set, including the correlation parameters of each single meteorological type to different fault types and the correlation parameters of each comprehensive meteorological type to different fault types. Since the calculation scale and numerical range of the single meteorological correlation parameters and the comprehensive meteorological correlation parameters may differ, direct fusion will affect the stability of subsequent model training; therefore, normalization processing is required for each.

[0079] Specifically, for a single meteorological correlation parameter, the Min-Max normalization method is used to map it to the [0,1] interval, that is, "normalized single correlation parameter = (single meteorological correlation parameter - minimum value of single parameter) ÷ (maximum value of single parameter - minimum value of single parameter)".

[0080] Similarly, the same normalization process is applied to the comprehensive meteorological correlation parameters to obtain the normalized correlation parameters for each comprehensive meteorological type. The normalization process preserves the relative magnitude relationship between the parameters while eliminating dimensional differences, making the risk contribution of different types of meteorological factors comparable.

[0081] Finally, after normalization, the normalized relevant parameters of each single meteorological type and the normalized relevant parameters of each comprehensive meteorological type are structurally integrated according to dimensions such as equipment category, fault type, and meteorological type to form a complete meteorological risk parameter set. This parameter set includes both the independent risk contribution of a single meteorological factor and the compound risk effect of multiple meteorological couplings.

[0082] Furthermore, a meteorological early warning agent is trained and obtained based on the aforementioned inherent parameter set, risk information set, and meteorological risk parameter set, including:

[0083] Construct a meteorological early warning intelligent agent, wherein the meteorological early warning intelligent agent includes a single meteorological branch and a comprehensive meteorological branch;

[0084] Using the single meteorological feature, the comprehensive meteorological feature, and the inherent parameter set as inputs, and the risk information set and the meteorological risk parameter set as supervision, the meteorological early warning agent is trained separately to obtain the trained meteorological early warning agent.

[0085] In this embodiment, firstly, a network architecture for a meteorological early warning agent is constructed. This agent adopts a dual-branch parallel structure to handle the differentiated influence mechanisms of single meteorological factors and comprehensive meteorological factors respectively. The single meteorological branch is designed as a multilayer perceptron structure. The input layer receives a single meteorological feature vector and the device's inherent parameter vector. The hidden layer uses the ReLU activation function to extract nonlinear features, and the output layer generates the failure risk probability under single meteorological conditions. The comprehensive meteorological branch adopts a graph neural network structure, using multiple meteorological types in the comprehensive meteorological features as graph nodes and the coupling relationships between meteorological factors as edge connections. It learns the complex interaction patterns of multiple meteorological factors through a graph attention mechanism and outputs the composite failure risk probability under comprehensive meteorological conditions. The outputs of the two branches are integrated by a weighted fusion layer. The weights are dynamically adjusted according to the real-time meteorological scene. When the comprehensive meteorological features are detected to be active, the weight ratio of the comprehensive meteorological branch is automatically increased.

[0086] Furthermore, using single meteorological features, comprehensive meteorological features, and inherent parameter sets as model inputs, and historical fault labels from the risk information set and correlation quantification values ​​from the meteorological risk parameter set as supervision signals, the meteorological early warning agent is trained end-to-end. A multi-task learning framework is employed during training, with the primary task being fault risk level classification and the auxiliary task being meteorological correlation parameter regression. The model's generalization ability is improved by sharing underlying feature representations. The loss function is designed as a weighted combination of classification cross-entropy loss and regression mean squared error loss, and a Focal Loss mechanism is introduced to alleviate the class imbalance problem caused by sparse fault samples. The Adam optimizer is used for parameter updates, and the learning rate is dynamically adjusted using a cosine annealing strategy. An early stopping mechanism is implemented during training to prevent overfitting.

[0087] For example, the initial learning rate was set to 0.001, the batch size was set to 64, the maximum number of training rounds was 200, and early stopping was triggered when the validation set loss did not decrease for 10 consecutive rounds. The final classification accuracy of the model on the training set reached more than 95%, indicating that the meteorological early warning agent has good fault risk identification ability and generalization performance.

[0088] Specifically, the input data is first preprocessed by batch normalization. Then, a dual-branch network is used to extract single meteorological risk features and comprehensive meteorological coupling features, respectively. A feature fusion layer is then used to generate a comprehensive risk representation. Finally, the fault risk level of the equipment and the corresponding meteorological risk contribution decomposition within a preset future time window are output. After training, the meteorological early warning agent has the ability to perceive real-time meteorological risks of heterogeneous power equipment. Based on the input real-time meteorological monitoring data and equipment operating status, it can automatically identify key risk equipment and generate differentiated early warning strategies.

[0089] Specifically, meteorological safety risk warnings include:

[0090] Obtain the current meteorological characteristics and the current inherent parameter set of the current power equipment as the input feature set;

[0091] Branch matching is performed based on the current meteorological feature set, and the meteorological feature set and the current inherent parameter set are input into the meteorological early warning agent to obtain predicted fault information and predicted meteorological risk parameters.

[0092] In this embodiment, firstly, real-time meteorological monitoring data is acquired, including real-time values ​​and trends of individual meteorological elements such as temperature, humidity, wind speed, air pressure, and precipitation. Simultaneously, current operating status parameters of the target power equipment are collected, such as load current, equipment temperature, insulation monitoring data, and online partial discharge signals. Combined with the inherent parameter set in the equipment ledger, a complete input feature set is constructed. The input feature set is structured according to data type; the meteorological feature part is divided into single meteorological feature vectors and comprehensive meteorological feature vectors, and the equipment parameter part is divided into static inherent parameters and dynamic operating parameters.

[0093] Secondly, intelligent branch matching is performed based on the current meteorological feature set to determine whether the current meteorological conditions trigger a comprehensive meteorological feature. Specifically, when a single meteorological feature vector simultaneously meets the composition conditions of two or more comprehensive meteorological types, and the time difference between the occurrence of each component meteorological element is within a preset tolerance range, the comprehensive meteorological feature is determined to be in an active state, and the comprehensive meteorological branch is activated to participate in the inference. If only a single meteorological element exceeds the standard but does not form an effective combination, only the single meteorological branch is activated. The branch matching result dynamically determines the weight allocation strategy of the dual-branch network to ensure that the model focuses on the currently dominant meteorological risk mechanism.

[0094] Among them, branch matching based on the current meteorological feature set includes:

[0095] When the current meteorological feature set is a single meteorological feature, the input feature set is input into a single meteorological branch;

[0096] When the current meteorological feature set is a comprehensive meteorological feature, the input feature set is split to obtain a single meteorological feature and the current inherent parameter set, and then integrated as the split input feature set.

[0097] The split input feature set and the input feature set are respectively input into the meteorological early warning intelligent agent to obtain multiple predicted fault information and multiple predicted meteorological risk parameters. Based on the inherent parameter set, the coefficient of the same type of fault is obtained. The multiple predicted meteorological risk parameters are weighted and synthesized to obtain the predicted fault information and predicted meteorological risk parameters. The coefficient of the same type of fault is obtained based on the inherent parameter set and the risk information set.

[0098] In this embodiment, the current meteorological feature set is first analyzed in real time to identify the types of meteorological features it contains. If the analysis results show that there is only a single meteorological feature, that is, a certain meteorological element exceeds the standard independently without forming an effective combination with other elements, then the complete input feature set is directly imported into a single meteorological branch for reasoning. This branch focuses on mining the monotonic or threshold response relationship between a single meteorological factor and equipment failure, and outputs the failure risk prediction result based on independent meteorological conditions.

[0099] If the analysis results indicate the existence of comprehensive meteorological characteristics, meaning that multiple meteorological elements simultaneously meet the combination conditions and form a coupled situation, then the intelligent splitting operation of the input feature set is executed.

[0100] Specifically, the comprehensive meteorological feature is broken down into its constituent individual meteorological feature subsets. Each subset retains the time boundary information from the original comprehensive meteorological feature, while extracting the current intrinsic parameter set as shared equipment background information, and integrating them to form multiple split input feature sets. For example, when the comprehensive meteorological feature of "high temperature, high humidity, and strong wind" is detected to be active, it is split into three split input feature sets: high temperature + intrinsic parameters, high humidity + intrinsic parameters, and strong wind + intrinsic parameters, while retaining the complete original input feature set of "high temperature, high humidity, and strong wind + intrinsic parameters".

[0101] Subsequently, the split input feature sets and the original input feature sets are input into the meteorological early warning agent in parallel, activating both the single meteorological branch and the comprehensive meteorological branch for multi-dimensional reasoning. The single meteorological branch independently calculates for each split input feature set, generating predicted fault information and predicted meteorological risk parameters under the individual effects of each meteorological element; the comprehensive meteorological branch calculates the composite risk output under multiple coupled meteorological conditions based on the original input feature set. This yields multiple prediction results, including risk predictions under individual meteorological scenarios and coupled risk predictions under a comprehensive meteorological scenario.

[0102] Furthermore, based on historical correlation analysis of the inherent parameter set and the risk information set, the similarity failure coefficient is obtained. The calculation process of the similarity failure coefficient is as follows: Historical failure samples in the risk information set that have the same equipment category, voltage level, and operating period as the current equipment are statistically analyzed. The proportion distribution of failures caused by single meteorological factors and failures caused by combined meteorological factors in the samples is analyzed, and the weighting coefficient is determined accordingly. The specific calculation formula is "Similarity failure coefficient = Number of samples caused by combined meteorological factors ÷ (Number of samples caused by combined meteorological factors + Number of samples caused by single meteorological factors)". This coefficient reflects the sensitivity of similar equipment to meteorological coupling effects; the closer the coefficient is to 1, the more susceptible the equipment is to combined meteorological conditions.

[0103] Finally, multiple predicted meteorological risk parameters are weighted and synthesized using similar failure coefficients. Let the normalized risk parameters output by a single meteorological branch be {p1, p2, ..., p...} n If the normalized risk parameter output by the integrated meteorological branch is pc, then the weighted integrated predicted meteorological risk parameter P = α·pc + (1-α)·max{p1,p2,...,p n}, where α is the coefficient for similar faults. The weighted predicted fault information comprehensively considers the independent risk contribution of a single meteorological factor and the synergistic amplification effect of multiple meteorological couplings, making the early warning results more consistent with the historical fault patterns of similar equipment.

[0104] In summary, compared with existing technologies, this application achieves refined perception and intelligent early warning of meteorological safety risks of heterogeneous power equipment by constructing a dual-branch parallel processing architecture of single meteorological and comprehensive meteorological data.

[0105] In summary, the embodiments of this application have at least the following technical effects:

[0106] This application provides a meteorological safety risk early warning method for heterogeneous power equipment. First, it acquires the inherent parameter set and risk information set of the heterogeneous power equipment, establishing a correlation between individual equipment differences and fault history. Second, it assesses the correlation between a single meteorological factor and the risk information set, obtaining a single meteorological correlation parameter set to quantify the sensitivity of different equipment types to specific meteorological conditions. Third, it obtains a comprehensive meteorological correlation parameter set by randomly combining meteorological features and screening for comprehensive meteorological combinations with high synchronization rates, revealing the composite risk characteristics under the coupled effects of multiple meteorological factors. Finally, it integrates the single and comprehensive meteorological correlation parameter sets to form a meteorological risk parameter set, and trains a meteorological early warning agent based on the inherent parameter set, risk information set, and meteorological risk parameter set to achieve early warning considering equipment heterogeneity and meteorological coupling effects.

[0107] Through the above technical solution, this application breaks through the limitations of the traditional unified threshold early warning mode. By hierarchical correlation assessment and agent training, it significantly improves the accuracy and adaptability of meteorological safety risk early warning for heterogeneous power equipment.

[0108] Example 2, as Figure 2 As shown, based on the same inventive concept as the meteorological safety risk early warning method for heterogeneous power equipment provided in Embodiment 1, this application also provides a meteorological safety risk early warning system for heterogeneous power equipment, including:

[0109] Information acquisition module 11 is used to acquire the inherent parameter set and risk information set of heterogeneous power equipment;

[0110] The feature acquisition module 12 is used to acquire meteorological features corresponding to the risk information set, and to perform correlation assessment in combination with the risk information set to acquire a single meteorological correlation parameter set for heterogeneous power equipment.

[0111] The feature combination module 13 is used to randomly combine the meteorological features to obtain comprehensive meteorological features, and combine them with the risk information set to perform correlation assessment to obtain a comprehensive meteorological correlation parameter set for heterogeneous power equipment.

[0112] The risk warning module 14 is used to combine the single meteorological correlation parameter set and the comprehensive meteorological correlation parameter set to obtain a meteorological risk parameter set, and to train and obtain a meteorological warning agent based on the inherent parameter set, risk information set and meteorological risk parameter set to carry out meteorological safety risk warning.

[0113] In one embodiment, the information acquisition module 11 is specifically used for:

[0114] Obtain the inherent parameter set of heterogeneous power equipment, wherein the inherent parameter set includes equipment type and equipment parameter standards;

[0115] Obtain a risk information set for heterogeneous power equipment, wherein the risk information set includes fault type and fault period.

[0116] In one embodiment, the feature acquisition module 12 is specifically used for:

[0117] Obtain the meteorological characteristics corresponding to the risk information set, and perform correlation assessment in conjunction with the risk information set to obtain a single meteorological correlation parameter set for heterogeneous power equipment, including:

[0118] Obtain the meteorological characteristics of the risk information set corresponding to the fault period, wherein the meteorological characteristics include meteorological type and meteorological period;

[0119] Extract the meteorological type frequency corresponding to the same fault type in the risk information set to obtain single-type correlation parameters;

[0120] Extract the overlap duration of fault time periods and meteorological time periods corresponding to the same fault type in the risk information set, and obtain a single duration correlation parameter;

[0121] For each meteorological type, the single-type correlation parameter and the single-duration correlation parameter are calculated by weighting to obtain the single meteorological correlation parameter of the meteorological type to the fault type.

[0122] The single meteorological correlation parameter is integrated based on the inherent parameter set to obtain a single meteorological correlation parameter set.

[0123] In one embodiment, the feature combination module 13 is specifically used for:

[0124] The meteorological features are combined, and meteorological combinations with a synchronous occurrence rate greater than the occurrence rate threshold are selected to obtain comprehensive meteorological features, wherein the comprehensive meteorological features include comprehensive meteorological type and comprehensive meteorological time period;

[0125] Extract the frequency of the comprehensive meteorological type with the same fault type from the risk information set, and obtain the comprehensive type correlation parameter;

[0126] Extract the overlap duration between the fault time period corresponding to the same fault type in the risk information set and the comprehensive meteorological time period to obtain the comprehensive duration correlation parameter;

[0127] For each of the comprehensive meteorological types, the comprehensive type correlation parameter and comprehensive duration correlation parameter are calculated by weighting, and the comprehensive meteorological correlation parameter of the comprehensive meteorological type to the fault type is obtained;

[0128] The comprehensive meteorological correlation parameters are integrated based on the inherent parameter set to obtain the comprehensive meteorological correlation parameter set.

[0129] Furthermore, in one embodiment, the meteorological features are combined, and meteorological combinations with a synchronization rate greater than a threshold are selected to obtain comprehensive meteorological features, including:

[0130] All extracted single meteorological features are randomly combined in pairs or more to obtain several meteorological combinations. Each meteorological combination contains multiple meteorological types and the corresponding meteorological time periods for each meteorological type.

[0131] The total number of times all weather types occur simultaneously in each weather combination within a preset time range is calculated, along with the sum of the total number of times each weather type occurs individually within the preset time range.

[0132] The synchronous occurrence rate of each weather combination is obtained by dividing the total number of times each weather combination occurs synchronously by the sum of the total number of times the corresponding weather type occurs individually.

[0133] Meteorological combinations with a synchronous occurrence rate greater than the occurrence rate threshold are identified as comprehensive meteorological types. The meteorological time periods of each meteorological type in the effective meteorological combinations are extracted and combined as comprehensive meteorological time periods, and comprehensive meteorological characteristics are obtained by integration.

[0134] Furthermore, in one embodiment of the application, a meteorological risk parameter set is obtained by combining the single meteorological correlation parameter set and the comprehensive meteorological correlation parameter set, including:

[0135] Extract all single meteorological correlation parameters from the single meteorological correlation parameter set, and extract all comprehensive meteorological correlation parameters from the comprehensive meteorological correlation parameter set;

[0136] The single meteorological correlation parameters and the comprehensive meteorological correlation parameters were normalized to obtain the normalized correlation parameters for each single meteorological type and the normalized correlation parameters for each comprehensive meteorological type.

[0137] The normalized correlation parameters of each individual meteorological type and the normalized correlation parameters of each comprehensive meteorological type are integrated to form a meteorological risk parameter set.

[0138] Furthermore, a meteorological early warning agent is trained and obtained based on the aforementioned inherent parameter set, risk information set, and meteorological risk parameter set, including:

[0139] Construct a meteorological early warning intelligent agent, wherein the meteorological early warning intelligent agent includes a single meteorological branch and a comprehensive meteorological branch;

[0140] Using the single meteorological feature, the comprehensive meteorological feature, and the inherent parameter set as inputs, and the risk information set and the meteorological risk parameter set as supervision, the meteorological early warning agent is trained separately to obtain the trained meteorological early warning agent.

[0141] Furthermore, meteorological safety risk warnings will be issued, including:

[0142] Obtain the current meteorological characteristics and the current inherent parameter set of the current power equipment as the input feature set;

[0143] Branch matching is performed based on the current meteorological feature set, and the meteorological feature set and the current inherent parameter set are input into the meteorological early warning agent to obtain predicted fault information and predicted meteorological risk parameters.

[0144] Furthermore, in one embodiment of the application, branch matching based on the current meteorological feature set includes:

[0145] When the current meteorological feature set is a single meteorological feature, the input feature set is input into a single meteorological branch;

[0146] When the current meteorological feature set is a comprehensive meteorological feature, the input feature set is split to obtain a single meteorological feature and the current inherent parameter set, and then integrated as the split input feature set.

[0147] The split input feature set and the input feature set are respectively input into the meteorological early warning intelligent agent to obtain multiple predicted fault information and multiple predicted meteorological risk parameters. Based on the inherent parameter set, the coefficient of the same type of fault is obtained. The multiple predicted meteorological risk parameters are weighted and synthesized to obtain the predicted fault information and predicted meteorological risk parameters. The coefficient of the same type of fault is obtained based on the inherent parameter set and the risk information set.

Claims

1. A meteorological safety risk early warning method for heterogeneous power equipment, characterized in that, include: Obtain the inherent parameter set and risk information set of heterogeneous power equipment; Obtain the meteorological characteristics corresponding to the risk information set, and conduct a correlation assessment in combination with the risk information set to obtain a single meteorological correlation parameter set for heterogeneous power equipment; The meteorological features are randomly combined to obtain comprehensive meteorological features, and the correlation is evaluated in conjunction with the risk information set to obtain a comprehensive meteorological correlation parameter set for heterogeneous power equipment. By combining the single meteorological correlation parameter set and the comprehensive meteorological correlation parameter set, a meteorological risk parameter set is obtained. Based on the inherent parameter set, risk information set, and meteorological risk parameter set, a meteorological early warning agent is trained to perform meteorological safety risk early warning, including: A meteorological early warning intelligent agent is constructed, comprising a single meteorological branch and a comprehensive meteorological branch. The single meteorological branch adopts a multilayer perceptron structure, with the input layer receiving a single meteorological feature vector and an inherent parameter vector of the equipment, the hidden layer using the ReLU activation function, and the output layer generating the failure risk probability under a single meteorological condition. The comprehensive meteorological branch adopts a graph neural network structure, using multiple meteorological types in the comprehensive meteorological features as graph nodes, the coupling relationship between meteorological features as edge connections, and outputting the composite failure risk probability under comprehensive meteorological conditions. Using the single meteorological feature, the comprehensive meteorological feature, and the inherent parameter set as inputs, and the risk information set and the meteorological risk parameter set as supervision, the single meteorological branch and the comprehensive meteorological branch of the meteorological early warning agent are trained separately to obtain the trained meteorological early warning agent. A multi-task learning framework is used during the training process.

2. The meteorological safety risk early warning method for heterogeneous power equipment according to claim 1, characterized in that, Obtain the inherent parameter set and risk information set of heterogeneous power equipment, including: Obtain the inherent parameter set of heterogeneous power equipment, wherein the inherent parameter set includes equipment type and equipment parameter standards; Obtain a risk information set for heterogeneous power equipment, wherein the risk information set includes fault type and fault period.

3. The meteorological safety risk early warning method for heterogeneous power equipment according to claim 1, characterized in that, Obtain the meteorological characteristics corresponding to the risk information set, and perform correlation assessment in conjunction with the risk information set to obtain a single meteorological correlation parameter set for heterogeneous power equipment, including: Obtain the meteorological characteristics of the risk information set corresponding to the fault period, wherein the meteorological characteristics include meteorological type and meteorological period; Extract the meteorological type frequency corresponding to the same fault type in the risk information set to obtain single-type correlation parameters; Extract the overlap duration of fault time periods and meteorological time periods corresponding to the same fault type in the risk information set, and obtain a single duration correlation parameter; For each meteorological type, the single-type correlation parameter and the single-duration correlation parameter are calculated by weighting to obtain the single meteorological correlation parameter of the meteorological type to the fault type. Based on the inherent parameter set, the single meteorological correlation parameter is integrated to obtain a single meteorological correlation parameter set.

4. The meteorological safety risk early warning method for heterogeneous power equipment according to claim 1, characterized in that, The meteorological characteristics are randomly combined to obtain comprehensive meteorological characteristics, and correlation assessment is performed in conjunction with a risk information set to obtain a comprehensive meteorological correlation parameter set for heterogeneous power equipment, including: The meteorological features are combined, and meteorological combinations with a synchronous occurrence rate greater than the occurrence rate threshold are selected to obtain comprehensive meteorological features, wherein the comprehensive meteorological features include comprehensive meteorological type and comprehensive meteorological time period; Extract the frequency of the comprehensive meteorological type with the same fault type from the risk information set, and obtain the comprehensive type correlation parameter; Extract the overlap duration between the fault time period corresponding to the same fault type in the risk information set and the comprehensive meteorological time period to obtain the comprehensive duration correlation parameter; For each of the comprehensive meteorological types, the comprehensive type correlation parameter and comprehensive duration correlation parameter are calculated by weighting, and the comprehensive meteorological correlation parameter of the comprehensive meteorological type to the fault type is obtained; The comprehensive meteorological correlation parameters are integrated based on the inherent parameter set to obtain the comprehensive meteorological correlation parameter set.

5. The meteorological safety risk early warning method for heterogeneous power equipment according to claim 4, characterized in that, The meteorological features are combined, and meteorological combinations with a synchronization rate greater than a threshold are selected to obtain comprehensive meteorological features, including: All extracted single meteorological features are randomly combined in pairs or more to obtain several meteorological combinations. Each meteorological combination contains multiple meteorological types and the corresponding meteorological time periods for each meteorological type. The total number of times all weather types occur simultaneously in each weather combination within a preset time range is calculated, along with the sum of the total number of times each weather type occurs individually within the preset time range. The synchronous occurrence rate of each weather combination is obtained by dividing the total number of times each weather combination occurs synchronously by the sum of the total number of times the corresponding weather type occurs individually. Meteorological combinations with a synchronous occurrence rate greater than the occurrence rate threshold are identified as comprehensive meteorological types. The meteorological time periods of each meteorological type in the effective meteorological combinations are extracted and combined as comprehensive meteorological time periods, and comprehensive meteorological characteristics are obtained by integration.

6. The meteorological safety risk early warning method for heterogeneous power equipment according to claim 1, characterized in that, By combining the single meteorological correlation parameter set and the comprehensive meteorological correlation parameter set, a meteorological risk parameter set is obtained, including: Extract all single meteorological correlation parameters from the single meteorological correlation parameter set, and extract all comprehensive meteorological correlation parameters from the comprehensive meteorological correlation parameter set; The single meteorological correlation parameters and the comprehensive meteorological correlation parameters were normalized to obtain the normalized correlation parameters for each single meteorological type and the normalized correlation parameters for each comprehensive meteorological type. The normalized correlation parameters of each individual meteorological type and the normalized correlation parameters of each comprehensive meteorological type are integrated to form a meteorological risk parameter set.

7. The meteorological safety risk early warning method for heterogeneous power equipment according to claim 1, characterized in that, Conduct meteorological safety risk warnings, including: Obtain the current meteorological feature set and the current inherent parameter set of the current power equipment as the input feature set; Branch matching is performed based on the current meteorological feature set, and the meteorological feature set and the current inherent parameter set are input into the meteorological early warning agent to obtain predicted fault information and predicted meteorological risk parameters.

8. The meteorological safety risk early warning method for heterogeneous power equipment according to claim 7, characterized in that, Branch matching based on the current meteorological feature set includes: When the current meteorological feature set is a single meteorological feature, the input feature set is input into a single meteorological branch; When the current meteorological feature set is a comprehensive meteorological feature, the input feature set is split to obtain a single meteorological feature and the current inherent parameter set, and then integrated as the split input feature set. The split input feature set and the input feature set are respectively input into the meteorological early warning intelligent agent to obtain multiple predicted fault information and multiple predicted meteorological risk parameters. Based on the inherent parameter set, the coefficient of the same type of fault is obtained. The multiple predicted meteorological risk parameters are weighted and synthesized to obtain the predicted fault information and predicted meteorological risk parameters. The coefficient of the same type of fault is obtained based on the inherent parameter set and the risk information set.

9. A meteorological safety risk early warning system for heterogeneous power equipment, characterized in that, The method for early warning of meteorological safety risks for heterogeneous power equipment according to any one of claims 1-8 includes: The information acquisition module is used to acquire the inherent parameter set and risk information set of heterogeneous power equipment; The feature acquisition module is used to acquire meteorological features corresponding to the risk information set, and to perform correlation assessment in combination with the risk information set to acquire a single meteorological correlation parameter set for heterogeneous power equipment. The feature combination module is used to randomly combine the meteorological features to obtain comprehensive meteorological features, and combine them with the risk information set to perform correlation assessment to obtain a comprehensive meteorological correlation parameter set for heterogeneous power equipment. The risk warning module is used to obtain a meteorological risk parameter set by combining the single meteorological correlation parameter set and the comprehensive meteorological correlation parameter set, and to train a meteorological warning agent based on the inherent parameter set, risk information set and meteorological risk parameter set to carry out meteorological safety risk warning.

Images