Power distribution network transformer multi-level state feature decomposition diagnosis system and method

Abstract

This invention discloses a multi-level state feature decomposition diagnostic system and method for distribution network transformers, relating to the field of transformer diagnostic technology. It acquires the impact data that needs to be monitored for transformers from a large database, divides the transformer into multi-level state feature layers based on this data, and each layer includes corresponding observation data. Based on the observation data, a corresponding stability index and entropy weight are generated for each state feature layer. The stability indices of all state feature layers in the transformer are weighted and fused to obtain the transformer's comprehensive characteristic value. Anomaly diagnosis is performed on the transformer based on this comprehensive characteristic value, and corresponding management strategies are generated based on the anomaly diagnosis results. By constructing multi-level state feature layers, the diagnostic system can more comprehensively reflect the different influencing factors of transformer operation, improving the detail of monitoring and the accuracy of diagnosis.

Description

Technical Field

[0001] This invention relates to the field of transformer diagnostic technology, specifically to a multi-level state characteristic decomposition diagnostic system and method for distribution network transformers. Background Technology

[0002] As an important component of the power system, distribution transformers are responsible for converting high voltage to low voltage suitable for users, providing stable power to homes, factories, and commercial establishments. Distribution transformers operate in a complex environment and are easily affected by load fluctuations, changes in ambient temperature, aging and wear, and external factors. Diagnostic systems are mainly used to monitor and analyze the operating status of distribution transformers to ensure their safe and stable operation.

[0003] The existing technology has the following drawbacks:

[0004] Existing diagnostic systems focus only on a single dimension or a limited number of state characteristics, resulting in an insufficiently comprehensive assessment of transformer condition. Furthermore, they often ignore the variability and uncertainty of characteristic data during data processing, leading to a predominantly qualitative analysis of transformer health conditions, making precise quantification difficult and thus reducing the accuracy and comprehensiveness of transformer condition diagnosis.

[0005] Based on this, the present invention proposes a multi-level state feature decomposition diagnostic system and method for distribution network transformers. By constructing a multi-level state feature layer, it can more comprehensively reflect different influencing factors of transformer operation, thereby improving the detail of monitoring and the accuracy of diagnosis. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-level state characteristic decomposition and diagnostic system and method for distribution network transformers to address the shortcomings in the prior art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a multi-level state characteristic decomposition and diagnosis method for distribution network transformers, the diagnosis method comprising the following steps:

[0008] The diagnostic system obtains the impact data that needs to be monitored for the transformer through a large database. The impact data represents the data that affects the safe and stable operation of the transformer. Based on the impact data, a multi-level state feature layer is divided, and each state feature layer includes the corresponding observation data.

[0009] Based on the observed data, a corresponding stability index and entropy weight are generated for the state feature layer;

[0010] After weighted fusion of the stability indices of all state characteristic layers in the transformer, the comprehensive characteristic value of the transformer is obtained. Anomaly diagnosis of the transformer is performed based on the comprehensive characteristic value, and corresponding management strategies are generated in combination with the anomaly diagnosis results. The management strategies are then sent to the transformer management platform.

[0011] In a preferred embodiment, generating a corresponding stability index for the state feature layer based on observation data includes the following steps:

[0012] The duration of occurrence of small, normal, and large observation data for any component in the state feature layer observation data is obtained. The duration of occurrence of small, normal, and large observation data is summed to obtain the total duration. The probability of small observation data is obtained by dividing the duration of small observation data by the total duration, the probability of normal observation data is obtained by dividing the duration of normal observation data by the total duration, and the probability of large observation data is obtained by dividing the duration of large observation data by the total duration.

[0013] The stable value of the component is calculated based on the probability of undersized observation data, the probability of normal observation data, and the probability of undersized observation data. The expression is as follows: In the formula, W is the stable value, p(x1) is the probability of undersized observation data, p(x2) is the probability of normal observation data, and p(x3) is the probability of undersized observation data. The expression for calculating the stability index of the state feature layer is:

[0014] In the formula, SW is the stability index, n is the number of components included in the state characteristic layer, and W i Let be the stability value of the i-th component. The larger the stability index, the higher the overall stability of the state characteristic layer during transformer operation.

[0015] In a preferred embodiment, generating corresponding entropy weights for the state feature layer based on the observed data includes the following steps:

[0016] The duration of occurrence of small, normal, and large observation data for any component in the state feature layer observation data is obtained. The duration of occurrence of small, normal, and large observation data is summed to obtain the total duration. The probability of small observation data is obtained by dividing the duration of small observation data by the total duration, the probability of normal observation data is obtained by dividing the duration of normal observation data by the total duration, and the probability of large observation data is obtained by dividing the duration of large observation data by the total duration.

[0017] The weights of components are calculated based on the probabilities of undersized observations, normal observations, and oversized observations. The expression is as follows:

[0018] H=-(p(x1)log(p(x1))+p(x2)log(p(x2))+p(x3)log(p(x3))); Eq.

[0019] In this context, H represents the weights, p(x1) is the probability of smaller observed data, p(x2) is the probability of normal observed data, and p(x3) is the probability of larger observed data. Since the state feature layer includes multiple components, the entropy weight of the state feature layer is obtained by summing the weights of the multiple components. The expression is as follows:

[0020] In the formula, SH is the entropy weight, n is the number of components included in the state feature layer, and H... i Let be the weight of the i-th component.

[0021] In a preferred embodiment, the comprehensive characteristic value of the transformer is obtained by weighted fusion of the stability indices of all state characteristic layers in the transformer, including the following steps:

[0022] After weighted fusion of the stability indices of all state characteristic layers in the transformer, the comprehensive characteristic value of the transformer is obtained, expressed as: In the formula, tz s The comprehensive characteristic value is SW, where m is the number of state characteristic layers in the transformer. i Let SH be the stability index of the i-th state feature layer. i is the entropy weight of the i-th state feature layer.

[0023] In a preferred embodiment, transformer anomaly diagnosis based on comprehensive characteristic values ​​includes the following steps:

[0024] The comprehensive feature value is compared with a preset health threshold. The health threshold is used to diagnose whether the transformer is abnormal. If the comprehensive feature value is greater than or equal to the health threshold, the transformer is diagnosed as not being abnormal. If the comprehensive feature value is less than the health threshold, the transformer is diagnosed as being abnormal.

[0025] In a preferred embodiment, the diagnostic system obtains the impact data of the transformer that needs to be monitored through a large database, including the following steps:

[0026] The diagnostic system obtains the impact data that needs to be monitored for the transformer through a large database. The impact data includes temperature data, resistance data, current data, and vibration data.

[0027] Based on the influence data, a multi-level state feature layer is divided, based on temperature data, based on resistance data, based on current data, and based on vibration data, a vibration state feature layer is divided.

[0028] In a preferred embodiment, each state feature layer includes corresponding observation data. The observation data of the temperature state feature layer includes winding temperature, cooling oil temperature, and core temperature. The observation data of the resistance state feature layer includes winding resistance, bushing resistance, and contact resistance. The observation data of the current state feature layer includes winding current and load current. The observation data of the vibration state feature layer includes winding vibration, core vibration, and oil pump vibration.

[0029] A multi-level state characteristic decomposition and diagnosis system for distribution network transformers includes a data collection module, a parameter generation module, and an anomaly diagnosis module.

[0030] Data collection module: Acquires the impact data that needs to be monitored for the transformer through a large database. The impact data represents the data that affects the safe and stable operation of the transformer. Based on the impact data, a multi-level state feature layer is divided, and each state feature layer includes the corresponding observation data.

[0031] Parameter generation module: Generates corresponding stability indices and entropy weights for the state feature layer based on the observed data;

[0032] Anomaly Diagnosis Module: After weighted fusion of the stability indices of all state characteristic layers in the transformer, the module obtains the comprehensive characteristic value of the transformer. Based on the comprehensive characteristic value, the module performs anomaly diagnosis on the transformer and generates corresponding management strategies in combination with the anomaly diagnosis results. The management strategies are then sent to the transformer management platform.

[0033] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0034] This invention acquires the impact data of transformers that need to be monitored through a large database. Based on the impact data, it divides the transformer into multi-level state feature layers, each including corresponding observation data. A stability index and entropy weight are generated for each state feature layer based on the observation data. After weighted fusion of the stability indices of all state feature layers in the transformer, a comprehensive characteristic value of the transformer is obtained. Anomaly diagnosis is performed on the transformer based on this comprehensive characteristic value, and corresponding management strategies are generated based on the anomaly diagnosis results. By constructing multi-level state feature layers, the diagnostic system can more comprehensively reflect the different influencing factors of transformer operation, improving the detail of monitoring and the accuracy of diagnosis. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0036] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] Example 1: Please refer to Figure 1 As shown in this embodiment, the multi-level state characteristic decomposition and diagnosis method for distribution network transformers includes the following steps:

[0039] The diagnostic system acquires the impact data that needs to be monitored for the transformer through a large database. The impact data represents the data that affects the safe and stable operation of the transformer. Based on the impact data, it divides the transformer into multi-level state feature layers. Each state feature layer includes corresponding observation data (for example, when the impact data is temperature data, the state feature layer is divided into temperature feature layers, which includes observation data such as winding temperature and oil temperature). Based on the observation data, it generates corresponding stability indices and entropy weights for the state feature layers (the larger the stability index, the more stable the state feature layer is, and the smaller the impact on the stable operation of the transformer; the larger the entropy weight, the greater the uncertainty of the state feature layer). After weighted fusion of the stability indices of all state feature layers in the transformer, it obtains the comprehensive feature value of the transformer. Based on the comprehensive feature value, it performs anomaly diagnosis on the transformer and generates corresponding management strategies based on the anomaly diagnosis results. The management strategies are then sent to the transformer management platform.

[0040] This application acquires the impact data of transformers that need to be monitored through a large database. Based on the impact data, it divides the transformer into multi-level state characteristic layers, each including corresponding observation data. A stability index and entropy weight are generated for each state characteristic layer based on the observation data. After weighted fusion of the stability indices of all state characteristic layers in the transformer, a comprehensive characteristic value of the transformer is obtained. Anomaly diagnosis is performed on the transformer based on this comprehensive characteristic value, and corresponding management strategies are generated based on the anomaly diagnosis results. By constructing multi-level state characteristic layers, the diagnostic system can more comprehensively reflect the different influencing factors of transformer operation, improving the detail of monitoring and the accuracy of diagnosis.

[0041] Example 2: The diagnostic system obtains the impact data that needs to be monitored for the transformer through a large database. The impact data represents data that affects the safe and stable operation of the transformer. Based on the impact data, a multi-level state feature layer is divided, including the following steps:

[0042] The diagnostic system obtains the impact data that needs to be monitored for the transformer through a large database. The impact data includes temperature data, resistance data, current data, and vibration data.

[0043] Based on the influence data, a multi-level state feature layer is divided, based on temperature data, based on resistance data, based on current data, and based on vibration data, a vibration state feature layer is divided.

[0044] Impact data obtained from large databases is used for comprehensive monitoring and analysis of transformer conditions. This impact data can be categorized into multiple types (such as temperature data, resistance data, current data, vibration data, etc.), with each type corresponding to a state characteristic layer. Each state characteristic layer contains all observation points within that data type. The following is an example illustrating the multi-level state characteristic layers based on this data:

[0045] 1) Temperature state characteristic layer:

[0046] Data source: Various temperature-related monitoring data from transformers.

[0047] 1.1) Observational data:

[0048] Winding temperature: Monitors the temperature of the windings, reflecting their thermal state.

[0049] Oil temperature: Monitors the temperature of transformer oil, reflecting the effectiveness of the cooling system and the heat dissipation of the transformer.

[0050] Core temperature: Monitor the temperature of the core to assess whether it is overheating.

[0051] Bushing temperature: Monitor the temperature of the bushing to assess its insulation condition.

[0052] Function of the Condition Characteristic Layer: This layer assesses the thermal condition of the transformer using temperature data. Abnormal temperatures may indicate problems such as winding overload, insufficient oil cooling, or core overheating.

[0053] 2) Resistance state characteristic layer:

[0054] Data source: Resistance-related monitoring data, mainly used to detect the health status of transformer windings and insulation.

[0055] 2.2) Observational data:

[0056] Winding resistance: Monitor the resistance of the winding to detect whether it has been damaged or aged due to overheating or excessive current.

[0057] Insulation resistance: Monitor the insulation resistance of transformer windings, bushings and other components to assess the insulation condition.

[0058] Function of the condition characteristic layer: This layer assesses the electrical condition of the transformer through resistance data. Changes in resistance may indicate problems such as short circuits in the windings, poor contact, or insulation aging.

[0059] 3) Current state characteristic layer:

[0060] Data source: Current-related monitoring data used to assess transformer load conditions and current anomalies.

[0061] 3.3) Observational data:

[0062] Winding current: Monitor the winding current to determine if there is an overload, short circuit or current imbalance.

[0063] Load current: reflects the load condition of the transformer and is usually used to detect the trend of load changes.

[0064] Function of the status characteristic layer: This layer evaluates the current status of the transformer. Excessive current may indicate problems such as overload or short circuit, while unbalanced current may indicate system instability.

[0065] 4) Vibration state characteristic layer:

[0066] Data source: Vibration monitoring data of various components in the transformer, mainly used to assess the stability of the transformer's mechanical structure.

[0067] 4.4) Observational data:

[0068] Winding vibration: Monitor the vibration of the windings and check for looseness or unevenness.

[0069] Core vibration: Monitor whether the core is vibrating or loose due to foreign objects.

[0070] Overall structural vibration: Monitor the vibration of the entire transformer structure and assess its mechanical stability.

[0071] Function of the state characteristic layer: This layer assesses the mechanical stability of the transformer through vibration data. Abnormal vibration may indicate problems such as loose windings, damaged core, or structural instability.

[0072] Each state feature layer includes corresponding observation data (for example, when the influencing data is temperature data, the divided state features are temperature feature layers, and the temperature feature layers include observation data such as winding temperature and oil temperature), including the following steps:

[0073] Each state characteristic layer includes corresponding observation data. The observation data for the temperature state characteristic layer includes winding temperature, cooling oil temperature, and core temperature. The observation data for the resistance state characteristic layer includes winding resistance, bushing resistance, and contact resistance. The observation data for the current state characteristic layer includes winding current and load current. The observation data for the vibration state characteristic layer includes winding vibration, core vibration, and oil pump vibration.

[0074] 1. Temperature state characteristic layer observation data:

[0075] Winding temperature: Winding temperature refers to the operating temperature of the transformer windings. Excessively high winding temperatures can lead to aging and damage of the winding insulation materials, and may even cause fires or electrical faults. Therefore, winding temperature is a critical monitoring data point.

[0076] Cooling oil temperature: The temperature of the cooling oil directly affects the cooling effect of the transformer, thus affecting its operating efficiency and safety. If the cooling oil temperature is too high, it may indicate a cooling system malfunction or insufficient oil volume, limiting the transformer's heat dissipation capacity.

[0077] Core temperature: Changes in the core temperature reflect the thermal state of the transformer's core components. Excessively high core temperature may be due to overload, short circuit, or other abnormal causes leading to core damage and further affecting the transformer's performance.

[0078] Function: The observation data from the temperature condition characteristic layer is used to monitor the thermal condition of the transformer. If the temperature of a certain part is too high (such as winding temperature, oil temperature, etc.), it indicates that there may be a fault in that part or that maintenance is required.

[0079] For example, if the winding temperature reaches 90℃, while the cooling oil temperature is 85℃ and the core temperature is close to 80℃, it may mean that the cooling oil temperature is rising too quickly and the cooling effect is decreasing. The oil level and cooling system need to be checked.

[0080] 2. Observation data of the resistance state characteristic layer:

[0081] Winding resistance: Winding resistance is used to monitor the health of the windings. Resistance that is too low or too high may indicate aging of the winding insulation or a short circuit.

[0082] Bushing resistance: The resistance of the bushing is used to monitor the insulation performance of the transformer bushing. Changes in resistance reflect whether the bushing insulation material has aged or broken down.

[0083] Contact resistance: The resistance of tap changers or other mechanical contacts. Poor contact can lead to increased resistance, resulting in overheating or a risk of electrical fire.

[0084] Purpose: The observation data from the resistance condition characteristic layer is primarily used to assess the electrical health of transformers. Abnormal changes in resistance values ​​may indicate electrical faults such as winding insulation problems or bushing aging.

[0085] For example, if the winding resistance fluctuates significantly and the bushing resistance continues to rise, it may indicate that the insulation system is aging or that there is partial discharge.

[0086] If the contact resistance gradually increases, it may indicate a problem with the tap changer's contacts. Poor contact may lead to unstable current in the transformer.

[0087] 3. Observation data of current state characteristic layer:

[0088] Winding current: The current in the transformer windings can help determine the load condition and whether an overload has occurred. Excessive winding current may be a sign of overload, or it could be caused by a short circuit or poor contact.

[0089] Load current: The load current reflects the actual load on the transformer. Under normal circumstances, the load current should match the rated load of the transformer. If the load current is too high, it may mean that the transformer is overloaded or that the load distribution is uneven.

[0090] Function: The current status feature layer is used to monitor the load condition and current fluctuations of the transformer. Excessive or uneven current may cause equipment damage or failure.

[0091] For example, if the winding current consistently exceeds the rated value and the load current fluctuates significantly, it may indicate that the transformer is facing an overload risk or that the load distribution is uneven. The load should be adjusted or the transformer should be shut down for inspection as soon as possible. If the winding current and the load current are inconsistent, it may indicate an internal imbalance within the transformer, requiring inspection for winding damage or poor contact.

[0092] 4. Observation data of vibration state characteristic layer:

[0093] Winding vibration: Vibration data of the windings helps determine whether they have become loose or damaged. Abnormal vibration may be a signal that internal components of the transformer are loose, damaged, or that other mechanical faults have occurred.

[0094] Core vibration: Core vibration monitoring is used to detect mechanical problems or magnetic saturation. Excessive vibration can damage the core structure and affect the transformer's performance.

[0095] Oil pump vibration: Monitoring oil pump vibration can determine whether the cooling system is operating normally. Abnormal oil pump vibration may indicate a malfunction in the oil pump or poor oil flow, which may lead to a decrease in cooling efficiency.

[0096] Function: The vibration condition characteristic layer is used to assess the mechanical health of a transformer. Excessive or irregular vibrations usually indicate mechanical failure or loose components.

[0097] For example, if the amplitude of the detected winding vibration suddenly increases, and the core vibration also shows large fluctuations, it may indicate a mechanical structural problem in the transformer, such as a loose core or a mechanical fault in the winding. Increased oil pump vibration may be a warning signal of oil pump failure, requiring inspection of the oil pump's operating status and whether the oil flow is smooth.

[0098] The process of generating corresponding entropy weights for the state feature layer based on observation data includes the following steps:

[0099] Each state characteristic layer includes corresponding observation data. The observation data for the temperature state characteristic layer includes winding temperature, cooling oil temperature, and core temperature. The observation data for the resistance state characteristic layer includes winding resistance, bushing resistance, and contact resistance. The observation data for the current state characteristic layer includes winding current and load current. The observation data for the vibration state characteristic layer includes winding vibration, core vibration, and oil pump vibration.

[0100] The duration of occurrence of small, normal, and large observation data for any component in the state feature layer observation data is obtained. The duration of occurrence of small, normal, and large observation data is summed to obtain the total duration. The probability of small observation data is obtained by dividing the duration of small observation data by the total duration, the probability of normal observation data is obtained by dividing the duration of normal observation data by the total duration, and the probability of large observation data is obtained by dividing the duration of large observation data by the total duration.

[0101] The weights of components are calculated based on the probabilities of undersized observations, normal observations, and oversized observations. The expression is as follows:

[0102] H=-(p(x1)log(p(x1))+p(x2)log(p(x2))+p(x3)log(p(x3))); Eq.

[0103] In this context, H represents the weights, p(x1) is the probability of smaller observed data, p(x2) is the probability of normal observed data, and p(x3) is the probability of larger observed data. Since the state feature layer includes multiple components, the entropy weight of the state feature layer is obtained by summing the weights of the multiple components. The expression is as follows:

[0104] In the formula, SH is the entropy weight, n is the number of components included in the state feature layer, and H... i The weight of the i-th component is denoted as . The larger the entropy weight of the state feature layer, the greater the uncertainty of that state feature layer, which in turn has a greater impact on the transformer.

[0105] To better illustrate the above solution, the following example is provided in this application:

[0106] Taking the winding temperature in the temperature state feature layer as an example, the winding temperature usually has a preset standard temperature range (T). min ~T max If the current winding temperature is within the standard temperature range, it is recorded as standard observation data; if the current winding temperature is less than the first temperature threshold T, it is recorded as standard observation data. min If the current temperature of the winding is greater than the second temperature threshold T, it will be recorded as a smaller observed data point. max If so, it will be recorded as an oversized observation;

[0107] For example, the transformer winding temperature is preset with a first temperature threshold of 85 degrees Celsius and a second temperature threshold of 105 degrees Celsius. When the current winding temperature is greater than or equal to 85 degrees Celsius and less than or equal to 105 degrees Celsius, it is recorded as general observation data, and so on.

[0108] The generation methods for the observation data of the resistance state characteristic layer and the current state characteristic layer are the same as those for the temperature state characteristic layer, and will not be described in detail in this application.

[0109] The acquisition method of observation data in the vibration state characteristic layer is different from that in the temperature state characteristic layer. This is because in the transformer diagnosis process, the vibration amplitude of the components is usually better the smaller it is. Therefore, taking the core vibration in the vibration state characteristic layer as an example, there are preset first vibration threshold and second vibration threshold. If the current core vibration is less than or equal to the first vibration threshold, it indicates that there is no vibration abnormality and is recorded as low observation data. If the current core vibration is greater than the first vibration threshold and less than or equal to the second vibration threshold, it indicates that there is slight vibration and is recorded as general observation data. If the current core vibration is greater than the second vibration threshold, it indicates that there is severe vibration and is recorded as high observation data.

[0110] The duration of occurrence of general observation data is obtained by summing the duration of occurrence of each occurrence of general observation data, the duration of occurrence of slightly smaller observation data is obtained by summing the duration of occurrence of each occurrence of slightly larger observation data, and the duration of occurrence of slightly larger observation data is obtained by summing the duration of occurrence of each occurrence of larger observation data.

[0111] The process of generating a corresponding stability index for the state feature layer based on observation data includes the following steps:

[0112] Each state characteristic layer includes corresponding observation data. The observation data for the temperature state characteristic layer includes winding temperature, cooling oil temperature, and core temperature. The observation data for the resistance state characteristic layer includes winding resistance, bushing resistance, and contact resistance. The observation data for the current state characteristic layer includes winding current and load current. The observation data for the vibration state characteristic layer includes winding vibration, core vibration, and oil pump vibration.

[0113] The process involves obtaining the duration of occurrence of small, normal, and large observation data for any component in the state feature layer observation data. The total duration is obtained by summing these durations. The probability of small observation data is obtained by dividing the duration of small observation data by the total duration, the probability of normal observation data is obtained by dividing the duration of normal observation data by the total duration, and the probability of large observation data is obtained by dividing the duration of large observation data by the total duration.

[0114] The stable value of the component is calculated based on the probability of undersized observation data, the probability of normal observation data, and the probability of undersized observation data. The expression is as follows: In the formula, W is the stable value, p(x1) is the probability of undersized observation data, p(x2) is the probability of normal observation data, and p(x3) is the probability of undersized observation data. The expression for calculating the stability index of the state feature layer is:

[0115] In the formula, SW is the stability index, n is the number of components included in the state characteristic layer, and W i Let be the stability value of the i-th component. The larger the stability index, the higher the overall stability of the state characteristic layer during transformer operation.

[0116] The above formula for calculating component stability values ​​only applies to temperature state characteristic layers, resistance state characteristic layers, and current state characteristic layers. For vibration state characteristic layers, the expression for calculating component stability values ​​is: In the formula, W is the stable value, p(x1) is the probability of the smaller observed data, p(x2) is the probability of the normal observed data, and p(x3) is the probability of the larger observed data. This is because in the vibration state feature layer, the vibration of the component is usually better the smaller it is.

[0117] After weighted fusion of the stability indices of all state characteristic layers in the transformer, a comprehensive characteristic value of the transformer is obtained. Anomaly diagnosis is performed on the transformer based on the comprehensive characteristic value, and a corresponding management strategy is generated based on the anomaly diagnosis results. The management strategy is sent to the transformer management platform, including the following steps:

[0118] After weighted fusion of the stability indices of all state characteristic layers in the transformer, the comprehensive characteristic value of the transformer is obtained, expressed as: In the formula, tz s The comprehensive characteristic value is SW, where m is the number of state characteristic layers in the transformer. i Let SH be the stability index of the i-th state feature layer. i is the entropy weight of the i-th state feature layer. The larger the comprehensive feature value, the better the overall health of the surface transformer.

[0119] The comprehensive feature value is compared with a preset health threshold. The health threshold is used to diagnose whether the transformer is abnormal. If the comprehensive feature value is greater than or equal to the health threshold, the transformer is diagnosed as not being abnormal. If the comprehensive feature value is less than the health threshold, the transformer is diagnosed as being abnormal.

[0120] When an anomaly is detected in the transformer, the corresponding management strategy generated includes:

[0121] If the overall anomaly is caused by overload or unbalanced load, load regulation measures should be taken immediately, and the load should be transferred from the affected transformer to standby equipment or other healthy equipment through the load distribution system.

[0122] If the load is too large and the transformer is overloaded, the load disconnection protection mechanism should be activated to suspend or reduce the load operation to prevent the temperature from rising continuously and the equipment from being damaged by overload.

[0123] If the assessment results indicate that the overall anomaly has posed a significant threat to the health of the transformer (e.g., both temperature and resistance anomalies may occur, potentially leading to insulation aging or internal short circuits), an emergency shutdown should be performed for a comprehensive inspection.

[0124] Depending on the type of anomaly, the abnormal transformer can be isolated to prevent it from affecting the entire power distribution network.

[0125] If the transformer fails to return to normal operation, a backup transformer or backup power supply should be activated to ensure the stability of the power supply system.

[0126] Based on the diagnostic results, the resulting management strategy should respond quickly, accurately pinpoint the cause of the anomaly, and take practical and feasible measures. By scheduling loads, testing equipment, activating backup plans, or performing maintenance, transformer failures should be minimized, and the impact on the system reduced. A sound management strategy helps improve transformer operational safety, extend its service life, and reduce maintenance costs.

[0127] Example 3: The multi-level state characteristic decomposition and diagnosis system for distribution network transformers described in this example includes a data collection module, a parameter generation module, and an anomaly diagnosis module;

[0128] Data collection module: Acquires the impact data that needs to be monitored for the transformer through a large database. Impact data represents data that affects the safe and stable operation of the transformer. Based on the impact data, multi-level state feature layers are divided. Each state feature layer includes corresponding observation data (for example, when the impact data is temperature data, the state feature is divided into a temperature feature layer, which includes observation data such as winding temperature and oil temperature). The multi-level state feature layers and the corresponding observation data are sent to the parameter generation module.

[0129] Parameter generation module: Based on the observation data, it generates the corresponding stability index and entropy weight for the state feature layer (the larger the stability index, the more stable the state feature layer is, and the smaller the impact on the stable operation of the transformer; the larger the entropy weight, the greater the uncertainty of the state feature layer). The stability index and entropy weight are sent to the anomaly diagnosis module.

[0130] Anomaly Diagnosis Module: After weighted fusion of the stability indices of all state characteristic layers in the transformer, the module obtains the comprehensive characteristic value of the transformer. Based on the comprehensive characteristic value, the module performs anomaly diagnosis on the transformer and generates corresponding management strategies in combination with the anomaly diagnosis results. The management strategies are then sent to the transformer management platform.

[0131] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0132] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0133] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A multi-level state characteristic decomposition and diagnosis method for distribution network transformers, characterized in that: The diagnostic method includes the following steps: The diagnostic system obtains the impact data that needs to be monitored for the transformer through a large database. The impact data represents the data that affects the safe and stable operation of the transformer. Based on the impact data, a multi-level state feature layer is divided, and each state feature layer includes the corresponding observation data. Based on the observation data, a corresponding stability index and entropy weight are generated for the state feature layer, including: obtaining the duration of occurrence of small observation data, normal observation data, and large observation data for any component in the state feature layer observation data; summing the duration of occurrence of small observation data, normal observation data, and large observation data to obtain the total duration; dividing the duration of occurrence of small observation data by the total duration to obtain the probability of small observation data; dividing the duration of occurrence of normal observation data by the total duration to obtain the probability of normal observation data; and dividing the duration of occurrence of large observation data by the total duration to obtain the probability of large observation data. The stable value of the component is calculated based on the probability of undersized observation data, the probability of normal observation data, and the probability of undersized observation data. The expression is as follows: ， In the formula, For stable values, To account for the probability of underestimating the observed data, For general observation data probability, To inflate the probability of observed data, the expression for calculating the stability index of the state feature layer is: ， In the formula, To stabilize the index, The number of components included in the state feature layer. For the first The stability value of each component; the larger the stability index, the higher the overall stability of the state characteristic layer during transformer operation. The weights of components are calculated based on the probabilities of undersized observations, normal observations, and oversized observations. The expression is as follows: ； In the formula, As weight, To account for the probability of underestimating the observed data, For general observation data probability, To inflate the probability of observed data, since the state feature layer includes multiple components, the entropy weight of the state feature layer is obtained by summing the weights of the multiple components. The expression is as follows: ， In the formula, For entropy weights, The number of components included in the state feature layer. For the first The weight of each component; After weighted fusion of the stability indices of all state characteristic layers in the transformer, a comprehensive characteristic value of the transformer is obtained. Anomaly diagnosis is performed on the transformer based on this comprehensive characteristic value, and a corresponding management strategy is generated based on the anomaly diagnosis results. The management strategy is then sent to the transformer management platform. Obtaining the comprehensive characteristic value of the transformer includes the following steps: After weighted fusion of the stability indices of all state characteristic layers in the transformer, the comprehensive characteristic value of the transformer is obtained, expressed as: ， In the formula, For comprehensive eigenvalues, This represents the number of state characteristic layers in the transformer. For the first Stability index of each state feature layer For the first The entropy weights of each state feature layer.

2. The method for multi-level state characteristic decomposition and diagnosis of distribution network transformers according to claim 1, characterized in that: Transformer anomaly diagnosis based on comprehensive characteristic values ​​includes the following steps: The comprehensive feature value is compared with a preset health threshold. The health threshold is used to diagnose whether the transformer is abnormal. If the comprehensive feature value is greater than or equal to the health threshold, the transformer is diagnosed as not being abnormal. If the comprehensive feature value is less than the health threshold, the transformer is diagnosed as being abnormal.

3. The method for multi-level state characteristic decomposition and diagnosis of distribution network transformers according to claim 2, characterized in that: The diagnostic system obtains the impact data that needs to be monitored for the transformer from a large database, including the following steps: The diagnostic system obtains the impact data that needs to be monitored for the transformer through a large database. The impact data includes temperature data, resistance data, current data, and vibration data. Based on the influence data, a multi-level state feature layer is divided, based on temperature data, based on resistance data, based on current data, and based on vibration data, a vibration state feature layer is divided.

4. The method for multi-level state characteristic decomposition and diagnosis of distribution network transformers according to claim 3, characterized in that: Each state characteristic layer includes corresponding observation data. The observation data for the temperature state characteristic layer includes winding temperature, cooling oil temperature, and core temperature. The observation data for the resistance state characteristic layer includes winding resistance, bushing resistance, and contact resistance. The observation data for the current state characteristic layer includes winding current and load current. The observation data for the vibration state characteristic layer includes winding vibration, core vibration, and oil pump vibration.

5. A multi-level state characteristic decomposition diagnostic system for distribution network transformers, used to implement the diagnostic method described in any one of claims 1-4, characterized in that: It includes a data collection module, a parameter generation module, and an anomaly diagnosis module; Data collection module: Acquires the impact data that needs to be monitored for the transformer through a large database. The impact data represents the data that affects the safe and stable operation of the transformer. Based on the impact data, a multi-level state feature layer is divided, and each state feature layer includes the corresponding observation data. Parameter generation module: Generates corresponding stability indices and entropy weights for the state feature layer based on the observed data; Anomaly Diagnosis Module: After weighted fusion of the stability indices of all state characteristic layers in the transformer, the module obtains the comprehensive characteristic value of the transformer. Based on the comprehensive characteristic value, the module performs anomaly diagnosis on the transformer and generates corresponding management strategies in combination with the anomaly diagnosis results. The management strategies are then sent to the transformer management platform.

Images