A method for defect identification, condition grading and active safety protection of multi-parameter integrated power transformers

Abstract

The present invention belongs to the technical field of power transformer condition monitoring, evaluation and safety control, and particularly relates to a method for defect identification, state classification and active safety protection of a multi-parameter fusion power transformer, including: obtaining the operating environmental conditions, operating parameters, thermal defect state parameters, discharge defect state parameters and winding deformation defect state parameters of the power transformer by installing intelligent sensors and monitoring devices; multi-source data fusion and value information extraction of the power transformer; online defect identification of the power transformer and tracing of defect causes; state classification, safety margin and safety tolerance capacity evaluation of the power transformer; and active safety regulation and protection based on the safety margin of the power transformer. The present invention can effectively improve the effect of power transformer defect identification, enhance the objectivity and comprehensiveness of power transformer safety assessment, and improve the initiative and pertinence of power transformer safety regulation and protection measures, thereby being conducive to ensuring the safe and stable operation of the power grid and power transformers.

Description

Technical Field

[0001] The invention belongs to the technical field of power transformer condition monitoring, evaluation and safety control, and particularly relates to a method for defect identification, state grading and active safety protection of a multi-parameter fusion power transformer. Background Art

[0002] A power transformer is the most basic electrical equipment for power transmission and transformation, and its operation safety directly affects the operation safety of the entire power system. Applying various technical means to detect the defects existing in the power transformer at an early stage and taking timely defect suppression or elimination measures can effectively avoid major accidents. The operation state of a power transformer is jointly determined by multiple characteristic indexes. Existing national standards and industry standards have detailed specifications for various electrical characteristic parameters of power transformers. However, the existing standards and specifications usually make judgments only under a single performance index, and cannot effectively identify the types of power transformer defects. At the same time, they cannot comprehensively and objectively reflect the safety state and tolerance ability of the power transformer, and cannot provide a reference for power grid regulation and equipment safety protection setting. When there are defects such as thermal defects, discharge defects or winding deformation inside the power transformer, the measured values of various monitored quantities of the power transformer will be different from those of the equipment during normal operation, and the severity of the defect is also related to the measured value or change rate of each monitored quantity. Therefore, configuring intelligent sensors or multi-physical quantity monitoring devices on the power transformer, analyzing and processing the obtained monitoring information, and using it to guide the operation and regulation of the power grid play an important role in ensuring the safe and stable operation of the power transformer and the power grid.

[0003] In view of the problem that the traditional power transformer defect identification method cannot quickly obtain high-value information and detect the potential correlation between the defect information attributes, the Chinese patent with publication number CN105843210A discloses the "Power Transformer Defect Information Data Mining Method", which uses the Apriori algorithm to mine the association rules between the defect-related factors of the power transformer, where the defect-related factors of the power transformer considered include the manufacturer, operating years and defect type. In view of the problem that the existing power transformer fault diagnosis method uses a single means to judge the degree of defects or faults of the power transformer, which cannot fully reflect the defects or faults of the power transformer and is prone to misjudgment, the Chinese patent with publication number CN114111886A discloses "A Power Transformer Defect Online Diagnosis Method Using Operation Information", which judges the type and degree of defects and faults of the power transformer based on the deviation between the measured temperature of the power transformer and the temperature output by the heat accumulation model and the equivalent circuit parameter identification results of the power transformer. The above method introduces limited feature parameters when identifying power transformer defects, lacks necessary supervision when training the identification model, and also does not comprehensively consider the requirements of expert experience and power transformer operation procedures. In addition, although the above method can identify whether there are partial defects in the power transformer, it cannot identify the cause of the defect, nor can it evaluate the safety margin and tolerance of the defective power transformer, and cannot provide guidance for the operation control and protection setting of the power transformer. The Chinese patent with publication number CN113078615A discloses "Large Power Transformer Active Protection Method and Device", which mainly constructs protection action criteria for serious defects and internal faults of power transformers. This method cannot perceive the existence of defects before minor defects of power transformers evolve into serious defects, nor can it identify the cause of defects when there are minor defects in power transformers, which is not conducive to curbing the defects of power transformers in the embryonic stage as soon as possible, and also does not provide an evaluation method for the safety margin and safety tolerance of defective power transformers.

[0004] When there are slight defects inside the power transformer, the power transmission and conversion capacity of the power transformer will not drop sharply. At this time, targeted defect suppression or defect elimination measures can be taken to suppress the speed of development of power transformer defects or even eliminate the defects, thereby preventing the power transformer defects from evolving into faults. However, the existing power transformer safety protection methods are usually conservative fixed value protections that do not take into account the actual operating status of the power transformer, the defect tolerance of the power transformer, or the repairability of defective power transformers. Usually, the power transformer is shut down when it has slight defects, which is not conducive to ensuring the reliability and continuity of power supply to customers. Summary of the invention

[0005] To solve the above technical problems, the present invention proposes a multi-parameter fusion method for defect identification, state classification, and active safety protection of power transformers, including:

[0006] S1: Obtain the operating environmental conditions, operating parameters, thermal defect state parameters, discharge defect state parameters, and winding deformation defect state parameters of the power transformer by installing intelligent sensors and monitoring devices;

[0007] S2: Utilize the multi-source data and defect state information obtained by the sensors and monitoring devices, complete data preprocessing through data format conversion, data cleaning, data noise and duplicate data elimination, and data normalization, realize multi-source data fusion by associating, combining, and integrating the preprocessed data and information, extract the valuable information hidden in the multi-source data by combining rule reasoning and causal mapping relationship matching, and obtain a structured and networked knowledge graph by processing, merging, and linking the relationships and attributes of the valuable information;

[0008] S3: Construct an online defect identification and defect cause tracing decision tree model for power transformers based on the structured and networked knowledge graph, and carry out online defect identification and defect cause tracing of power transformers according to the online defect identification and defect cause tracing decision tree model for power transformers;

[0009] S4: By considering the dynamic superposition effect of the internal and external parameters of the power transformer, combining the change law of the safety performance of the power transformer, and simultaneously considering the tolerance limits of the load rate, winding hot spot temperature, top oil temperature, discharge severity, and winding deformation amount of the power transformer, characterize the safety domain of the power transformer based on multi-parameter feature fusion, establish a hyperplane space safety domain model for the power transformer including a conservative layer, a transition layer, and a dangerous layer, and combine this safety domain model to classify the state of the power transformer according to the monitoring values of the operating parameters and state parameters of the power transformer, and simultaneously evaluate the safety margin and safety tolerance ability of the power transformer;

[0010] S5: When the power transformer has a safety margin and sufficient safety tolerance time, allow the power transformer to continue operating with defects, and simultaneously adopt targeted defect suppression or defect elimination measures; when the safety margin or safety tolerance time of the power transformer is insufficient, the power transformer is shut down and maintenance is arranged.

[0011] Advantages of the present invention: By installing intelligent sensors and multi-physical quantity on-line monitoring devices inside and outside the power transformer, the present invention collects the real-time operation parameters and environmental conditions of the power transformer, senses the temperature and magnetic field distribution inside the power transformer, obtains the thermal, electrical, and acoustic information generated during the operation of the power transformer, and then uses multi-source monitoring information and historical defect data to form a knowledge graph through multi-source data mining, fusion, and key information extraction, and guides the construction of a method for identifying power transformer defects and tracing the causes of defects, further evaluating the defect level, safety margin, and tolerance ability of the power transformer, and finally taking targeted power transformer active safety regulation and protection methods aiming at suppressing and eliminating power transformer defects while ensuring the continuity and reliability of customer power supply; the present invention can effectively improve the state perception ability of the power transformer, improve the effect and intelligent level of power transformer defect identification, enhance the objectivity and comprehensiveness of power transformer safety assessment, and improve the initiative and pertinence of power transformer safety regulation and protection measures, which is conducive to ensuring the safe and stable operation of the power grid and power transformer. Description of the Drawings

[0012] Figure 1 It is a block diagram for implementing the multi-parameter fusion power transformer defect identification, state grading, and active safety protection method of the present invention;

[0013] Figure 2 It is a decision tree model for on-line identification of power transformer defects and tracing the causes of defects with multi-parameter fusion of the present invention;

[0014] Figure 3 It is a flow chart for identifying the causes of power transformer defects based on data mining of the present invention;

[0015] Figure 4 It is a schematic diagram of the safety domain and state grading of the power transformer of the present invention. Detailed Embodiment

[0016] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work shall fall within the protection scope of the present invention.

[0017] In this embodiment, a multi-parameter fusion power transformer defect identification, state grading, and active safety protection method is disclosed;

[0018] As Figure 1 shown, a multi-parameter fusion power transformer defect identification, state grading, and active safety protection method includes:

[0019] S1: Obtain the operating environmental conditions, operating parameters, thermal defect state parameters, discharge defect state parameters, and winding deformation defect state parameters of the power transformer by installing intelligent sensors and monitoring devices;

[0020] S2: Utilize the multi-source data and defect state information obtained by the sensors and monitoring devices, complete data preprocessing through data format conversion, data cleaning, data noise and duplicate data elimination, and data normalization, realize multi-source data fusion by correlating, combining, and integrating the preprocessed data and information, discover and extract the valuable information hidden in the multi-source data by combining rule reasoning and causal mapping relationship matching, and finally form a structured and networked knowledge graph through processing, merging, relationship and attribute linking of the valuable information;

[0021] S3: Construct an online identification decision tree model for power transformer defects and a decision tree model for tracing the causes of defects based on the structured and networked knowledge graph, and carry out online identification of power transformer defects and tracing of the causes of defects according to the online identification decision tree model for power transformer defects and the decision tree model for tracing the causes of defects;

[0022] S4: By considering the dynamic superposition effect of internal and external parameters of the power transformer, combining with the change law of the safety performance of the power transformer, and at the same time considering the tolerance limits of the load rate, winding hot spot temperature, top oil temperature, discharge severity, and winding deformation amount of the power transformer, characterize the safety domain of the power transformer based on multi-parameter feature fusion, establish a hyperplane space safety domain model for the power transformer including a conservative layer, a transition layer, and a dangerous layer, and combine this safety domain model to realize the state classification of the power transformer according to the monitoring values of the operating parameters and state parameters of the power transformer, and at the same time evaluate the safety margin and safety tolerance ability of the power transformer;

[0023] S5: When the power transformer has a safety margin and sufficient safety tolerance time, allow the power transformer to continue operating with defects, and at the same time take targeted defect suppression or defect elimination measures; when the safety margin or safety tolerance time of the power transformer is insufficient, stop the power transformer and arrange for maintenance.

[0024] During the specific implementation process, the intelligent sensors and monitoring devices to be installed include: voltage transformers, current transformers, high-frequency current transformers, ultra-high frequency sensors, acoustic sensors, infrared temperature measurement devices, fiber optic temperature measurement devices, top oil temperature gauges, oil-in-gas monitoring devices, free gas monitoring devices, and magnetic flux leakage monitoring devices; the obtained operating environmental conditions include: the operating environmental temperature of the power transformer, the heat dissipation method and conditions of the power transformer; the obtained operating parameters include: voltage, current (load factor); the obtained thermal defect state parameters include: winding hot spot temperature, top oil temperature, temperature of metal components in contact with fiber insulation materials, hot spot temperature of metal components; the obtained discharge defect state parameters include: discharge high-frequency and ultra-high frequency signals, discharge acoustic signal, type and quantity of discharged gas; the obtained winding deformation defect state parameter is: the magnetic field distribution inside the power transformer.

[0025] During the specific implementation process, the online identification of power transformer defects and the decision tree model for tracing the causes of defects are as Figure 2 shown, including: an input layer, a decision layer, and a result layer;

[0026] The online monitoring data and historical operation data in the input layer mainly include the load factor of the power transformer, harmonic frequency and content, degree of bias magnetic field, heat dissipation conditions, multi-point temperature measurement data, ultra-high frequency signals, content of dissolved gas and free gas in oil, multi-point magnetic flux leakage measured data, structural information, and historical outage and maintenance data;

[0027] The decision layer uses the data and information in the input layer, combines with the pre-classification method of power transformer defect types based on the threshold range of characteristic parameters, as well as the thermal defect identification model, discharge defect identification model, and winding deformation defect identification model of the power transformer, to diagnose the defect types of the power transformer and the causes of the defects;

[0028] During specific implementation, the type and gas content ratio of gas generated inside the power transformer are selected as the characteristic parameters for pre-classifying the power transformer defect types;

[0029] The result layer outputs the power transformer defect types, causes, and severity according to the identification and evaluation results of the decision layer, and obtains the safety margin and safety tolerance of the power transformer.

[0030] During the specific implementation process, in the decision layer of the decision tree model, the pre-classification method of power transformer defect types based on the threshold range of characteristic parameters uses the type and gas content ratio of gas generated inside the power transformer to initially determine whether the defects existing in the power transformer are thermal defects, discharge defects, or both types of defects exist simultaneously, and then traces the causes of the defects for single-type defects respectively;

[0031] The criteria used for pre-classifying defect types include:

[0032]

[0033] Among them, l(C 2 H 2 )、l(C 2 H 4 )、l(CH 4 )、l(H 2 ) respectively represent the unit volume contents of acetylene, ethylene, methane and hydrogen, and ζ 1 、ζ 2 respectively represent the first and second division thresholds.

[0034] Preferably, according to the gas generation test results of power transformer oil and solid insulation materials under different temperatures and different discharge forms, take ζ 1 = 0.1, ζ 2 = 1.

[0035] In the specific implementation process, the power transformer thermal defect identification model includes: the dynamic identification layer of power transformer thermal defects under abnormal conditions, the diagnosis layer of power transformer winding defects based on hierarchical partitioning, and the diagnosis layer of power transformer structural component defects based on hierarchical partitioning;

[0036] The dynamic identification layer of power transformer thermal defects under abnormal conditions obtains the temperature prediction values of each point inside the power transformer through temperature field simulation based on the operation parameters and heat dissipation conditions of the power transformer; compare the multi-point temperature measurement values of the power transformer with the temperature prediction values of the corresponding points. If the temperatures are approximately the same, the overheating of the power transformer is caused by abnormal conditions such as harmonics, bias magnetism, overload, and heat dissipation; if the temperatures are inconsistent, determine whether the reason for the overheating problem of the power transformer is winding overheating or structural component overheating through the diagnosis layer of power transformer winding defects based on hierarchical partitioning and the diagnosis layer of power transformer structural component defects based on hierarchical partitioning; for the overheating problems that cannot be determined by the above methods, classify them as unknown local overheating outside the measuring point area; the diagnosis models of power transformer windings and structural components involved in the diagnosis layer can be trained from the historical thermal defect data of the power transformer;

[0037] The temperature consistency determination rule includes:

[0038]

[0039] Among them, i represents the temperature measurement point number; Θ represents the set of all temperature measurement point numbers; θ est,i represents the temperature prediction value at the measuring point i obtained by temperature field simulation; θ mea,i represents the actual measured temperature value at the measuring point i; ET represents the allowable error tolerance.

[0040] Preferably, considering that the measurement accuracy of the current optical fiber temperature measurement system can be controlled within ±1°C, and taking into account the influence of other interference factors on temperature measurement, the error tolerance ET = 2°C is taken.

[0041] In the specific implementation process, the power transformer discharge defect identification model in the decision-making layer of the decision tree model includes: distinguishing the severity of the discharge by combining the gas production type inside the power transformer and the proportion of gas content, and determining whether the discharge defect belongs to partial discharge, breakdown, or arcing fault; for partial discharge defects, using ultra-high frequency signal pattern recognition to determine the type of partial discharge, and the types of partial discharge include: air gap discharge, surface discharge, floating electrode discharge, and corona discharge.

[0042] The division criteria for the severity of the discharge include:

[0043]

[0044] Among them, l(C 2 H 2 ) and l(C 2 H 4 ) respectively represent the unit volume content of acetylene and ethylene; ζ 1 and ζ 3 respectively represent the first and third division thresholds.

[0045] Preferably, according to the gas production test results of power transformer oil and solid insulation materials under different discharge forms, ζ 1 = 0.1 and ζ 3 = 3 are taken.

[0046] In the specific implementation process, the power transformer winding deformation defect identification model in the decision-making layer of the decision tree model includes: judging whether the power transformer is in abnormal conditions such as inrush current, over-excitation, and DC bias magnetic field that cause magnetic field distortion inside the power transformer based on the operating parameters of the power transformer; considering the magnetic field distortion inside the power transformer caused by abnormal conditions, using the magnetic leakage data of the power transformer collected during historical winding deformation defects to train a winding deformation defect evaluation and positioning model based on magnetic leakage characteristics to determine whether the winding deformation belongs to radial deformation or axial deformation, and at the same time evaluate the maximum short-circuit current withstand capacity of the power transformer under the current state.

[0047] In the specific implementation process, the power transformer thermal defect identification model, discharge defect identification model, and winding deformation defect identification model involved in the decision tree model based on historical defect data mining, such as Figure 3As shown, it is applied to the traceability of the defect causes of power transformers; this method uses the data obtained from the on-line monitoring system of power transformers and the stored historical data to train a defect cause identification model. The model establishes a non-linear mapping relationship between the monitoring data and the defect causes. Inputting the real-time monitoring data into the model can diagnose the defect causes of power transformers, providing guidance for formulating targeted defect suppression measures.

[0048] In the specific implementation process, the safety domain model of the power transformer in the hyperplane space is characterized in the form of a radar chart, as Figure 4 shown; then, combined with this safety domain model, the state classification of the power transformer is realized according to the operation parameters and the monitoring values of the state parameters of the power transformer, and at the same time, the safety margin and the safety tolerance ability of the power transformer are evaluated.

[0049] Combined with the safety domain model of the power transformer in the hyperplane space, the state classification of the power transformer is realized according to the operation parameters and the monitoring values of the state parameters of the power transformer, including:

[0050] Establish severity evaluation functions for thermal defects, discharge defects, and winding deformation defects of power transformers with multi-parameter fusion. By setting the threshold ranges of each evaluation function, the safety domain hyperplane space is divided into three layers: the conservative layer, the transition layer, and the dangerous layer; according to the real-time operation parameters of the power transformer and the monitoring data of each state parameter, calculate the values of the severity evaluation functions of each defect, and then determine the safety layer where the power transformer is located; when the values of the evaluation functions of various defects of the power transformer are all within the conservative layer, there are no defects inside the power transformer and it is in a normal operation state; when there is a certain defect in the power transformer, the value of the severity evaluation function corresponding to the defect will be within the transition layer; when the internal defect of the power transformer will turn into or has developed into a fault after a short time, the value of the severity evaluation function corresponding to the defect will be within the dangerous layer;

[0051] Among them:

[0052] Defect severity evaluation function:

[0053] Conservative layer:

[0054] Transition layer:

[0055] Dangerous layer:

[0056] Among them, T, A, and W respectively represent the severity evaluation functions of thermal defects, discharge defects, and winding deformation defects of power transformers; x 1 …x m are the normalized monitoring values of each state parameter; the matrix λ is the contribution degree matrix of the state monitoring quantity to the defect; T tr 、Atr , W tr respectively represent the critical values of the severity function for the transformation of thermal defects, discharge defects, and winding deformation defects from the conservative layer to the transition layer; T cr , A cr , W cr respectively represent the critical values of the severity function for the transformation of thermal defects, discharge defects, and winding deformation defects from the transition layer to the dangerous layer; (T, A, W) represents the point in the safety domain hyperplane space determined by the values of each defect evaluation function; Ω 正常 , Ω 过渡 , Ω 危险 respectively represent the regions corresponding to the conservative layer, the transition layer, and the dangerous layer in the safety domain hyperplane space, which are represented by the sets of all points satisfying the respective division threshold conditions.

[0057] In the specific implementation process, in combination with this safety domain model, the safety margin and safety tolerance ability of the power transformer are evaluated based on the operating parameters and state parameter monitoring values of the power transformer, including:

[0058] The safety margin of the power transformer is characterized by the difference between the critical values of the severity functions of each defect for the transition from defect to fault and the current evaluation values of the severity functions of each defect of the power transformer; the safety tolerance ability of the thermal defects and discharge defects of the power transformer is reflected by the safety tolerance time of the power transformer. The safety tolerance time of the power transformer under thermal defects and discharge defects is determined by the time required for the corresponding defect severity function of the power transformer to transition from the current value to the critical value of the severity function equal to the transition from the transition layer to the dangerous layer of the power transformer, and is obtained by using the numerical integration method; when the power transformer has both thermal defects and discharge defects at the same time, the overall safety tolerance time of the power transformer is determined by the minimum value of the thermal defect safety tolerance time and the discharge defect safety tolerance time; the safety tolerance ability of the winding deformation defect of the power transformer is measured by the magnitude of the maximum short-circuit current that the power transformer can withstand;

[0059] Among them:

[0060] Safety margin:

[0061] Thermal defect safety tolerance time:

[0062] Discharge defect safety tolerance time:

[0063] Overall safety tolerance time of the power transformer when there are both thermal defects and discharge defects: t s = min{t s,T , t s,A}

[0064] Among them, SM T (tn )、SM A (t n )、SM W (t n ) respectively represent the safety margins of thermal defects, discharge defects, and winding deformation defects evolving into faults at time t; T(t n ), A(t n ), W(t n ) respectively represent the values of the severity evaluation functions of thermal defects, discharge defects, and winding deformation defects of the power transformer at time t n ; T n , A cr , W cr respectively represent the critical values of the severity functions of thermal defects, discharge defects, and winding deformation defects transitioning from the transition layer to the dangerous layer; Δt cr , Δt T respectively represent the integration step sizes used when calculating the safety tolerance times of thermal defects and discharge defects; ΔT and ΔA respectively represent the changes in the severity functions of thermal defects and discharge defects of the power transformer within a single integration step; f A T [x(t n ), T(t n )], f A [x(t n ), A(t n )] respectively represent the relationships between the rates of change of the severity functions of thermal defects and discharge defects at time t n and the current state monitoring information and the severity of the defects; T(t 0 ), A(t 0 ) respectively represent the values of the severity functions of thermal defects and discharge defects at the initial moment; t s,T , t s,A respectively represent the safety tolerance times of thermal defects and discharge defects of the power transformer; t s represents the overall safety tolerance time of the power transformer.

[0065] Preferably, considering that the evolution rate of thermal defects is relatively slow, to ensure the calculation accuracy of the safety tolerance time, the value range of the integration step size Δt T for thermal defects is 1 min to 5 min; considering that the evolution rate of discharge defects is relatively fast, the value range of the integration step size Δt A for discharge defects is 0.1 s to 10 s; the relationship between the rate of change of the thermal defect severity function and the current state monitoring information and the defect severity f T [x(t n ), T(t n ​)]It can be derived by considering the influence of abnormal working conditions and defects on the basis of the temperature rise model of power transformers; the change rate of the discharge defect severity function varies with the current state monitoring information and the defect severity, f A [x(t n ), A(t n )] can be obtained by parameter fitting using historical defect data.

[0066] In the specific implementation process, the targeted defect suppression or elimination measures for power transformers include:

[0067] Combined with the identification results of the causes of power transformer defects, for the thermal defects of power transformers, forced heat dissipation, spraying water, placing ice cubes, load transfer or load shedding, harmonic compensation, and neutral point regulation measures are taken to suppress the development of thermal defects of power transformers; for the discharge defects of power transformers, through active load reduction and insulation medium purification, the development of discharge defects into severity is avoided; for the winding deformation defects of power transformers, combined with the short-circuit current withstand capacity of the windings, through unit rescheduling, network reconstruction or installation of current-limiting equipment, the deterioration of the winding deformation defects of power transformers is avoided.

[0068] In the specific implementation process, when the safety margin or safety tolerance time of the power transformer is insufficient, the triggering conditions for shutting down the power transformer and arranging maintenance include:

[0069] SM T (t n ) ≤ 0 or SM A (t n ) ≤ 0 or SM W (t n ) ≤ 0 or t s ≤ 0

[0070] Among them, SM T (t n ), SM A (t n ), and SM W (t n ) respectively represent the safety margins of thermal defects, discharge defects, and winding deformation defects from evolving into faults; t s represents the overall safety tolerance time of the power transformer.

[0071] The present invention can comprehensively perceive the operating conditions of power transformers, cooperate with data mining, fusion and value information extraction, improve the effect of power transformer defect identification and its intelligent level, enhance the objectivity and comprehensiveness of power transformer safety assessment, improve the initiative and pertinence of power transformer safety regulation and protection measures, and is conducive to ensuring the safe and stable operation of the power grid and power transformers.

[0072] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-parameter fusion power transformer defect identification, state classification and active safety protection method, It is characterized in that include: S1: Obtain the operating environment conditions, operating parameters, thermal defect state parameters, discharge defect state parameters and winding deformation defect state parameters of the power transformer by installing intelligent sensors and monitoring devices; S2: Using multi-source data and defect status information obtained by sensors and monitoring devices, complete data preprocessing through data format conversion, data cleaning, data noise and duplicate data elimination, and data normalization. Multi-source data fusion is achieved by associating, combining, and integrating the preprocessed data and information. Combined with rule reasoning and causal mapping relationship matching, the value information implicit in the multi-source data is extracted. By processing, merging, and linking relationships and attributes of the value information, a structured and networked knowledge graph is obtained; S3: Based on the structured and networked knowledge graph, a decision tree model for online identification of power transformer defects and defect cause tracing is constructed, and online identification of power transformer defects and defect cause tracing is carried out based on the decision tree model; S4: By considering the dynamic superposition of internal and external parameters of the power transformer, combined with the law of change of the safety performance of the power transformer, and considering the load rate of the power transformer, the hot spot temperature of the winding, the top oil temperature, the severity of the discharge, and the tolerance limit of the winding deformation, the safety domain of the power transformer is characterized based on the fusion of multi-parameter features, and a hyperplane space safety domain model of the power transformer including the conservative layer, the transition layer and the dangerous layer is established. Combined with the safety domain model, the status of the power transformer is graded according to the operating parameters and state parameter monitoring values ​​of the power transformer, and the safety margin and safety tolerance capacity of the power transformer are evaluated at the same time; S5: When the power transformer has a safety margin and sufficient safety tolerance time, the power transformer is allowed to continue operating with defects, and targeted defect suppression or defect elimination measures are taken; when the power transformer has an insufficient safety margin or safety tolerance time, the power transformer is shut down and arranged for maintenance.

2. According to the multi-parameter fusion power transformer defect identification, state classification and active safety protection method described in claim 1, It is characterized in that The S1 specifically includes: Intelligent sensors and monitoring devices, including: voltage transformers, current transformers, voiceprint sensors, infrared temperature measuring devices, optical fiber temperature measuring devices, top oil thermometers, gas in oil monitoring devices, free gas monitoring devices and magnetic leakage monitoring devices; The operating environment conditions obtained include: the operating environment temperature of the power transformer, the heat dissipation method and heat dissipation conditions of the power transformer; The operating parameters obtained include: voltage and current; The thermal defect state parameters obtained include: winding hot spot temperature, top oil temperature, metal part temperature in contact with fiber insulation material, and metal part hot spot temperature; The acquired discharge defect state parameters include: discharge high frequency and ultra-high frequency signals, discharge soundprint signals, discharge gas generation type and value; The winding deformation defect state parameters obtained are: magnetic field distribution inside the power transformer.

3. A method for defect identification, condition grading, and active safety protection of a multi-parameter fusion power transformer according to claim 1, characterized in that, the on-line identification of power transformer defects and the decision tree model for tracing the causes of defects include: an input layer, a decision layer, and a result layer; the on-line monitoring data and historical operation data in the input layer include the load rate, harmonic frequency and content, bias magnetization degree, heat dissipation condition, multi-point temperature measurement data, ultra-high frequency signal, dissolved gas and free gas content in oil, multi-point magnetic leakage measurement data, structure information, and historical outage and maintenance data of the power transformer; the decision layer uses the data and information in the input layer to pre-divide the types of power transformer defects, and diagnoses the types of power transformer defects and the causes of defects through a power transformer thermal defect identification model, a discharge defect identification model, and a winding deformation defect identification model; the result layer outputs the types, causes, and severity of power transformer defects based on the identification and evaluation results of the decision layer, and obtains the safety margin and safety tolerance of the power transformer.

4. A method for defect identification, condition grading, and active safety protection of a multi-parameter fusion power transformer according to claim 3, characterized in that, using the data and information in the input layer to pre-divide the types of power transformer defects, including: using the types of gas generated inside the power transformer and the proportion of gas content to preliminarily determine whether the defects existing in the power transformer are thermal defects, discharge defects, or both types of defects exist at the same time, and then tracing the causes of defects for single-type defects respectively; the criteria for pre-dividing defect types include: where l(C 2 H 2 ), l(C 2 H 4 ), l(CH 4 ), and l(H 2 ) respectively represent the unit volume contents of acetylene, ethylene, methane, and hydrogen, and ζ 1 , ζ 2 respectively represent the first and second partitioning thresholds.

5. A method for defect identification, condition grading, and active safety protection of a multi-parameter fusion power transformer according to claim 3, characterized in that, the power transformer thermal defect identification model includes: a dynamic identification layer for power transformer thermal defects under abnormal conditions, a diagnosis layer for power transformer winding defects based on hierarchical partitioning, and a diagnosis layer for power transformer structural component defects based on hierarchical partitioning; the dynamic identification layer for power transformer thermal defects under abnormal conditions obtains the temperature prediction values of each point inside the power transformer through temperature field simulation according to the operation parameters and heat dissipation conditions of the power transformer; comparing the multi-point temperature measurement values of the power transformer with the temperature prediction values of the corresponding points, if the temperatures are approximately the same, the overheating of the power transformer is caused by abnormal conditions such as harmonics, bias magnetization, overload, and heat dissipation; if the temperatures are inconsistent, determine whether the cause of the overheating problem of the power transformer is winding overheating or structural component overheating through the diagnosis layer for power transformer winding defects based on hierarchical partitioning and the diagnosis layer for power transformer structural component defects based on hierarchical partitioning; for the overheating problems that cannot be determined by the above method, classify them as unknown local overheating outside the measurement point area; the temperature consistency determination rule includes: where \(i\) represents the temperature measurement point number; \(\Theta\) represents the set of all temperature measurement point numbers; \(\theta\) est,i represents the predicted temperature value at the measurement point \(i\) obtained from the temperature field simulation; \(\theta\) mea,i represents the actual measured temperature value at the measurement point \(i\); \(ET\) represents the allowable error tolerance.

6. A method for defect identification, condition grading, and active safety protection of a multi-parameter fusion power transformer according to claim 3, characterized in that, The power transformer discharge defect identification model includes: differentiating the severity of the discharge by combining the types of gas generated inside the power transformer and the proportion of gas content, and determining whether the discharge defect belongs to partial discharge, breakdown or arcing fault; for partial discharge defects, using ultra-high frequency signal pattern recognition to determine the type of partial discharge, and the types of partial discharge include: air gap discharge, surface discharge, floating electrode discharge, and corona discharge. The classification criteria for the severity of the discharge include: where l(C 2 H 2 ) and l(C 2 H 4 ) represent the unit volume contents of acetylene and ethylene, respectively; ζ 1 and ζ 3 represent the first and third division thresholds, respectively.

7. A multi-parameter fusion power transformer defect identification, state grading and active safety protection method according to claim 3, characterized in that The power transformer winding deformation defect identification model includes: determining whether the power transformer is in abnormal conditions such as inrush current, over-excitation, and DC bias based on the operating parameters of the power transformer; considering the magnetic field distortion inside the power transformer caused by abnormal conditions, determining whether the winding deformation is radial deformation or axial deformation, and simultaneously evaluating the maximum short-circuit current withstand capacity of the power transformer under the current state.

8. A multi-parameter fusion power transformer defect identification, state grading and active safety protection method according to claim 1, characterized in that Combined with the power transformer hyperplane space safety domain model, grading the state of the power transformer based on the operating parameters of the power transformer and the monitored values of the state parameters, including: Establishing severity evaluation functions for thermal defects, discharge defects, and winding deformation defects of the power transformer with multi-parameter fusion. By setting the threshold ranges of each evaluation function, the safety domain hyperplane space is divided into three layers: a conservative layer, a transition layer, and a dangerous layer; based on the real-time operating parameters of the power transformer and the monitored data of each state parameter, calculating the values of the severity evaluation functions for each defect, and determining the safety layer where the power transformer is located; when the values of the evaluation functions for various defects of the power transformer are all within the conservative layer, there are no defects inside the power transformer and it is in a normal operating state; when there is a certain defect in the power transformer, the value of the severity evaluation function corresponding to the defect will be within the transition layer; when the internal defect of the power transformer will turn into or has developed into a fault after a short time, the value of the severity evaluation function corresponding to the defect will be within the dangerous layer. Wherein: Defect severity assessment function: Conservative layer: Intermediate layer: Dangerous layer: Among them, T, A, and W respectively represent the severity evaluation functions of thermal defects, discharge defects, and winding deformation defects of power transformers; x 1 …x m are the monitored values of each state parameter after normalization; the matrix λ is the contribution degree matrix of the state monitoring quantity to the defect; T tr 、A tr 、W tr respectively represent the critical values of the severity functions of thermal defects, discharge defects, and winding deformation defects when changing from the conservative layer to the transition layer; T cr 、A cr 、W cr respectively represent the critical values of the severity functions of thermal defects, discharge defects, and winding deformation defects when changing from the transition layer to the dangerous layer; (T, A, W) represents the point in the safety domain hyperplane space determined by the values of each defect evaluation function; Ω 正常 、Ω 过渡 、Ω 危险 respectively represent the regions corresponding to the conservative layer, the transition layer, and the dangerous layer in the safety domain hyperplane space, and are represented by the set of all points that meet the respective division threshold conditions.

9. A multi-parameter fusion power transformer defect identification, state grading and active safety protection method according to claim 1, characterized in that Combined with this safety domain model, evaluating the safety margin and safety withstand capacity of the power transformer based on the operating parameters of the power transformer and the monitored values of the state parameters, including: Establish a severity evaluation function for thermal defects, discharge defects, and winding deformation defects of power transformers with multi-parameter fusion. The safety margin of the power transformer is characterized by the difference between the critical values of the severity functions of each defect during the transition from defect to fault and the evaluation values of the current severity functions of each defect of the power transformer. The safety tolerance of the power transformer for thermal defects and discharge defects is reflected by the safety tolerance time of the power transformer. The safety tolerance time of the power transformer under thermal defects and discharge defects is determined by the time required for the corresponding defect severity function of the power transformer to transition from the current value to the critical value of the severity function when the power transformer transitions from the transition layer to the dangerous layer, and is obtained by using the numerical integration method. When the power transformer has both thermal defects and discharge defects at the same time, the overall safety tolerance time of the power transformer is determined by the minimum value of the safety tolerance time of the thermal defect and the safety tolerance time of the discharge defect. The safety tolerance of the winding deformation defect of the power transformer is measured by the magnitude of the maximum short-circuit current that the power transformer can withstand; Wherein: Safety margin: Thermal defect safety tolerance time: Discharge defect safety tolerance time: Overall safe tolerance time of a power transformer with both thermal defects and discharge defects: t s = min{t s,T , t s,A} Among them, SM T (t n )、SM A (t n )、SM W (t n ) respectively represent the safety margins of the thermal defect, discharge defect, and winding deformation defect at time t n evolving into a fault; T(t n ), A(t n ), W(t n ) respectively represent the values of the severity evaluation functions of the thermal defect, discharge defect, and winding deformation defect of the power transformer at time t n ; T cr , A cr , W cr respectively represent the critical values of the severity functions of the thermal defect, discharge defect, and winding deformation defect transitioning from the transition layer to the dangerous layer; Δt T , Δt A respectively represent the integration step sizes used when calculating the safety tolerance times of the thermal defect and the discharge defect; ΔT and ΔA respectively represent the changes in the severity functions of the thermal defect and the discharge defect of the power transformer within a single integration step; f T [x(t n ), T(t n )], f A [x(t n ), A(t n )] respectively represent the relationships between the rates of change of the severity functions of the thermal defect and the discharge defect at time t n and the current state monitoring information and the defect severity; T(t 0 ), A(t 0 ) respectively represent the values of the severity functions of the thermal defect and the discharge defect at the initial moment; t s,T , t s,A respectively represent the safety tolerance times of the thermal defect and the discharge defect of the power transformer; t s represents the overall safety tolerance time of the power transformer.

10. A method for defect identification, state classification, and active safety protection of a multi-parameter fusion power transformer according to claim 1, characterized in that The targeted defect suppression or defect elimination measures for the power transformer include: Combined with the identification results of the defect causes of the power transformer, for the thermal defects of the power transformer, take measures such as forced heat dissipation, water spraying, ice placement, load transfer or load shedding, harmonic compensation, and neutral point regulation to suppress the development of thermal defects of the power transformer; for the discharge defects of the power transformer, avoid the development of discharge defects to severity by active load reduction and insulation medium purification; for the winding deformation defects of the power transformer, combined with the short-circuit current tolerance of the winding, through unit rescheduling, network reconstruction or installation of current-limiting equipment, to avoid the deterioration of the winding deformation defects of the power transformer; When the safety margin or safety tolerance time of the power transformer is insufficient, the triggering conditions for shutting down the power transformer and arranging maintenance include: SM T (t n ) ≤ 0 or SM A (t n ) ≤ 0 or SM W (t n ) ≤ 0 or t s ≤ 0 Among them, SM T (t n )、SM A (t n )、SM W (t n ) respectively represent the safety margins of the thermal defect, discharge defect, and winding deformation defect evolving into a fault; t s represents the overall safety tolerance time of the power transformer.

Images