A transformer partial discharge positioning method based on photoelectromagnetic signal fusion

Abstract

The application discloses a transformer partial discharge positioning method based on photoelectromagnetic signal fusion. The method fuses partial discharge ultrahigh frequency, high frequency current and micro-light three signals, and adaptively selects the optimal positioning strategy through analyzing the ultrahigh frequency preliminary positioning result and the response of different sensors. Specifically, firstly, the arrival time difference method is adopted based on the ultrahigh frequency signal to carry out preliminary positioning, and the three-dimensional model is automatically used to distinguish whether the preliminary discharge point is located inside or outside the winding, and then the multi-end high frequency current positioning or the light-electric combined positioning strategy is selected, and finally, the result is visualized and output in the transformer three-dimensional model. The application can significantly improve the transformer partial discharge positioning precision, and provides a reliable basis for transformer fault diagnosis.

Description

Technical Field

[0001] This invention belongs to the field of power equipment condition monitoring and fault location technology, specifically relating to a transformer partial discharge location method based on opto-electromagnetic signal fusion. Background Technology

[0002] Precise location of partial discharge is a key technical aspect of ensuring the safe and stable operation of power transformers. With the development of equipment condition monitoring and fault diagnosis technologies, partial discharge has been identified as one of the important causes of transformer insulation accidents. Achieving rapid and accurate spatial location of partial discharge helps in timely fault diagnosis and effective prevention.

[0003] However, the internal structure of transformers is complex, and the commonly used methods for partial discharge detection and location in the field all have significant limitations: optical measurement is easily blocked by the medium, has a limited detection range, and is extremely sensitive to the sensor observation position; ultra-high frequency method has low location accuracy due to the refraction, diffraction and attenuation of electromagnetic waves inside the transformer; high frequency current method is mostly used for locating winding defects and identifying discharge phases, but it is difficult to locate the discharge in three-dimensional space.

[0004] In summary, the internal environment of a transformer is complex, and no single detection technology can guarantee the absolute reliability of the positioning results. Therefore, integrating multi-parameter information such as optical, electrical, and magnetic data, and using multiple technologies in synergy to compensate for their respective shortcomings, holds promise for overcoming the bottleneck of single technologies and providing a feasible solution for achieving high-precision and high-reliability positioning of partial discharges. Summary of the Invention

[0005] To overcome the shortcomings of the existing technology, this invention provides a transformer partial discharge location method based on the fusion of three signals: partial discharge micro-light, ultra-high frequency, and high frequency current (optical, electrical, and magnetic). Essentially, it is an intelligent joint location scheme. Its core is to automatically select the best location technology for precise location by analyzing the preliminary location results and the responses of different sensors, so as to solve the problems of complex internal structure of transformers and unreliability of single detection methods.

[0006] According to one aspect of the present invention, a method for locating partial discharge in a transformer based on the fusion of opto-electromagnetic signals is provided, comprising: Step 1: Using partial discharge signals collected by multiple UHF sensors, the UHF positioning equations are constructed and solved using the time difference of arrival algorithm to obtain preliminary positioning coordinates characterizing the location of the discharge source. Step 2: Compare the obtained preliminary positioning coordinates in the three-dimensional structural model of the transformer, and automatically determine whether the position indicated by the preliminary positioning coordinates is located in the internal region of the winding. If it is located inside the winding, start the multi-terminal high-frequency current positioning mode and proceed to Step 3; if it is located outside the winding, start the optical-electric joint positioning process and proceed to Step 4. Step 3: Analyze the time-frequency characteristics of the high-frequency current signal of each phase, and combine it with the winding circuit model to locate the specific phase and turn where the discharge occurs. Use the location result as the target location coordinates and proceed to step 5. Step 4: Check whether all low-light sensors have captured effective partial discharge light signals. Based on the detection results, convert the response state of each low-light sensor into a spatial determination criterion. Use all spatial determination criteria as spatial constraints to resolve and optimize the UHF positioning equations from Step 1 with constraints. Obtain partial discharge positioning coordinates with higher accuracy than the initial positioning coordinates as the target positioning coordinates, and proceed to Step 5. Step 5: Visualize the target location coordinates in the transformer 3D model and output a complete diagnostic report including spatial location, region, and positioning mode.

[0007] As a further technical solution, the number of UHF sensors involved in the positioning in step 1 shall not be less than 4. Before positioning, a three-dimensional rectangular coordinate system shall be established for the transformer, and the coordinates of each UHF sensor shall be recorded. x i , y i , z i The target optimization equation of the UHF positioning equation set is: , Where N is the number of UHF sensors involved in the positioning; e i The signal originates from the power source and propagates to the next... i The difference between the actual propagation distance and the straight-line propagation distance of an ultra-high frequency sensor; c It is the propagation speed of ultra-high frequency signals in transformer oil; t 1 represents the time it takes for the signal to travel from the power source to the first UHF sensor; x i , y i , z i It is the first i Coordinates of an ultra-high frequency sensor; The signal originates from the power source and propagates to the next... i The time difference between the first UHF sensor and the second UHF sensor; x , y , z It is the three-dimensional spatial coordinate of the power source; X min , Y min , Z min, X max , Y max , Z max It is the boundary value of the transformer oil tank.

[0008] As a further technical solution, the automatic determination of whether the preliminary positioning coordinates are located within the winding internal region in step 2 includes: The transformer winding regions are simplified in three-dimensional space as cylindrical geometric models with specific radii and heights. By defining the radial and axial boundary ranges of each cylinder, a mathematical expression for the winding space is constructed. For the initial positioning coordinates, if they satisfy the mathematical condition of being located within the closed interval defined by the geometric model of any winding cylinder, the discharge source is determined to be located inside the winding.

[0009] Furthermore, by traversing all winding cylinder geometric models, if the target point coordinates are located within the closed interval defined by any winding cylinder geometric model, the discharge point is determined to be located inside the winding; otherwise, it is determined to be located outside the winding.

[0010] As a further technical solution, the mathematical conditions within the closed interval defined by the geometric model of the winding cylinder are as follows: , in,( x m , y m ( ) represents the projected coordinates of the center axis of a certain phase transformer winding. R m The outermost radius of the winding. H m, min This is the bottom height. H m, max This is the top height.

[0011] As a further technical solution, the execution of the multi-terminal high-frequency current positioning mode includes: First, an accurate high-frequency equivalent circuit model of the transformer winding is constructed. This model treats the winding as a passive chain network composed of resistors, inductors, capacitance to ground, and inter-turn capacitance, which can be cascaded through several identical or similar basic units to accurately reflect the electrical characteristics of the winding under high-frequency signals.

[0012] Secondly, in the equivalent circuit model, simulated partial discharge power supplies are set at different numbered wire discs, and multi-location and multi-type enumeration injections are performed. After each injection, the high-frequency current response signals at the beginning and end of the transformer winding are acquired, and time-domain and frequency-domain analyses are performed on them respectively to extract various characteristic parameters, including but not limited to time-frequency distribution (such as short-time Fourier transform STFT characteristics), energy, peak value, and time delay.

[0013] Next, high-frequency current characteristic quantities that have a clear ability to distinguish the discharge phase are selected as the basis for determining the discharge phase. At the same time, high-frequency current characteristic quantities that show a monotonically increasing or decreasing change with the increase of the wire plate number are selected. Subsequently, a quantitative functional relationship between "current characteristic quantity - wire plate number" is established by mathematical fitting method.

[0014] Finally, based on the functional relationship obtained from the above fitting, the characteristic quantities corresponding to the actual detected fault current signal are substituted into the equation to calculate the specific coil number where the discharge occurred, thus achieving precise coil-level positioning.

[0015] As a further technical solution, the screening of high-frequency current characteristic quantities that can distinguish between discharge phases includes: First, a feature sample library covering different simulated discharge locations of each phase winding is constructed. Each sample contains multi-dimensional time-frequency features extracted from high-frequency current signals and their corresponding true phase labels. Then, a feature importance evaluation algorithm (such as the random forest algorithm) is applied to quantitatively calculate the contribution of each feature dimension to phase classification, identifying the subset of features that maximizes the separation between different phase samples in the feature space—that is, features with clear discriminative power. These selected features will serve as the core input for determining the discharge phase in the subsequent pattern recognition model.

[0016] As a further technical solution, step 3 employs a mathematical fitting method to establish a quantitative functional relationship between "current characteristic quantity - wire pancake number" and achieves precise positioning at the coil level. The specific method is as follows: From numerous high-frequency current characteristics, characteristic parameters that monotonically change with increasing coil number were selected and denoted as standard parameters. a i All standard parameters constitute the input matrix. A The pie chart number is used as the output vector. B ,Depend on Perform linear regression to obtain weights p and bias q A quantitative functional relationship between "current characteristic quantity and coil number" is obtained. Based on this functional relationship, the characteristic quantity corresponding to the actual detected fault current signal is substituted into the formula, and the specific coil number where the discharge occurred can be calculated in reverse, thus achieving precise coil-level positioning.

[0017] As a further technical solution, step 4, checking whether all micro-light sensors have captured valid partial discharge light signals, includes: Set amplitude threshold V th With time window Δ T If a certain sensor is in ΔT The amplitude of the light pulse signal captured within the time window continuously exceeds V th If the pulse timing is synchronized with the partial discharge pulse detected by the ultra-high frequency, then it is determined that the sensor has captured an effective partial discharge optical signal.

[0018] As a further technical solution, the conversion of the response state of each low-light sensor into a spatial determination criterion includes: Because optical signals attenuate rapidly in transformer oil, the detection range of low-light sensors is limited, and their detection performance is affected by the obstruction of solid media. The effective detection range of each low-light sensor can be considered in three-dimensional space as a region relative to its installation location (…). x j , y j , z j Using the sphere as its center and its maximum effective detection range R light A sphere with radius denoted as is defined. Based on the signal detection results, the following spatial decision constraints are generated: If the first j If a micro-light sensor detects an effective light signal, then the discharge point ( x , y , z The region is located within the spherical domain and has the following constraints: , The set of all sensors that detected valid light signals is: M detected .

[0019] If the first j If a low-light sensor fails to detect a valid light signal, then the discharge point ( x , y , z The sphere is located outside the sphere's domain, and its constraints are: , The set of all sensors that did not detect a valid light signal is: M undetected .

[0020] The decisions made by all sensors will collectively constitute a set of inequality constraints.

[0021] As a further technical solution, step 4 involves resolving and optimizing the UHF positioning equations from step 1 with constraints. The specific target optimization equations are as follows: , In the formula, N It refers to the number of UHF sensors involved in the positioning process; ei The signal originates from the power source and propagates to the next... i The difference between the actual propagation distance and the straight-line propagation distance of an ultra-high frequency sensor; c It is the propagation speed of ultra-high frequency signals in transformer oil; t 1 represents the time it takes for the signal to travel from the power source to the first UHF sensor; x i , y i , z i It is the first i Coordinates of an ultra-high frequency sensor; The signal originates from the power source and propagates to the next... i The time difference between the first UHF sensor and the second UHF sensor; x , y , z () represents the three-dimensional spatial coordinates of the discharge source; X min , Y min , Z min , X max , Y max , Z max These are the boundary values ​​of the transformer oil tank. x j , y j , z j )and( x k , y k , z k ) are respectively the first j The and the first k Coordinates of a low-light sensor; R light This represents the maximum effective detection distance of the low-light sensor. M detected and M undetected These are the sets of sensors that detected and those that did not detect valid light signals, respectively.

[0022] The problem is solved using a numerical optimization algorithm, and the solution obtained is ( x , y , z The final high-precision positioning coordinates are obtained after the fusion of photoelectric signals.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows: The transformer partial discharge localization method provided by this invention effectively overcomes the limitations of single-signal interference or obstruction by integrating three signals: ultra-high frequency (UHF), high-frequency current, and low-light. This significantly improves the stability of localization. The method possesses an intelligent decision-making mechanism that automatically compares the preliminary localization results with the 3D model, adaptively selecting either the high-frequency current localization mode inside the winding or the optical-electrical joint localization mode outside the winding, thereby optimizing the strategy and improving efficiency. In the optical-electrical joint localization, the signal response states of each low-light sensor are innovatively transformed into explicit spatial constraints, which are then used to optimize the solution of the UHF localization equations. This process effectively suppresses multiple solutions and localization ambiguity, achieving higher-precision spatial localization. Finally, the localization results can be visually displayed in the transformer's 3D model, and a structured diagnostic report containing spatial location, region, and localization mode is output, providing a clear and reliable basis for on-site fault diagnosis and handling. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a flowchart illustrating a transformer partial discharge localization method based on opto-electromagnetic signal fusion, provided as an embodiment of the present invention.

[0026] Figure 2 The fitting curve of high-frequency current characteristic quantity - line pie number provided in the embodiment of the present invention. Detailed Implementation

[0027] The terms “comprising” and “having”, and any variations thereof, in the specification, claims, and accompanying drawings of this invention are intended to cover a non-exclusive inclusion, such as a process, method, system, product, or apparatus that includes a series of steps or units, not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0028] 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. In addition, the technical features of the various embodiments or individual embodiments provided by the present invention can be arbitrarily combined to form new technical solutions. Such combinations are not bound by the order of steps and / or structural composition patterns, but must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0029] The transformer partial discharge localization method based on opto-electromagnetic signal fusion provided by this invention designs an intelligent decision-making algorithm process that integrates multi-source information. Through feature analysis and information combination of ultra-high frequency, high frequency current, and optical signals, it adaptively selects the optimal localization strategy. Finally, the localization results are visualized in a digital 3D model that accurately simulates the complex structure of the transformer's internal windings and core.

[0030] This embodiment uses a test oil-immersed transformer with dimensions of 475×211×267cm to illustrate the implementation of the method of the present invention. The transformer has 6 ultra-high frequency sensors, 4 low-light detection sensors, and high-frequency current sensors installed on the end screen grounding wire and neutral point grounding wire of each phase bushing.

[0031] Two types of discharge defects were designed inside the transformer: First, tree-like scratches were applied to the outer screen of the A-phase high-voltage winding and a suitable amount of carbon traces were coated to simulate a surface discharge fault to ground caused by the outer edge insulation paperboard of the winding. The coordinates of the discharge source were (128, 54, 112) cm. Second, the paper insulation of the adjacent coils of the 23rd and 24th coils in the middle of the A-phase high-voltage winding was damaged and rewound with damp insulation paper containing metal particles to simulate the insulation defect between coils of the high-voltage winding of the transformer.

[0032] See Figure 1 The operation steps of a transformer partial discharge location method based on opto-electromagnetic signal fusion are as follows: Step 1: Preliminary spatial positioning using the UHF method Six ultra-high frequency (UHF) sensors deployed within a transformer were used to synchronously acquire partial discharge electromagnetic wave signals. First, a three-dimensional Cartesian coordinate system covering the transformer was established, and the coordinates of each UHF sensor were precisely measured: S1 (255, 0, 110) cm, S2 (369, 0, 204) cm, S3 (475, 138, 256) cm, S4 (140, 211, 110) cm, S5 (220, 211, 205) cm, S6 (0, 81, 163) cm. Based on the time difference information of the signal arrival at each sensor, the target optimization equation was constructed and solved. , In the formula, e i The signal originates from the power source and propagates to the next... i The difference between the actual propagation distance and the straight-line propagation distance of an ultra-high frequency sensor; c This is the propagation speed of ultra-high frequency signals in transformer oil, taken as 20 cm / ns; t 1 represents the time it takes for the signal to travel from the power source to the first UHF sensor; x i , y i , z i It is the first i Coordinates of an ultra-high frequency sensor; The signal originates from the power source and propagates to the next... i The time difference between the first UHF sensor and the second UHF sensor; x , y , z It refers to the three-dimensional spatial coordinates of the power source.

[0033] Using an iterative optimization algorithm, the preliminary spatial coordinates of the discharge point were obtained. For a surface discharge fault to ground originating from the outer edge of the winding insulation cardboard, the location was (129, 48, 98) cm. For an insulation defect between the high-voltage winding discs, the location was (103, 146, 92) cm.

[0034] Step 2: Selection of intelligent identification and positioning strategy for discharge area The preliminary positioning coordinates obtained in step 1 are compared with the pre-established three-dimensional cylindrical model of the transformer winding to intelligently select subsequent positioning strategies. This example uses the A-phase winding model, with the following parameters: center axis projection coordinates (128, 106) cm, outermost radius of the winding 52 cm, and axial height ranging from 70 cm to 225 cm. The preliminary positioning point is then determined. x , y , zThe mathematical condition for whether a part is located inside the winding is: .

[0035] For surface discharge faults to ground occurring on the outer edge of the winding insulation paperboard, if the discharge point is determined to be outside the winding, the process proceeds to the optical-electrical joint positioning procedure, i.e., step 4. For insulation defects between high-voltage winding discs, if the mathematical condition of being located inside the winding is met, the multi-terminal high-frequency current positioning mode is activated, and the process proceeds to step 3.

[0036] Step 3: Multi-terminal high-frequency current positioning Regarding the internal discharge of the winding, firstly, an accurate high-frequency equivalent circuit model of the transformer winding is constructed. This model treats the winding as a passive chain network composed of resistors, inductors, capacitance to ground, inter-turn capacitance, etc., which can be cascaded through several identical or similar basic units to accurately reflect the electrical characteristics of the winding under high-frequency signals.

[0037] Secondly, in the equivalent circuit model, simulated partial discharge power supplies are set at different numbered wire discs, and multi-location and multi-type enumeration injections are performed. After each injection, the high-frequency current response signals at the beginning and end of the transformer winding are acquired, and time-domain and frequency-domain analyses are performed on them respectively to extract various characteristic parameters, including but not limited to time-frequency distribution (such as short-time Fourier transform STFT characteristics), energy, peak value, and time delay.

[0038] To select features with clear distinguishing capabilities for different discharge phases, each sample contains multi-dimensional time-frequency features extracted from high-frequency current signals and their corresponding true phase labels. Subsequently, a random forest algorithm is applied to quantitatively calculate the contribution of each feature dimension to phase classification, identifying the subset of features that maximizes the separation between different phase samples in the feature space—those with clear distinguishing capabilities. These selected features will serve as the core input for determining the discharge phase in the identification model.

[0039] Furthermore, characteristic quantities that exhibit monotonically increasing or decreasing changes with the pie number are selected and denoted as standard parameters. a i All standard parameters constitute the input matrix. A The pie chart number is used as the output vector. B ,Depend on Perform linear regression to obtain weights p and bias q Establish a quantitative functional relationship between "current characteristic quantity and wire number".

[0040] Finally, based on the functional relationship obtained from the above fitting, the characteristic quantities corresponding to the actual detected fault current signal are substituted into the equation to calculate the specific coil number where the discharge occurred, thus achieving precise coil-level positioning.

[0041] For the three-phase double-winding transformer with YD connection in this embodiment, the measurable current on a certain phase includes: the current of the first section of the high-voltage winding. High-voltage winding end current Low-voltage winding start current The current I Subscript S Indicates the casing side (starting end), E Indicates the neutral point side (terminus). φ Represents separation, H Represents the high-voltage side. L This represents the low-pressure side.

[0042] Calculations revealed that the total energy of the current at the beginning of each phase is an effective criterion for distinguishing between different discharge phases. The calculation method is as follows: , In the formula, express φ Total energy of the phase-starting current; I Represents the magnitude of the current; n Represents the sampling point.

[0043] In this embodiment, an insulation defect was set between the high-voltage winding discs of phase A. The measured three-phase energies were 0.273, 0.105 and 0.073, respectively. The total energy of the current at the beginning of phase A was the highest. Based on this, it was accurately determined that the discharge occurred in phase A.

[0044] Furthermore, calculations revealed the maximum time-frequency domain (STFT) value of the winding start-up current. Corresponding time The relationship between the winding number and the coil number is close to a linear function, making it ideal for positioning. The calculation method for the time corresponding to the maximum value of the winding start current STFT is as follows: , Finally, based on the established quantitative function relationship of "current characteristic quantity - coil number", the characteristic quantity of the actual detected signal is substituted into the function to calculate the specific coil number where the discharge occurred, thus achieving precise coil-level positioning. In this embodiment, according to = 3.38 μs. Based on the fitted curve, the positioning error was calculated to be 2.36 line segments. The fitted curve is shown below. Figure 2 As shown.

[0045] Step 4: Optical-Electronic Joint Positioning Regarding external discharge of the winding: First, check whether all low-light sensors have captured a valid partial discharge optical signal. The specific method is as follows: set an amplitude threshold of 1 V and a time window of 100 ns. If the amplitude of the optical pulse signal captured by a sensor within the 100 ns time window continuously exceeds 1 V, and its pulse timing is synchronized with the partial discharge pulse detected by the ultra-high frequency, then the sensor is determined to have captured a valid partial discharge optical signal.

[0046] Upon testing, only L1 (128, 0, 135) cm detected a significant partial discharge light signal among the four optical sensors, while the other three optical sensors, L2 (475, 100, 166) cm, L3 (90, 211, 215) cm, and L4 (0, 155, 155) cm, did not detect any light signal.

[0047] Because optical signals attenuate rapidly in transformer oil, the detection range of low-light sensors is limited, and their detection performance is affected by the obstruction of solid media. Therefore, the effective detection range of each low-light sensor is considered in three-dimensional space as a spherical domain centered at its installation location and with a radius of 60 cm for its maximum effective detection distance. Based on the signal detection results, the following spatial decision constraints are generated: , The decisions made by all sensors collectively constitute a set of inequality constraints. Combining all the above decision criteria, these are used as spatial constraints to resolve and optimize the UHF positioning equations from step 1 with constraints. The specific objective optimization equation is: , In the formula, e i The signal originates from the power source and propagates to the next... i The difference between the actual propagation distance and the straight-line propagation distance of an ultra-high frequency sensor; c This is the propagation speed of ultra-high frequency signals in transformer oil, taken as 20 cm / ns; t 1 represents the time it takes for the signal to travel from the power source to the first UHF sensor; x i , y i , z i It is the first i Coordinates of an ultra-high frequency sensor; The signal originates from the power source and propagates to the next... i The time difference between the first UHF sensor and the second UHF sensor; x , y , z It refers to the three-dimensional spatial coordinates of the power source.

[0048] A numerical optimization algorithm was used to solve the problem, and the solution (127, 54, 110) cm is the final high-precision positioning coordinate after photoelectric signal fusion. The positioning error is approximately 2 cm, which is about 13 cm less than the initial positioning position obtained by the UHF method.

[0049] Step 5: Visualization of Location Results and Generation of Diagnostic Report The final location coordinates are dynamically displayed in the 3D model of the transformer, and a structured diagnostic report containing information such as the discharge location, the region, and the location mode used is output, providing an intuitive and reliable basis for fault assessment and handling.

[0050] In summary, this invention provides a transformer partial discharge localization method based on the fusion of optical and electromagnetic signals. This method integrates three signals: ultra-high frequency (UHF) partial discharge, high-frequency current, and low-light signals. By analyzing the preliminary UHF localization results and the responses of different sensors, it adaptively selects the optimal localization strategy. Specifically, it first performs preliminary localization using the time difference of arrival (TDOA) method based on the UHF signal, and then automatically determines whether the preliminary discharge point is located inside or outside the winding based on the three-dimensional model. Subsequently, it selects either multi-terminal high-frequency current localization or a combined optical-electrical localization strategy. The final result is visualized and output in the three-dimensional model of the transformer. This invention can significantly improve the accuracy of transformer partial discharge localization and provide a reliable basis for transformer fault diagnosis.

[0051] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A method for locating partial discharge in a transformer based on the fusion of opto-electromagnetic signals, characterized in that, include: Step 1: Using partial discharge signals collected by multiple UHF sensors, the UHF positioning equations are constructed and solved using the time difference of arrival algorithm to obtain preliminary positioning coordinates characterizing the location of the discharge source. Step 2: Compare the obtained preliminary positioning coordinates in the three-dimensional structural model of the transformer, and automatically determine whether the position indicated by the preliminary positioning coordinates is located in the internal region of the winding. If it is located inside the winding, start the multi-terminal high-frequency current positioning mode and proceed to Step 3; if it is located outside the winding, start the optical-electric joint positioning process and proceed to Step 4. Step 3: Analyze the time-frequency characteristics of the high-frequency current signal of each phase, and combine it with the winding circuit model to locate the specific phase and turn where the discharge occurs. Use the location result as the target location coordinates and proceed to step 5. Step 4: Check whether all low-light sensors have captured effective partial discharge light signals. Based on the detection results, convert the response state of each low-light sensor into a spatial determination criterion. Use all spatial determination criteria as spatial constraints to resolve and optimize the UHF positioning equations from Step 1 with constraints. Obtain partial discharge positioning coordinates with higher accuracy than the initial positioning coordinates as the target positioning coordinates, and proceed to Step 5. Step 5: Visualize and output the target positioning coordinates in the 3D model of the transformer.

2. The transformer partial discharge localization method based on opto-electromagnetic signal fusion according to claim 1, characterized in that, The target optimization equation for the UHF positioning equation set in step 1 is: ， Where N is the number of UHF sensors involved in the positioning; e i The signal originates from the power source and propagates to the next... i The difference between the actual propagation distance and the straight-line propagation distance of an ultra-high frequency sensor; c It is the propagation speed of ultra-high frequency signals in transformer oil; t 1 represents the time it takes for the signal to travel from the power source to the first UHF sensor; x i , y i , z i It is the first i Coordinates of an ultra-high frequency sensor; The signal originates from the power source and propagates to the next... i The time difference between the first UHF sensor and the second UHF sensor; x , y , z It is the three-dimensional spatial coordinate of the power source; X min , Y min , Z min , X max , Y max , Z max It is the boundary value of the transformer oil tank.

3. The transformer partial discharge localization method based on opto-electromagnetic signal fusion according to claim 1, characterized in that, Step 2, which involves automatically determining whether the preliminary positioning coordinates are located within the winding's internal region, includes: The transformer winding regions are simplified into cylindrical geometric models with radius and height in three-dimensional space. For the initial positioning coordinates, if they satisfy the mathematical condition of being located within the closed interval defined by the cylindrical geometric model of any winding, then the discharge source is determined to be located inside the winding.

4. The transformer partial discharge localization method based on opto-electromagnetic signal fusion according to claim 3, characterized in that, The mathematical condition located within the closed interval defined by the geometric model of the winding cylinder is: ， in,( x m , y m ( ) represents the projected coordinates of the center axis of a certain phase transformer winding. R m The outermost radius of the winding. H m, min This is the bottom height. H m, max This is the top height.

5. The transformer partial discharge localization method based on opto-electromagnetic signal fusion according to claim 1, characterized in that, The execution of the multi-terminal high-frequency current positioning mode includes: Construct a high-frequency equivalent circuit model of the transformer winding; In the equivalent circuit model, a partial discharge power source is simulated at wire discs with different numbers to obtain the high-frequency current response signals at the beginning and end of the transformer winding and extract their characteristic parameters. High-frequency current characteristic quantities that can distinguish the discharge phase are selected as the basis for determining the discharge phase. High-frequency current characteristics that exhibit monotonically increasing or decreasing changes with increasing wire disc number are selected, and a quantitative functional relationship between "current characteristic quantity - wire disc number" is established using mathematical fitting method; Based on the aforementioned quantitative functional relationship, the characteristic quantities corresponding to the actual detected fault current signals are substituted into the equation to calculate the specific line number where the discharge occurred.

6. The transformer partial discharge localization method based on opto-electromagnetic signal fusion according to claim 5, characterized in that, The high-frequency current characteristic quantities that can distinguish between different discharge phases include: A feature sample library covering different simulated discharge locations of each phase winding is constructed. A feature importance evaluation algorithm is applied to calculate the contribution of each feature dimension to phase classification and identify the feature subset that can produce the maximum separation of different phase samples.

7. The transformer partial discharge localization method based on opto-electromagnetic signal fusion according to claim 5, characterized in that, The establishment of the quantitative functional relationship between "current characteristic quantity - wire disc number" includes: Characteristic parameters that monotonically change with increasing coil number are selected from high-frequency current characteristic quantities and denoted as standard parameters. a i The input matrix A consists of all standard parameters, and the pie chart numbers serve as the output vector B. This is achieved through linear regression. Establish a functional relationship, where p As weight, q For bias.

8. The transformer partial discharge localization method based on opto-electromagnetic signal fusion according to claim 1, characterized in that, Step 4, which involves checking whether all the low-light sensors have captured a valid partial discharge light signal, includes: Set amplitude threshold V th With time window Δ T If a certain sensor is in Δ T The amplitude of the light pulse signal captured within the time window continuously exceeds V th If the pulse timing is synchronized with the partial discharge pulse detected by the ultra-high frequency, then it is determined that the sensor has captured an effective partial discharge optical signal.

9. A transformer partial discharge localization method based on opto-electromagnetic signal fusion according to claim 1 or 8, characterized in that, The process of converting the response state of each low-light sensor into a spatial determination criterion includes: The effective detection range of each low-light sensor is considered to be based on its installation location ( x j , y j , z j Using the sphere as its center and its maximum effective detection range R light A spherical domain with radius . If the first j If a micro-light sensor detects a valid light signal, then the power source is constrained within the spherical domain. If no valid light signal is detected, the constraint that the power source is located outside the spherical domain is applied. .

10. The transformer partial discharge localization method based on opto-electromagnetic signal fusion according to claim 9, characterized in that, Step 4 involves resolving and optimizing the UHF positioning equations with constraints. The objective optimization equation is: ， in, N It refers to the number of UHF sensors involved in the positioning process; e i The signal originates from the power source and propagates to the next... i The difference between the actual propagation distance and the straight-line propagation distance of an ultra-high frequency sensor; c It is the propagation speed of ultra-high frequency signals in transformer oil; t 1 represents the time it takes for the signal to travel from the power source to the first UHF sensor; x i , y i , z i It is the first i Coordinates of an ultra-high frequency sensor; The signal originates from the power source and propagates to the next... i The time difference between the first UHF sensor and the second UHF sensor; x , y , z () represents the three-dimensional spatial coordinates of the discharge source; X min , Y min , Z min , X max , Y max , Z max These are the boundary values ​​of the transformer oil tank. x j , y j , z j )and( x k , y k , z k ) are respectively the first j The and the first k Coordinates of a low-light sensor; R light This represents the maximum effective detection distance of the low-light sensor. M detected and M undetected These are the sets of sensors that detected and those that did not detect valid light signals, respectively.

Images