Method and device for diagnosing winding state of power transformer

Abstract

The invention discloses a power transformer winding state diagnosis method and device, and relates to the technical field of transformer winding insulation state diagnosis. The method comprises the following steps: acquiring a transient signal and a steady-state signal of a power transformer winding, performing dimension conversion on the transient signal and extracting a feature parameter to obtain an energy feature vector, and performing image conversion on the steady-state signal and extracting a feature parameter to obtain an image feature vector; therefore, the state of the power transformer winding is diagnosed by combining the transient signal and the steady-state signal. An energy feature vector and an image feature vector are obtained by extracting feature parameters of a transient signal and a steady-state signal respectively, and features about the state of the transformer winding in the transient signal and the steady-state signal are mined in the form of the feature vectors, so that the state of the power transformer winding can be represented more accurately; the combination of the transient signal and the steady signal is realized in the form of the feature vector, so that the accuracy of the state diagnosis of the power transformer winding is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of transformer winding insulation state diagnosis, and in particular relates to a method and device for diagnosing the state of a power transformer winding. Background Art

[0002] As society's demand for electric energy grows, the scale of the power system is increasing. In the power system, power transformers play a vital role, such as converting voltage and current, so the safe and stable operation of power transformers is crucial. In the stable operation of power transformers, its insulation performance determines the power quality of the transformer. When the insulation performance deteriorates seriously, it will affect the operation of the power system. Therefore, timely diagnosis of the winding status of the power transformer and judging its insulation performance are critical to the stable operation of the power system.

[0003] Injecting excitation signals into transformers and analyzing the changes in response signals is a hot topic in the current research of detecting transformer winding faults. Currently, the most widely used methods mostly use a single transient signal or steady-state signal, and only focus on one aspect of the transient signal or steady-state signal, which will result in the lack of device information of the power transformer winding state, and fail to fully reflect the state information of the transformer winding, making the diagnosis accuracy of the power transformer winding state low. Therefore, there is an urgent need for a method and device for diagnosing the power transformer winding state to solve the defects in the prior art. Summary of the invention

[0004] The present invention aims to provide a method and device for diagnosing the state of a power transformer winding, so as to solve the above-mentioned technical problems, extract characteristic vectors of transient signals and steady-state signals of the power transformer winding, realize the judgment combination of transient signals and steady-state signals, and improve the accuracy of the diagnosis of the state of the power transformer winding.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a method for diagnosing the winding state of a power transformer, comprising:

[0006] Collect transient and steady-state signals of power transformer windings;

[0007] Constructing an average curve of the transient signal according to the transient signal, and obtaining an eigencomponent corresponding to the transient signal in combination with a preset judgment function, so as to extract an energy eigenvector from the eigencomponent;

[0008] Converting the steady-state signal into a feature image according to a preset signal conversion algorithm, and extracting an image feature vector from the feature image;

[0009] The state factor of the power transformer winding is calculated according to the energy feature vector and the image feature vector, and the diagnosis result of the power transformer winding state is obtained in combination with a preset threshold.

[0010] It can be understood that, compared with the prior art, the present invention realizes the diagnosis of the winding state of the power transformer by combining the transient signal and the steady-state signal by collecting the transient signal and the steady-state signal of the power transformer winding, performing dimension conversion on the transient signal and extracting the characteristic parameters to obtain the energy characteristic vector, performing image conversion on the steady-state signal and extracting the characteristic parameters to obtain the image characteristic vector. The energy characteristic vector and the image characteristic vector are obtained by respectively extracting the characteristic parameters of the transient signal and the steady-state signal, and the characteristics of the transformer winding state in the transient signal and the steady-state signal are mined in the form of characteristic vectors, so that the winding state of the power transformer can be more accurately characterized; the transient signal and the steady-state signal are combined in the form of characteristic vectors, thereby improving the accuracy of the diagnosis of the winding state of the power transformer.

[0011] As a preferred solution, the method of constructing an average curve of the transient signal according to the transient signal, and obtaining the intrinsic component corresponding to the transient signal in combination with a preset judgment function to extract an energy characteristic vector from the intrinsic component specifically includes:

[0012] Obtaining a maximum value point envelope and a minimum value point envelope of the transient signal, and constructing an average curve of the transient signal according to the maximum value point envelope and the minimum value point envelope;

[0013] Determine the transient signal and its average curve according to a preset judgment function to obtain an eigencomponent corresponding to the transient signal;

[0014] An energy eigenvector is extracted from the eigencomponents.

[0015] As a preferred solution, judging the transient signal and its average curve according to a preset judgment function to obtain the intrinsic component corresponding to the transient signal specifically includes:

[0016] Acquire candidate components according to the transient signal and the average curve;

[0017] Determine whether the candidate component meets the requirements of a preset judgment function;

[0018] When the candidate component meets the requirements of the preset judgment function, the candidate component is used as the eigencomponent; otherwise, the transient signal is updated according to the candidate component to complete the update of the candidate component, until the transient signal is in the preset requirements, the transient signal and its average curve are judged, and the eigencomponent corresponding to the transient signal is obtained;

[0019] Among them, the preset judgment function is specifically:

[0020]

[0021] Among them, h k (t) is the kth candidate component; T is time.

[0022] This preferred solution converts the dimension of the transient signal to obtain the intrinsic component of the transient signal and then obtain the energy eigenvector; the characteristics of the transformer winding state in the transient signal are mined in the form of a eigenvector, so that the power transformer winding state can be more accurately characterized from a transient perspective.

[0023] As a preferred solution, converting the steady-state signal into a feature image according to a preset signal conversion algorithm, and extracting an image feature vector from the feature image, specifically includes:

[0024] Acquire vibration parameter information of the steady-state signal, and convert the steady-state signal into a feature image in combination with a preset signal conversion algorithm;

[0025] Acquire pixel information of the feature image, and calculate the gray gradient mean square error of the feature image based on the pixel information;

[0026] An image feature vector is extracted from the feature image according to the grayscale gradient mean square error.

[0027] As a preferred solution, the step of acquiring the vibration parameter information of the steady-state signal and converting the steady-state signal into a feature image in combination with a preset signal conversion algorithm specifically includes:

[0028] Acquiring vibration parameter information of the steady-state signal;

[0029] Determine polar coordinate radius and rotation angle information according to the vibration parameter information;

[0030] The steady-state signal is converted into a characteristic image according to the polar coordinate radius and the rotation angle information.

[0031] This preferred solution converts the steady-state signal into a feature image, thereby obtaining the grayscale gradient mean square error of the feature image, and then obtaining the image feature vector; the characteristics of the transformer winding state in the steady-state signal are mined in the form of a feature vector, so that the power transformer winding state can be more accurately characterized from a steady-state perspective.

[0032] As a preferred solution, the calculating of the state factor of the power transformer winding according to the energy feature vector and the image feature vector specifically includes:

[0033] The state factor of the power transformer winding is calculated according to the energy feature vector and the image feature vector; wherein the expression of the state factor is:

[0034]

[0035] Among them, λ is the state factor, T is the energy feature vector, and U is the image feature vector.

[0036] This preferred solution realizes the combination of transient signals and steady-state signals in the form of characteristic vectors, thereby improving the accuracy of diagnosis of the power transformer winding state.

[0037] Accordingly, an embodiment of the present invention provides a device for diagnosing the state of a power transformer winding, comprising: a signal acquisition module, an energy feature vector extraction module, an image feature vector extraction module and a power transformer winding state diagnosis module;

[0038] Wherein, the signal acquisition module is used to collect transient signals and steady-state signals of the power transformer winding;

[0039] The energy characteristic vector extraction module is used to construct an average curve of the transient signal according to the transient signal, and obtain the intrinsic component corresponding to the transient signal in combination with a preset judgment function to extract the energy characteristic vector from the intrinsic component;

[0040] The image feature vector extraction module is used to convert the steady-state signal into a feature image according to a preset signal conversion algorithm, and extract an image feature vector from the feature image;

[0041] The power transformer winding state diagnosis module is used to calculate the state factor of the power transformer winding according to the energy feature vector and the image feature vector, and obtain the diagnosis result of the power transformer winding state in combination with a preset threshold.

[0042] As a preferred solution, the energy feature vector extraction module includes: an energy feature vector extraction unit;

[0043] The energy feature vector extraction unit is used to obtain the maximum value point envelope and the minimum value point envelope of the transient signal, and construct an average curve of the transient signal according to the maximum value point envelope and the minimum value point envelope;

[0044] Determine the transient signal and its average curve according to a preset judgment function to obtain an eigencomponent corresponding to the transient signal;

[0045] An energy eigenvector is extracted from the eigencomponents.

[0046] As a preferred solution, the image feature vector extraction module includes: an image feature vector extraction unit;

[0047] The image feature vector extraction unit is used to obtain the vibration parameter information of the steady-state signal, and convert the steady-state signal into a feature image in combination with a preset signal conversion algorithm;

[0048] Acquire pixel information of the feature image, and calculate the gray gradient mean square error of the feature image based on the pixel information;

[0049] An image feature vector is extracted from the feature image according to the grayscale gradient mean square error.

[0050] As a preferred solution, the power transformer winding state diagnosis module comprises: a power transformer winding state diagnosis unit;

[0051] The power transformer winding state diagnosis unit is used to calculate the state factor of the power transformer winding according to the energy feature vector and the image feature vector; wherein the expression of the state factor is:

[0052]

[0053] Among them, λ is the state factor, T is the energy feature vector, and U is the image feature vector.

[0054] It can be understood that, compared with the prior art, the present device realizes the diagnosis of the winding state of the power transformer by combining the transient signal and the steady-state signal by collecting the transient signal and the steady-state signal, performing dimension conversion on the transient signal and extracting the characteristic parameters to obtain the energy characteristic vector, performing image conversion on the steady-state signal and extracting the characteristic parameters to obtain the image characteristic vector. The energy characteristic vector and the image characteristic vector are obtained by respectively extracting the characteristic parameters of the transient signal and the steady-state signal, and the characteristics of the transformer winding state in the transient signal and the steady-state signal are mined in the form of characteristic vectors, so that the winding state of the power transformer can be more accurately characterized; the transient signal and the steady-state signal are combined in the form of characteristic vectors, thereby improving the accuracy of the diagnosis of the winding state of the power transformer. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 : A flowchart of a method for diagnosing the winding state of a power transformer provided by an embodiment of the present invention;

[0056] Figure 2 : A schematic diagram of the structure of a device for diagnosing the winding status of a power transformer provided by an embodiment of the present invention;

[0057] Figure 3 : A schematic diagram of the structure of a diagnostic test platform for power transformer winding status provided by an embodiment of the present invention;

[0058] Among them, 201: signal acquisition module; 202: energy feature vector extraction module; 203: image feature vector extraction module; 204: power transformer winding state diagnosis module; 301: programmable power supply; 302: A phase high voltage side bushing; 303: B phase high voltage side bushing; 304: C phase high voltage side bushing; 305: A phase low voltage side bushing; 306: B phase low voltage side bushing; 307: C phase low voltage side bushing; 308: A phase winding; 309: B phase winding; 310: C phase winding; 311: acceleration sensor; 312: signal acquisition device; 313: voiceprint vibration detection box; 314: host computer. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] Embodiment 1

[0061] Please refer to Figure 1 , is a flowchart of a method for diagnosing the winding status of a power transformer provided in an embodiment of the present invention, including steps S101 to S104.

[0062] Step S101: collecting transient signals and steady-state signals of power transformer windings.

[0063] Step S102: constructing an average curve of the transient signal according to the transient signal, and acquiring an eigencomponent corresponding to the transient signal in combination with a preset judgment function, so as to extract an energy eigenvector from the eigencomponent.

[0064] In this embodiment, the method of constructing an average curve of the transient signal according to the transient signal and obtaining the intrinsic component corresponding to the transient signal in combination with a preset judgment function to extract an energy characteristic vector from the intrinsic component specifically includes:

[0065] Obtaining a maximum value point envelope and a minimum value point envelope of the transient signal, and constructing an average curve of the transient signal according to the maximum value point envelope and the minimum value point envelope;

[0066] Determine the transient signal and its average curve according to a preset judgment function to obtain an eigencomponent corresponding to the transient signal;

[0067] An energy eigenvector is extracted from the eigencomponents.

[0068] In an optional embodiment, the step of obtaining a maximum value point envelope and a minimum value point envelope of the transient signal, and constructing an average curve of the transient signal according to the maximum value point envelope and the minimum value point envelope, specifically includes:

[0069]

[0070] Among them, u(t) is the envelope curve of the maximum point of the transient signal x(t), v(t) is the envelope curve of the minimum point of the transient signal x(t); m1(t) is the average curve.

[0071] In this embodiment, judging the transient signal and its average curve according to a preset judgment function to obtain the intrinsic component corresponding to the transient signal specifically includes:

[0072] Acquire candidate components according to the transient signal and the average curve;

[0073] Determine whether the candidate component meets the requirements of a preset judgment function;

[0074] When the candidate component meets the requirements of the preset judgment function, the candidate component is used as the eigencomponent; otherwise, the transient signal is updated according to the candidate component to complete the update of the candidate component, until the transient signal is in the preset requirements, the transient signal and its average curve are judged, and the eigencomponent corresponding to the transient signal is obtained;

[0075] Among them, the preset judgment function is specifically:

[0076]

[0077] Among them, h k (t) is the kth candidate component; T is time.

[0078] In an optional embodiment, the expression of the candidate component is:

[0079] h1(t)=x(t)-m1(t),

[0080] Among them, m1(t) is the average curve, x(t) is the transient signal, and h1(t) is the alternative component.

[0081] In some optional implementations of this embodiment, when the candidate component h1(t) satisfies the judgment rule of the judgment function, the candidate component h1(t) is used as the first eigencomponent c1(t); when the candidate component h1(t) does not satisfy the judgment rule of the judgment function, the candidate component h1(t) is used as the transient signal x(t), thereby updating the transient signal x(t); and the candidate component h1(t) is updated according to the updated transient signal x(t); until the transient signal meets the preset requirements, it means that the transient signal is a monotonic function or can no longer be decomposed into an eigencomponent that meets the requirements of the judgment function. The eigencomponent corresponding to the transient signal is c i (t), i = 2, 3, ..., k, where t is the test time (i.e., the duration of the transient signal collected).

[0082] In an optional embodiment, extracting the energy eigenvector from the eigencomponent specifically includes:

[0083] Calculate the energy E of the first k eigencomponents i :

[0084]

[0085] (t=1,2,…,N; i=1,2,…,k),

[0086] Among them, E i is energy, c i (t) is the eigenvalue.

[0087] According to the energy E of the first k eigencomponents i Construct the energy eigenvector:

[0088] T=[E1,E2,…,E k ],

[0089] Where T is the energy eigenvector, E1, E2, …, E k For energy.

[0090] This embodiment converts the dimension of the transient signal to obtain the intrinsic component of the transient signal and then obtains the energy eigenvector; the characteristics of the transformer winding state in the transient signal are mined in the form of a eigenvector, so that the power transformer winding state can be more accurately characterized from a transient perspective.

[0091] Step S103: converting the steady-state signal into a feature image according to a preset signal conversion algorithm, and extracting an image feature vector from the feature image.

[0092] In this embodiment, converting the steady-state signal into a feature image according to a preset signal conversion algorithm, and extracting an image feature vector from the feature image specifically includes:

[0093] Acquire vibration parameter information of the steady-state signal, and convert the steady-state signal into a feature image in combination with a preset signal conversion algorithm;

[0094] Acquire pixel information of the feature image, and calculate the gray gradient mean square error of the feature image based on the pixel information;

[0095] An image feature vector is extracted from the feature image according to the grayscale gradient mean square error.

[0096] In this embodiment, the step of acquiring the vibration parameter information of the steady-state signal and converting the steady-state signal into a feature image in combination with a preset signal conversion algorithm specifically includes:

[0097] Acquiring vibration parameter information of the steady-state signal;

[0098] Determine polar coordinate radius and rotation angle information according to the vibration parameter information;

[0099] The steady-state signal is converted into a characteristic image according to the polar coordinate radius and the rotation angle information.

[0100] It should be noted that the conversion of the steady-state signal into a characteristic image can be achieved through the SDP transformation algorithm. The SDP transformation algorithm (Symmetric Dot Pattern) is an algorithm that converts a one-dimensional signal into a two-dimensional symmetrical image; it is an image representation method based on a polar coordinate system, which can convert a one-dimensional signal into a symmetrical petal image under two-dimensional polar coordinates.

[0101] In an optional embodiment, obtaining vibration parameter information of the steady-state signal and converting the steady-state signal into a feature image in combination with a preset signal conversion algorithm specifically includes:

[0102]

[0103] Among them, γ(i) is the polar coordinate radius, α(i) and β(i) are the rotation angles along the initial line in counterclockwise and clockwise directions, and x i is the vibration parameter at time i, x i+Δτ is the vibration parameter at time i+Δτ, x max is the maximum value of the vibration parameter, x min is the minimum value of the vibration parameter, Δτ is the time interval, is the initial line angle, take Δτ=3.

[0104] In an optional embodiment, the acquiring pixel information of the feature image and calculating the grayscale gradient mean square error of the feature image based on the pixel information specifically includes:

[0105]

[0106] Among them, H(m,n) represents the number of pixels of the feature image, N represents the total number of pixels of the feature image, and U i is the mean square error of grayscale gradient.

[0107] In an optional embodiment, extracting the image feature vector from the feature image according to the grayscale gradient mean square error specifically includes:

[0108] U=[U1,U2,…,U k ] T ,

[0109] Among them, U is the image feature vector, U1,U2,…,U k are the mean square error of grayscale gradient.

[0110] This embodiment converts the steady-state signal into a feature image, thereby obtaining the grayscale gradient mean square error of the feature image, and then obtaining the image feature vector; the characteristics of the transformer winding state in the steady-state signal are mined in the form of a feature vector, so that the power transformer winding state can be more accurately characterized from a steady-state perspective.

[0111] Step S104: Calculate the state factor of the power transformer winding according to the energy feature vector and the image feature vector, and obtain the diagnosis result of the power transformer winding state in combination with a preset threshold.

[0112] In this embodiment, calculating the state factor of the power transformer winding according to the energy feature vector and the image feature vector specifically includes:

[0113] The state factor of the power transformer winding is calculated according to the energy feature vector and the image feature vector; wherein the expression of the state factor is:

[0114]

[0115] Among them, λ is the state factor, T is the energy feature vector, and U is the image feature vector.

[0116] This embodiment realizes the combination of transient signal and steady-state signal in the form of characteristic vector, thereby improving the accuracy of diagnosis of power transformer winding state.

[0117] In an optional embodiment, the preset threshold is 2.5. If the state factor λ≥2.5, the diagnosis result of the power transformer winding state is that the insulation state of the transformer winding is good and no insulation fault occurs; if λ<2.5, the diagnosis result of the power transformer winding state is that the insulation performance of the transformer winding is degraded or an insulation fault occurs, and the corresponding wiring method or the performance of the insulation material needs to be checked.

[0118] This embodiment collects transient signals and steady-state signals of the power transformer winding, performs dimension conversion on the transient signal and extracts characteristic parameters to obtain an energy characteristic vector, performs image conversion on the steady-state signal and extracts characteristic parameters to obtain an image characteristic vector, thereby realizing the diagnosis of the winding state of the power transformer by combining the transient signal and the steady-state signal. The energy characteristic vector and the image characteristic vector are obtained by respectively extracting the characteristic parameters of the transient signal and the steady-state signal, and the features of the transformer winding state in the transient signal and the steady-state signal are mined in the form of characteristic vectors, so that the winding state of the power transformer can be more accurately characterized; the transient signal and the steady-state signal are combined in the form of characteristic vectors, thereby improving the accuracy of the diagnosis of the winding state of the power transformer.

[0119] Embodiment 2

[0120] Please refer to Figure 2 , which is a structural schematic diagram of a power transformer winding state diagnosis device provided in an embodiment of the present invention, comprising: a signal acquisition module 201, an energy feature vector extraction module 202, an image feature vector extraction module 203 and a power transformer winding state diagnosis module 204.

[0121] The signal acquisition module 201 is used to collect transient signals and steady-state signals of the power transformer winding.

[0122] The energy characteristic vector extraction module 202 is used to construct an average curve of the transient signal according to the transient signal, and obtain the intrinsic component corresponding to the transient signal in combination with a preset judgment function to extract the energy characteristic vector from the intrinsic component.

[0123] In this embodiment, the energy feature vector extraction module 202 includes: an energy feature vector extraction unit;

[0124] The energy feature vector extraction unit is used to obtain the maximum value point envelope and the minimum value point envelope of the transient signal, and construct an average curve of the transient signal according to the maximum value point envelope and the minimum value point envelope;

[0125] Determine the transient signal and its average curve according to a preset judgment function to obtain an eigencomponent corresponding to the transient signal;

[0126] An energy eigenvector is extracted from the eigencomponents.

[0127] In this embodiment, the energy feature vector extraction unit includes: an eigencomponent extraction subunit;

[0128] The intrinsic component extraction subunit is used to obtain candidate components according to the transient signal and the average curve;

[0129] Determine whether the candidate component meets the requirements of a preset judgment function;

[0130] When the candidate component meets the requirements of the preset judgment function, the candidate component is used as the eigencomponent; otherwise, the transient signal is updated according to the candidate component to complete the update of the candidate component, until the transient signal is in the preset requirements, the transient signal and its average curve are judged, and the eigencomponent corresponding to the transient signal is obtained;

[0131] Among them, the preset judgment function is specifically:

[0132]

[0133] Among them, h k (t) is the kth candidate component; T is time.

[0134] The image feature vector extraction module 203 is used to convert the steady-state signal into a feature image according to a preset signal conversion algorithm, and extract an image feature vector from the feature image.

[0135] In this embodiment, the image feature vector extraction module 203 includes: an image feature vector extraction unit;

[0136] The image feature vector extraction unit is used to obtain the vibration parameter information of the steady-state signal, and convert the steady-state signal into a feature image in combination with a preset signal conversion algorithm;

[0137] Acquire pixel information of the feature image, and calculate the gray gradient mean square error of the feature image based on the pixel information;

[0138] An image feature vector is extracted from the feature image according to the grayscale gradient mean square error.

[0139] In this embodiment, the image feature vector extraction unit includes: a feature image conversion subunit;

[0140] The characteristic image conversion subunit is used to obtain the vibration parameter information of the steady-state signal;

[0141] Determine polar coordinate radius and rotation angle information according to the vibration parameter information;

[0142] The steady-state signal is converted into a characteristic image according to the polar coordinate radius and the rotation angle information.

[0143] The power transformer winding state diagnosis module 204 is used to calculate the state factor of the power transformer winding according to the energy feature vector and the image feature vector, and obtain the diagnosis result of the power transformer winding state in combination with a preset threshold.

[0144] In this embodiment, the power transformer winding state diagnosis module 204 includes: a power transformer winding state diagnosis unit;

[0145] The power transformer winding state diagnosis unit is used to calculate the state factor of the power transformer winding according to the energy feature vector and the image feature vector; wherein the expression of the state factor is:

[0146]

[0147] Among them, λ is the state factor, T is the energy feature vector, and U is the image feature vector.

[0148] This embodiment collects transient signals and steady-state signals of the power transformer winding, performs dimension conversion on the transient signal and extracts characteristic parameters to obtain an energy characteristic vector, performs image conversion on the steady-state signal and extracts characteristic parameters to obtain an image characteristic vector, thereby realizing the diagnosis of the winding state of the power transformer by combining the transient signal and the steady-state signal. The energy characteristic vector and the image characteristic vector are obtained by respectively extracting the characteristic parameters of the transient signal and the steady-state signal, and the features of the transformer winding state in the transient signal and the steady-state signal are mined in the form of characteristic vectors, so that the winding state of the power transformer can be more accurately characterized; the transient signal and the steady-state signal are combined in the form of characteristic vectors, thereby improving the accuracy of the diagnosis of the winding state of the power transformer.

[0149] Embodiment 3

[0150] Please refer to Figure 3 , which is a structural schematic diagram of a diagnostic test platform for the winding status of a power transformer provided in an embodiment of the present invention; the diagnostic test platform includes: a programmable power supply 301, an A-phase high-voltage side bushing 302, a B-phase high-voltage side bushing 303, a C-phase high-voltage side bushing 304, an A-phase low-voltage side bushing 305, a B-phase low-voltage side bushing 306, a C-phase low-voltage side bushing 307, an A-phase winding 308, a B-phase winding 309, a C-phase winding 310, an acceleration sensor 311, a signal acquisition device 312, a voiceprint vibration detection box 313 and a host computer 314.

[0151] Wherein, the signal acquisition device 312 is used to collect transient signals of the power transformer winding;

[0152] The acceleration sensor 311 and the soundprint vibration detection box 313 are used to collect steady-state signals of the power transformer winding;

[0153] The host computer 314 is used to construct an average curve of the transient signal according to the transient signal, and obtain the intrinsic component corresponding to the transient signal in combination with a preset judgment function to extract an energy characteristic vector from the intrinsic component;

[0154] The host computer 314 is also used to convert the steady-state signal into a feature image according to a preset signal conversion algorithm, and extract an image feature vector from the feature image;

[0155] The host computer 314 is further used to calculate the state factor of the power transformer winding according to the energy feature vector and the image feature vector, and obtain the diagnosis result of the power transformer winding state in combination with a preset threshold.

[0156] In an optional embodiment, the output end of the programmable power supply 301 is connected to the input ends of the A-phase high-voltage side bushing 302 , the B-phase high-voltage side bushing 303 , and the C-phase high-voltage side bushing 304 , respectively.

[0157] In an optional embodiment, the programmable power supply 301 is used to generate a transient signal and inject it into the A-phase high-voltage side bushing 302 , the B-phase high-voltage side bushing 303 , and the C-phase high-voltage side bushing 304 .

[0158] In an optional embodiment, the input end of the signal acquisition device 312 is respectively connected to the A-phase high-voltage side bushing 302 , the B-phase high-voltage side bushing 303 and the C-phase high-voltage side bushing 304 ; the output end of the signal acquisition device 312 is connected to the input end of the host computer 314 .

[0159] In an optional embodiment, the input end of the signal acquisition device 312 is connected to the A-phase low-voltage side bushing 305 , the B-phase low-voltage side bushing 306 , and the C-phase low-voltage side bushing 307 ; the output end of the signal acquisition device 312 is connected to the input end of the host computer 314 .

[0160] In an optional embodiment, the input end of the voiceprint vibration detection box 313 is connected to the output end of the acceleration sensor 311 ; the output end of the voiceprint vibration detection box 313 is connected to the input end of the host computer 314 .

[0161] In an optional embodiment, the acceleration sensor 311 is used to collect vibration signals of the A-phase winding 308 , the B-phase winding 309 , and the C-phase winding 310 , and transmit the vibration signals to the voiceprint vibration detection box 313 .

[0162] It should be noted that the A-phase high-voltage side bushing 302, the B-phase high-voltage side bushing 303, the C-phase high-voltage side bushing 304, the A-phase low-voltage side bushing 305, the B-phase low-voltage side bushing 306, the C-phase low-voltage side bushing 307, the A-phase winding 308, the B-phase winding 309 and the C-phase winding 310 belong to the structural composition of the power transformer, and there are currently many technologies for describing the structure of the power transformer, which will not be described in detail in the embodiments of the present invention.

[0163] In an optional embodiment, the working principle of the diagnostic test platform includes:

[0164] The programmable power supply 301 generates a transient signal and inputs it into the A-phase high-voltage side bushing 302, the B-phase high-voltage side bushing 303 and the C-phase high-voltage side bushing 304 respectively. The transient signal then passes through the A-phase low-voltage side bushing 305, the B-phase low-voltage side bushing 306 and the C-phase low-voltage side bushing 307 and enters the signal acquisition device 312, and is transmitted to the host computer 314 through the signal acquisition device 312;

[0165] The transformer in the diagnostic test platform generates vibration signals during operation. The acceleration sensor 311 collects the vibration signals of the A-phase winding 308, the B-phase winding 309 and the C-phase winding 310, and transmits them to the soundprint vibration detection box 313. The steady-state signal is obtained through the soundprint vibration detection box 313 and transmitted to the host computer 314.

[0166] The host computer 314 constructs an average curve of the transient signal according to the transient signal, and obtains the intrinsic component corresponding to the transient signal in combination with a preset judgment function to extract an energy characteristic vector from the intrinsic component;

[0167] The host computer 314 converts the steady-state signal into a feature image according to a preset signal conversion algorithm, and extracts an image feature vector from the feature image;

[0168] The host computer 314 calculates the state factor of the power transformer winding according to the energy feature vector and the image feature vector, and obtains the diagnosis result of the power transformer winding state in combination with a preset threshold.

[0169] In summary, the embodiment of the present invention realizes the diagnosis of the winding state of the power transformer by combining the transient signal and the steady-state signal by collecting the transient signal and the steady-state signal of the power transformer winding, performing dimension conversion on the transient signal and extracting characteristic parameters to obtain the energy characteristic vector, performing image conversion on the steady-state signal and extracting characteristic parameters to obtain the image characteristic vector. The energy characteristic vector and the image characteristic vector are obtained by respectively extracting the characteristic parameters of the transient signal and the steady-state signal, and the features of the transformer winding state in the transient signal and the steady-state signal are mined in the form of characteristic vectors, so that the winding state of the power transformer can be more accurately characterized; the transient signal and the steady-state signal are combined in the form of characteristic vectors, thereby improving the accuracy of the diagnosis of the winding state of the power transformer.

[0170] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for diagnosing the winding status of a power transformer, characterized in that: include: Collect transient and steady-state signals of power transformer windings; Constructing an average curve of the transient signal according to the transient signal, and obtaining an eigencomponent corresponding to the transient signal in combination with a preset judgment function, so as to extract an energy eigenvector from the eigencomponent; Converting the steady-state signal into a feature image according to a preset signal conversion algorithm, and extracting an image feature vector from the feature image; The state factor of the power transformer winding is calculated according to the energy feature vector and the image feature vector, and the diagnosis result of the power transformer winding state is obtained in combination with a preset threshold.

2. A method for diagnosing the winding status of a power transformer as claimed in claim 1, characterized in that: The step of constructing an average curve of the transient signal according to the transient signal, and obtaining an intrinsic component corresponding to the transient signal in combination with a preset judgment function to extract an energy characteristic vector from the intrinsic component specifically includes: Obtaining a maximum value point envelope and a minimum value point envelope of the transient signal, and constructing an average curve of the transient signal according to the maximum value point envelope and the minimum value point envelope; Determine the transient signal and its average curve according to a preset judgment function to obtain an eigencomponent corresponding to the transient signal; An energy eigenvector is extracted from the eigencomponents.

3. A method for diagnosing the winding status of a power transformer as claimed in claim 1 or 2, characterized in that: The step of determining the transient signal and its average curve according to a preset determination function to obtain an intrinsic component corresponding to the transient signal specifically includes: Acquire candidate components according to the transient signal and the average curve; Determine whether the candidate component meets the requirements of a preset judgment function; When the candidate component meets the requirements of the preset judgment function, the candidate component is used as the eigencomponent; otherwise, the transient signal is updated according to the candidate component to complete the update of the candidate component, until the transient signal is in the preset requirements, the transient signal and its average curve are judged, and the eigencomponent corresponding to the transient signal is obtained; Among them, the preset judgment function is specifically: Among them, h k (t) is the kth candidate component; T is time.

4. A method for diagnosing the winding status of a power transformer as claimed in claim 1, characterized in that: The step of converting the steady-state signal into a feature image according to a preset signal conversion algorithm, and extracting an image feature vector from the feature image, specifically includes: Acquire vibration parameter information of the steady-state signal, and convert the steady-state signal into a feature image in combination with a preset signal conversion algorithm; Acquire pixel information of the feature image, and calculate the gray gradient mean square error of the feature image based on the pixel information; An image feature vector is extracted from the feature image according to the grayscale gradient mean square error.

5. A method for diagnosing the state of a power transformer winding according to claim 4, characterized in that: The step of obtaining the vibration parameter information of the steady-state signal and converting the steady-state signal into a feature image in combination with a preset signal conversion algorithm specifically includes: Acquiring vibration parameter information of the steady-state signal; Determine polar coordinate radius and rotation angle information according to the vibration parameter information; The steady-state signal is converted into a characteristic image according to the polar coordinate radius and the rotation angle information.

6. A method for diagnosing the state of a power transformer winding according to claim 1, characterized in that: The calculating the state factor of the power transformer winding according to the energy feature vector and the image feature vector specifically includes: The state factor of the power transformer winding is calculated according to the energy feature vector and the image feature vector; wherein the expression of the state factor is: Among them, λ is the state factor, T is the energy feature vector, and U is the image feature vector.

7. A device for diagnosing the winding status of a power transformer, characterized in that: include: Signal acquisition module, energy feature vector extraction module, image feature vector extraction module and power transformer winding state diagnosis module; Wherein, the signal acquisition module is used to collect transient signals and steady-state signals of the power transformer winding; The energy characteristic vector extraction module is used to construct an average curve of the transient signal according to the transient signal, and obtain the intrinsic component corresponding to the transient signal in combination with a preset judgment function to extract the energy characteristic vector from the intrinsic component; The image feature vector extraction module is used to convert the steady-state signal into a feature image according to a preset signal conversion algorithm, and extract an image feature vector from the feature image; The power transformer winding state diagnosis module is used to calculate the state factor of the power transformer winding according to the energy feature vector and the image feature vector, and obtain the diagnosis result of the power transformer winding state in combination with a preset threshold.

8. A device for diagnosing the winding status of a power transformer as claimed in claim 7, characterized in that: The energy feature vector extraction module comprises: an energy feature vector extraction unit; The energy feature vector extraction unit is used to obtain the maximum value point envelope and the minimum value point envelope of the transient signal, and construct an average curve of the transient signal according to the maximum value point envelope and the minimum value point envelope; Determine the transient signal and its average curve according to a preset judgment function to obtain an eigencomponent corresponding to the transient signal; An energy eigenvector is extracted from the eigencomponents.

9. A device for diagnosing the winding status of a power transformer as claimed in claim 7, characterized in that: The image feature vector extraction module comprises: an image feature vector extraction unit; The image feature vector extraction unit is used to obtain the vibration parameter information of the steady-state signal, and convert the steady-state signal into a feature image in combination with a preset signal conversion algorithm; Acquire pixel information of the feature image, and calculate the gray gradient mean square error of the feature image based on the pixel information; An image feature vector is extracted from the feature image according to the grayscale gradient mean square error.

10. A device for diagnosing the winding status of a power transformer as claimed in claim 7, characterized in that: The power transformer winding state diagnosis module comprises: a power transformer winding state diagnosis unit; The power transformer winding state diagnosis unit is used to calculate the state factor of the power transformer winding according to the energy feature vector and the image feature vector; wherein the expression of the state factor is: Among them, λ is the state factor, T is the energy feature vector, and U is the image feature vector.

Images