A method for determining AC loss of transformer windings

By designing a customized time-sharing prediction model and radial basis neural network, combining historical data and configuration information, intelligently predicting the AC loss of the transformer winding, solving the problem of insufficient loss calculation accuracy in the existing technology, and achieving high-precision distributed monitoring and over-limit warning.

CN118604691BActive Publication Date: 2024-11-26HUANGHUA POWER SUPPLY COMPANY OF STATE GRID QINGHAI ELECTRIC POWER +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410664645.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-27
Publication Date
2024-11-26
Estimated Expiration
2044-05-27

AI Technical Summary

Technical Problem

It is difficult for the prior art to accurately calculate the accuracy in the future time interval of the AC loss of the transformer winding, resulting in temperature changes affecting the continuous changes of the primary and secondary winding resistance, inductance and other parameters, resulting in inaccurate calculation results.

Method used

Design a time-sharing prediction model with customized structures, use radial-based neural network for multiple learning, combine historical temperature data, configuration information and input voltage to intelligently predict winding AC losses, and transmit it to the remote server through wireless communication to realize distributed monitor.

Benefits of technology

It improves the measurement accuracy of the AC loss of the transformer winding, provides targeted management and early warning mechanisms, and ensures distributed monitoring and over-limit early warning of different transformers.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118604691B_ABST
    Figure CN118604691B_ABST
Patent Text Reader

Abstract

The present invention relates to a method for determining the AC loss of a transformer winding, and relates to the field of electric power facilities. The method comprises: using a time-sharing prediction model to intelligently predict the AC loss of the winding of a target transformer in the current time segment based on the reference data corresponding to multiple historical time segments before the current time segment, the configuration information of the target transformer, and the input AC voltage value of the target transformer; synchronously transmitting the AC loss of the winding and the transformer number of the target transformer to the power facility server. Through the present invention, it is possible to perform reliable prediction of the AC loss of the winding of the target transformer in the future time segment on the basis of a customized time-sharing prediction model, and the prediction result and the transformer number of the target transformer are packaged and transmitted to the power facility server, thereby solving the technical problem that the AC loss of the winding of different transformers in the future time segments is difficult to accurately calculate.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of electric power facilities, and in particular to a method for determining AC loss of a transformer winding. Background Art

[0002] As one of the important AC power facilities, the transformer, like all equipment, is not perfect. Although the ideal transformer has no loss, the real transformer has power loss, and the output power of the transformer is always slightly less than the input power of the transformer. These power losses eventually become heat that must be removed from the transformer. For example, during the use of the transformer, its winding AC loss is related to the primary winding resistance, primary winding inductance, secondary winding capacitance, secondary winding inductance, primary winding current and secondary winding current. The transformer winding AC loss can be calculated using a standard calculation model with primary winding resistance, primary winding inductance, secondary winding capacitance, secondary winding inductance, primary winding current and secondary winding current as input parameters.

[0003] For example, the Chinese invention patent publication CN116184272A proposes a high-voltage and high-frequency planar transformer and a winding loss evaluation method thereof, wherein the high-voltage and high-frequency planar transformer winding loss evaluation method comprises: determining the winding magnetic field boundary conditions according to the winding current distribution characteristics; determining the copper layer current distribution according to the winding magnetic field boundary conditions; and determining the winding loss of the high-frequency and high-voltage planar transformer according to the copper layer current distribution. The present invention can improve the voltage field of planar transformer applications and improve the calculation accuracy of winding loss.

[0004] For example, the Chinese invention patent publication CN117783961A proposes a method and device for determining the AC loss of a high-frequency transformer winding, wherein the method and device can decompose the two-dimensional magnetic field in the core window to obtain multiple one-dimensional magnetic fields, and divide the first distributed winding of the first hollow winding on the vertical section of the core window into regions according to the magnetic field direction of the first magnetic field and the shape type of the first hollow winding to obtain multiple regional windings, and determine the first current component in each regional winding, the first current component corresponding to the first magnetic field, and determine the AC loss of the unit length winding in the core window according to the first current component in the regional winding. The present invention provides a method for determining the AC loss of the unit length winding in the core window, and effectively realizes the calculation of the AC loss of the unit length winding in the core window.

[0005] However, since the power consumed by the transformer from the beginning of use to the uninterrupted use process will be continuously converted into heat, the temperature of various components including the primary winding and the secondary winding will continue to change, which in turn causes the primary winding resistance, primary winding inductance, secondary winding capacitance and secondary winding inductance to continue to change. Therefore, the transformer winding AC loss in the future time interval calculated using the standard calculation mode is often not an accurate value. It can be seen that how to improve the measurement accuracy of the transformer winding AC loss in the future time interval is one of the problems that need to be solved at present. Summary of the invention

[0006] In order to solve the technical problems in the related fields, the present invention provides a method for determining the AC loss of a transformer winding, designs a time-sharing prediction model with a corresponding customized structure for a target transformer, and selectively screens a plurality of sufficient and in-depth basic data for performing intelligent prediction of the AC loss of the winding of the target transformer in a current time segment which is a future time segment. The AC loss of the winding obtained by the intelligent prediction is packaged together with the transformer number of the target transformer and wirelessly transmitted to a remote power facility server, thereby providing a high-precision solution mechanism for measuring the AC loss of the winding of the transformer in the future time interval, and realizing distributed monitoring of different transformers.

[0007] According to the present invention, a method for determining AC loss of a transformer winding is provided, the method comprising:

[0008] Measure the reference data corresponding to the target transformer in the multiple historical time segments before the current time segment, wherein the single temperature data corresponding to each historical time segment of the target transformer is the primary winding wire temperature, the secondary winding wire temperature, the primary winding current value, the secondary winding current value and the transformer oil temperature of the target transformer at each evenly spaced time in the historical time segment;

[0009] Acquire various configuration information of the target transformer, wherein the various configuration information of the target transformer is the number of primary winding turns, the zero-degree resistance of the primary winding, the zero-degree inductance of the primary winding, the number of secondary winding turns, the zero-degree resistance of the secondary winding, the zero-degree inductance of the secondary winding, the volume and weight of transformer oil;

[0010] Performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after the multiple learning operations, outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, wherein the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer;

[0011] The time-sharing prediction model is used to intelligently predict the winding AC loss of the target transformer in the current time segment based on the reference data corresponding to multiple historical time segments before the current time segment, the configuration information of the target transformer and the input AC voltage value of the target transformer;

[0012] Packing the AC loss of the winding of the target transformer obtained by intelligent prediction in the current time segment together with the transformer number of the target transformer to obtain a data packet, and wirelessly transmitting the data packet to a remote power facility server through a wireless communication link;

[0013] Wherein, measuring the reference data corresponding to the target transformer in the multiple historical time segments before the current time segment, the single temperature data corresponding to the target transformer in each historical time segment is the primary winding wire temperature, secondary winding wire temperature, primary winding current value, secondary winding current value and transformer oil body temperature of the target transformer at each evenly spaced moment in the historical time segment, and also includes: the current time segment takes the current moment as the starting time point, and the current time segment and the multiple historical time segments before the current time segment form a complete time interval;

[0014] Wherein, measuring the reference data corresponding to the target transformer in the multiple historical time segments before the current time segment, the single temperature data corresponding to the target transformer in each historical time segment is the primary winding wire temperature, secondary winding wire temperature, primary winding current value, secondary winding current value and transformer oil body temperature of the target transformer at each evenly spaced moment in the historical time segment, and further includes: the current time segment is the same as the duration of each time segment in the multiple historical time segments before the current time segment;

[0015] Among them, various configuration information of the target transformer is obtained, and the various configuration information of the target transformer is the primary winding turns, primary winding zero-degree resistance, primary winding zero-degree inductance, secondary winding turns, secondary winding zero-degree resistance, secondary winding zero-degree inductance, transformer oil volume and weight of the target transformer, including: the primary winding zero-degree resistance, primary winding zero-degree inductance, secondary winding zero-degree resistance and secondary winding zero-degree inductance of the target transformer are the primary winding resistance, primary winding inductance, secondary winding resistance and secondary winding inductance of the target transformer measured at zero degrees respectively.

[0016] It can be seen that the present invention has at least the following outstanding substantive features:

[0017] First, in order to perform intelligent prediction of the winding AC loss of the target transformer in the current time segment as the future time segment, a time-sharing prediction model with a corresponding customized structure is designed for the target transformer. The time-sharing prediction model is a radial basis neural network after completing multiple learning operations, and the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer, thereby ensuring the validity and stability of the intelligent prediction data;

[0018] Second, in order to perform intelligent prediction of the winding AC loss of the target transformer in the current time segment as the future time segment, multiple basic information is selectively selected in a sufficient and in-depth manner, wherein the multiple basic information includes reference data corresponding to multiple historical time segments of the target transformer before the current time segment, various configuration information of the target transformer, and the input AC voltage value of the target transformer;

[0019] Third: Specifically, the single temperature data corresponding to each historical time segment of the target transformer is the primary winding wire temperature, secondary winding wire temperature, primary winding current value, secondary winding current value and transformer oil temperature of the target transformer at each evenly spaced moment in the historical time segment, and the various configuration information of the target transformer is the primary winding turns, primary winding zero-degree resistance, primary winding zero-degree inductance, secondary winding turns, secondary winding zero-degree resistance, secondary winding zero-degree inductance, transformer oil volume and weight of the target transformer. The targeted screening of multiple basic information in a sufficient and in-depth manner further ensures the effectiveness and stability of the intelligent prediction data.

[0020] Fourthly, in each learning operation performed on the radial basis neural network, the AC loss of the winding of the target transformer obtained by measurement in a certain historical time segment is used as a single output content of the radial basis neural network, and the reference data corresponding to multiple historical time segments before the certain historical time segment, the configuration information of the target transformer and the input AC voltage value of the target transformer are used as multiple input contents of the radial basis neural network to complete this learning operation, so as to ensure the learning effect of each learning operation of the radial basis neural network;

[0021] Fifth: The winding AC loss of the target transformer in the current time segment, which is intelligently predicted by a customized time-sharing prediction model based on multiple basic information, is packaged together with the transformer number of the target transformer to obtain a data packet, and the data packet is wirelessly transmitted to a remote power facility server through a wireless communication link, thereby realizing distributed monitoring of different transformers, and providing valuable reference content for targeted management and early warning of whether there is excessive winding AC loss in different transformers in future time segments. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The embodiments of the present invention will be described below with reference to the accompanying drawings, wherein:

[0023] Figure 1 It is a technical flow chart of the method for determining the AC loss of transformer windings according to the present invention.

[0024] Figure 2 The present invention is a flowchart of a method for determining AC loss of a transformer winding according to a first embodiment of the present invention.

[0025] Figure 3 The present invention is a flowchart of a method for determining AC loss of a transformer winding according to a second embodiment of the present invention.

[0026] Figure 4 The present invention is a flowchart of a method for determining AC loss of a transformer winding according to a third embodiment of the present invention.

[0027] Figure 5 The fourth embodiment of the present invention is a flowchart of a method for determining AC loss of a transformer winding.

[0028] Figure 6 It is a flowchart of the steps of a method for determining AC loss of a transformer winding according to a fifth embodiment of the present invention. DETAILED DESCRIPTION

[0029] like Figure 1 As shown, a technical flow chart of a method for determining AC loss of a transformer winding according to the present invention is given.

[0030] like Figure 1 As shown, the present invention is used to solve the technical problem that it is difficult to accurately calculate the AC loss of the transformer winding in the future. The specific technical process of the present invention is as follows:

[0031] Technical process A: A time-sharing prediction model with a corresponding customized structure is designed for intelligent prediction of the winding AC loss of the target transformer in the future time segment. The time-sharing prediction model is a radial basis neural network after completing multiple learning operations;

[0032] like Figure 1 As shown, the radial basis neural network includes an input layer, an output layer and an intermediate layer, wherein the intermediate layer is arranged between the input layer and the output layer;

[0033] For example, the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer, so that time-sharing prediction models with different structures are designed for different transformers, ensuring the validity and stability of intelligent prediction data;

[0034] For example, in each learning operation performed on the radial basis function neural network, the AC loss of the winding of the target transformer obtained by measurement in a certain historical time segment is used as a single output content of the radial basis function neural network, and the reference data corresponding to multiple historical time segments before the certain historical time segment, the various configuration information of the target transformer, and the input AC voltage value of the target transformer are used as multiple input contents of the radial basis function neural network to complete this learning operation, so as to ensure the learning effect of each learning operation of the radial basis function neural network;

[0035] Specifically, the future time segment is selected as the current time segment, the current time segment takes the current time as the starting time, and belongs to the current unfinished time segment, that is, the future time segment;

[0036] Technical process B: targeted selection of sufficient and in-depth basic information for intelligent prediction of winding AC loss of target transformer in future time segments;

[0037] For example, the multiple pieces of basic information include reference data corresponding to multiple historical time segments of the target transformer before the current time segment, various configuration information of the target transformer, and an input AC voltage value of the target transformer;

[0038] Specifically, the single temperature data corresponding to each historical time segment of the target transformer is the primary winding wire temperature, the secondary winding wire temperature, the primary winding current value, the secondary winding current value and the transformer oil temperature of the target transformer at each uniformly spaced time in the historical time segment, and the various configuration information of the target transformer is the primary winding turns, the primary winding zero-degree resistance, the primary winding zero-degree inductance, the secondary winding turns, the secondary winding zero-degree resistance, the secondary winding zero-degree inductance, the transformer oil volume and weight of the target transformer;

[0039] The above-mentioned comprehensive and in-depth targeted screening of multiple basic information further ensures the effectiveness and stability of intelligent prediction data;

[0040] Technical process C: The customized time-sharing prediction model designed for the target transformer using technical process A intelligently predicts the winding AC loss of the target transformer in the current time segment based on the sufficient and in-depth multiple basic information targeted by technical process B;

[0041] Technical process D: Packing the AC loss of the winding of the target transformer intelligently predicted in the current time segment by technical process C together with the transformer number of the target transformer to obtain a data packet, and wirelessly transmitting the data packet to a remote power facility server via a wireless communication link;

[0042] In this way, distributed monitoring of different transformers is achieved, providing valuable reference content for targeted management and early warning of whether there is excessive winding AC loss in different transformers in future time segments.

[0043] The key points of the present invention are: targeted design of each learning operation of the radial basis neural network, construction of time-sharing prediction models with different customized structures for different transformers, and targeted screening of multiple basic information for intelligent prediction of winding AC losses in future time segments of the target transformer.

[0044] The method for determining the AC loss of a transformer winding according to the present invention will be specifically described below by way of an embodiment.

[0045] First embodiment

[0046] Figure 2 The present invention is a flowchart of a method for determining AC loss of a transformer winding according to a first embodiment of the present invention.

[0047] like Figure 2 As shown, the method for determining the AC loss of the transformer winding comprises the following steps:

[0048] Step S101: measuring reference data of a target transformer corresponding to a plurality of historical time segments before a current time segment, wherein the single temperature data corresponding to each historical time segment of the target transformer is the primary winding wire temperature, the secondary winding wire temperature, the primary winding current value, the secondary winding current value and the transformer oil temperature of the target transformer at each evenly spaced moment in the historical time segment;

[0049] For example, multiple physical parameter detection devices may be selected to respectively detect the primary winding wire temperature, the secondary winding wire temperature, the primary winding current value, the secondary winding current value, and the transformer oil body temperature at each moment;

[0050] Step S102: Acquire various configuration information of the target transformer, wherein the various configuration information of the target transformer includes the number of primary winding turns, the zero-degree resistance of the primary winding, the zero-degree inductance of the primary winding, the number of secondary winding turns, the zero-degree resistance of the secondary winding, the zero-degree inductance of the secondary winding, the volume and weight of transformer oil;

[0051] Specifically, since the primary winding resistance, primary winding inductance, secondary winding resistance and secondary winding inductance of the target transformer correspond to different physical quantities at different temperatures, the primary winding resistance, primary winding inductance, secondary winding resistance and secondary winding inductance of the target transformer are constantly changing during the continuous use of the target transformer;

[0052] Here, the primary winding zero-degree resistance, the primary winding zero-degree inductance, the secondary winding zero-degree resistance and the secondary winding zero-degree inductance of the target transformer are physical quantities corresponding to the primary winding resistance, the primary winding inductance, the secondary winding resistance and the secondary winding inductance of the target transformer at an ambient temperature of zero degrees, respectively. The primary winding zero-degree resistance, the primary winding zero-degree inductance, the secondary winding zero-degree resistance and the secondary winding zero-degree inductance of the target transformer are taken as one of the multiple basic data, which can provide reference parameters for the primary winding resistance, the primary winding inductance, the secondary winding resistance and the secondary winding inductance of the target transformer;

[0053] Step S103: performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after completing the multiple learning operations, and outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, wherein the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer;

[0054] Specifically, the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer, including: the weight of the target transformer is 2.5 tons, the corresponding number of learning operations of the radial basis neural network is 100, the weight of the target transformer is 5.0 tons, the corresponding number of learning operations of the radial basis neural network is 150, the weight of the target transformer is 10.0 tons, the corresponding number of learning operations of the radial basis neural network is 200, and so on;

[0055] Step S104: using the time-sharing prediction model to intelligently predict the winding AC loss of the target transformer in the current time segment based on the reference data corresponding to the multiple historical time segments before the current time segment, the configuration information of the target transformer, and the input AC voltage value of the target transformer;

[0056] For example, an FPGA device may be selected to implement the simulation and test of a data processing process for intelligently predicting the AC loss of the winding of the target transformer in the current time segment using the time-sharing prediction model based on the reference data corresponding to the multiple historical time segments before the current time segment, the configuration information of the target transformer, and the input AC voltage value of the target transformer;

[0057] Step S105: Packing the AC loss of the winding of the target transformer obtained by intelligent prediction in the current time segment together with the transformer number of the target transformer to obtain a data packet, and wirelessly transmitting the data packet to a remote power facility server through a wireless communication link;

[0058] Specifically, the AC loss of the winding of the target transformer obtained by intelligent prediction in the current time segment is packaged together with the transformer number of the target transformer to obtain a data packet, and the data packet is wirelessly transmitted to a remote power facility server through a wireless communication link, including: the wireless communication link is a time division duplex communication link or a frequency division duplex communication link;

[0059] Wherein, measuring the reference data corresponding to the target transformer in the multiple historical time segments before the current time segment, the single temperature data corresponding to the target transformer in each historical time segment is the primary winding wire temperature, secondary winding wire temperature, primary winding current value, secondary winding current value and transformer oil body temperature of the target transformer at each evenly spaced moment in the historical time segment, and also includes: the current time segment takes the current moment as the starting time point, and the current time segment and the multiple historical time segments before the current time segment form a complete time interval;

[0060] For example, the current time segment takes the current moment as the starting time point, and the current time segment and multiple historical time segments before the current time segment form a complete time interval including: the current time segment is from 5 pm to 6 pm, the current moment is 5 pm, and it is obvious that the current time segment is a future time segment, and the multiple historical time segments before the current time segment are from 10 am to 11 am, 11 am to 12 noon, 12 noon to 1 pm, 1 pm to 2 pm, 2 pm to 3 pm, 3 pm to 4 pm, and 4 pm to 5 pm, a total of 7 historical time segments;

[0061] Wherein, measuring the reference data corresponding to the target transformer in the multiple historical time segments before the current time segment, the single temperature data corresponding to the target transformer in each historical time segment is the primary winding wire temperature, secondary winding wire temperature, primary winding current value, secondary winding current value and transformer oil body temperature of the target transformer at each evenly spaced moment in the historical time segment, and further includes: the current time segment is the same as the duration of each time segment in the multiple historical time segments before the current time segment;

[0062] For example, the duration of the current time segment being the same as that of each of the multiple historical time segments before the current time segment includes: the duration of the current time segment being the same as that of each of the multiple historical time segments before the current time segment, both being 1 hour;

[0063] Among them, obtaining various configuration information of the target transformer, the various configuration information of the target transformer is the primary winding turns, the primary winding zero-degree resistance, the primary winding zero-degree inductance, the secondary winding turns, the secondary winding zero-degree resistance, the secondary winding zero-degree inductance, the transformer oil volume and weight of the target transformer, including: the primary winding zero-degree resistance, the primary winding zero-degree inductance, the secondary winding zero-degree resistance and the secondary winding zero-degree inductance of the target transformer are the primary winding resistance, the primary winding inductance, the secondary winding resistance and the secondary winding inductance of the target transformer measured at zero degrees respectively;

[0064] Among them, performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after completing the multiple learning operations, and outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, and the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer, including: in each learning operation performed on the radial basis neural network, the AC loss of the winding of the target transformer obtained by measurement in a certain historical time segment is used as a single output content of the radial basis neural network, and each reference data corresponding to multiple historical time segments before the certain historical time segment of the target transformer, each configuration information of the target transformer and the input AC voltage value of the target transformer are used as multiple input contents of the radial basis neural network to complete this learning operation;

[0065] And wherein, the reference data corresponding to the target transformer in multiple historical time segments before the current time segment are measured, and the single temperature data corresponding to the target transformer in each historical time segment is the primary winding wire temperature, secondary winding wire temperature, primary winding current value, secondary winding current value and transformer oil body temperature of the target transformer at each evenly spaced moment in the historical time segment, including: the number of each evenly spaced moment in each historical time segment is proportional to the input AC voltage value of the target transformer.

[0066] Second embodiment

[0067] Figure 3 The present invention is a flowchart of a method for determining AC loss of a transformer winding according to a second embodiment of the present invention.

[0068] like Figure 3 As shown, compared with Figure 2 After step S105, the method further includes:

[0069] Step S106: receiving the AC loss of the winding of the target transformer in the current time segment obtained by intelligent prediction, and when the AC loss of the winding of the target transformer in the current time segment obtained by intelligent prediction is greater than or equal to a set loss threshold, executing an early alarm action for excessive AC loss of the winding in the current time segment;

[0070] For example, receiving the winding AC loss of the target transformer obtained by intelligent prediction in the current time segment, and when the winding AC loss of the target transformer obtained by intelligent prediction in the current time segment is greater than or equal to a set loss threshold, executing an early alarm action for the winding AC loss exceeding the limit in the current time segment includes: the early alarm action for the winding AC loss exceeding the limit in the current time segment is a photoelectric alarm action with a set flashing frequency.

[0071] Third embodiment

[0072] Figure 4 The present invention is a flowchart of a method for determining AC loss of a transformer winding according to a third embodiment of the present invention.

[0073] like Figure 4 As shown, compared with Figure 2 After step S104, the method further includes:

[0074] Step S107: receiving the AC loss of the winding of the target transformer in the current time segment obtained by intelligent prediction, and synchronously displaying the AC loss of the winding of the target transformer in the current time segment obtained by intelligent prediction and the transformer number of the target transformer;

[0075] For example, receiving the winding AC loss of the target transformer obtained by intelligent prediction in the current time segment, and synchronously displaying the winding AC loss of the target transformer obtained by intelligent prediction in the current time segment and the transformer number of the target transformer, including: the winding AC loss of the target transformer in the current time segment and the transformer number of the target transformer are respectively expressed in numerical form.

[0076] Fourth embodiment

[0077] Figure 5 The fourth embodiment of the present invention is a flowchart of a method for determining AC loss of a transformer winding.

[0078] like Figure 5 As shown, compared with Figure 2 After step S103, the method further includes:

[0079] Step S108: receiving the time-sharing prediction model, and completing the storage of the time-sharing prediction model by storing various model parameters of the time-sharing prediction model;

[0080] For example, receiving the time-sharing prediction model and completing the storage of the time-sharing prediction model by storing various model parameters of the time-sharing prediction model includes: selecting FLASH flash memory, TF storage chip or MMC storage chip to perform storage of various model parameters of the time-sharing prediction model.

[0081] Fifth embodiment

[0082] Figure 6 It is a flowchart of the steps of a method for determining AC loss of a transformer winding according to a fifth embodiment of the present invention.

[0083] like Figure 6 As shown, compared with Figure 2 After step S101, the method further includes:

[0084] Step S109: when the average temperature of the primary winding wire or the average temperature of the secondary winding wire of the target transformer at each evenly spaced time in a certain historical time segment is greater than or equal to the set temperature limit, executing an alarm action corresponding to the winding wire temperature exceeding the limit;

[0085] For example, an acoustic alarm device is selected to perform an alarm action corresponding to the winding wire temperature exceeding the limit when the average primary winding wire temperature or the average secondary winding wire temperature of the target transformer at each evenly spaced moment in a certain historical time segment is greater than or equal to the set temperature limit.

[0086] Next, various embodiments of the present invention will be further described.

[0087] In each of the above embodiments, optionally, in the method for determining the AC loss of the transformer winding:

[0088] Adopting the time-sharing prediction model to intelligently predict the winding AC loss of the target transformer in the current time segment based on the reference data corresponding to the multiple historical time segments before the current time segment, the configuration information of the target transformer and the input AC voltage value of the target transformer includes: synchronously inputting the reference data corresponding to the multiple historical time segments before the current time segment, the configuration information of the target transformer and the input AC voltage value of the target transformer into the time-sharing prediction model;

[0089] Specifically, a synchronous control device is selected to realize synchronous input of the reference data corresponding to the multiple historical time segments of the target transformer before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer to the time-sharing prediction model;

[0090] Wherein, synchronously inputting the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer into the time-sharing prediction model includes: before synchronously inputting the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer into the time-sharing prediction model, respectively performing numerical normalization processing on the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer;

[0091] Among them, using the time-sharing prediction model to intelligently predict the winding AC loss of the target transformer in the current time segment based on the reference data corresponding to multiple historical time segments before the current time segment, the configuration information of the target transformer and the input AC voltage value of the target transformer also includes: executing the time-sharing prediction model to obtain the winding AC loss of the target transformer in the current time segment output by the time-sharing prediction model;

[0092] Among them, executing the time-sharing prediction model to obtain the winding AC loss of the target transformer in the current time segment output by the time-sharing prediction model includes: the winding AC loss of the target transformer in the current time segment output by the time-sharing prediction model is a representation after numerical normalization.

[0093] In each of the above embodiments, optionally, in the method for determining the AC loss of the transformer winding:

[0094] Packing the AC loss of the winding of the target transformer obtained by intelligent prediction in the current time segment together with the transformer number of the target transformer to obtain a data packet, and wirelessly transmitting the data packet to a remote power facility server through a wireless communication link includes: the remote power facility server is a big data server, a cloud computing server or a blockchain server;

[0095] Alternatively, the winding AC loss of the target transformer obtained by intelligent prediction in the current time segment is packaged together with the transformer number of the target transformer to obtain a data packet, and the data packet is wirelessly transmitted to a remote power facility server through a wireless communication link, which also includes: using a big data service node, a cloud computing service node or a blockchain service node to respectively implement a big data server, a cloud computing server or a blockchain server.

[0096] Among them, the winding AC loss of the target transformer obtained by intelligent prediction in the current time segment is packaged together with the transformer number of the target transformer to obtain a data packet, and the data packet is wirelessly transmitted to a remote power facility server through a wireless communication link, which also includes: the winding AC loss of the target transformer obtained by intelligent prediction in the current time segment is packaged together with the transformer number of the target transformer into an IP data packet, and the IP data packet is wirelessly transmitted to a remote power facility server through a wireless communication link.

[0097] And in each of the above embodiments, optionally, in the method for determining the AC loss of the transformer winding:

[0098] Performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after the multiple learning operations, outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, wherein the number of learning operations of the radial basis neural network is positively associated with the weight of the target transformer further comprises: using a numerical mapping formula to represent a numerical mapping relationship of the positive association between the number of learning operations of the radial basis neural network and the weight of the target transformer;

[0099] For example, using a numerical mapping formula to represent the numerical mapping relationship of the number of learning operations of the radial basis neural network and the positive correlation of the weight of the target transformer includes: selecting a numerical simulation mode to complete the simulation and testing of a data processing process of using a numerical mapping formula to represent the numerical mapping relationship of the number of learning operations of the radial basis neural network and the positive correlation of the weight of the target transformer;

[0100] Among them, performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after completing the multiple learning operations, outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, and the number of learning operations of the radial basis neural network is positively associated with the weight of the target transformer also includes: in the numerical mapping formula, the weight of the target transformer is an input parameter of the numerical mapping formula;

[0101] Among them, performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after completing the multiple learning operations, outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, and the number of learning operations of the radial basis neural network is positively associated with the weight of the target transformer further comprising: in the numerical mapping formula, the number of learning operations of the radial basis neural network corresponding to the weight of the target transformer is the output parameter of the numerical mapping formula;

[0102] Wherein, the reference data corresponding to the multiple historical time segments before the current time segment of the target transformer are measured, and the single temperature data corresponding to each historical time segment of the target transformer is the primary winding wire temperature, the secondary winding wire temperature, the primary winding current value, the secondary winding current value and the transformer oil body temperature of the target transformer at each uniformly spaced moment in the historical time segment, including: the primary winding wire temperature, the secondary winding wire temperature and the transformer oil body temperature of the target transformer at any moment in any historical time segment are all measured in real time using a non-contact temperature sensor;

[0103] Among them, the primary winding wire temperature, the secondary winding wire temperature and the transformer oil body temperature of the target transformer at any time in any historical time segment are all measured in real time using non-contact temperature sensors, including: using three non-contact temperature sensors to respectively perform real-time measurement of the primary winding wire temperature, the secondary winding wire temperature and the transformer oil body temperature of the target transformer at any time in any historical time segment.

[0104] In addition, in the method for determining the AC loss of the transformer winding according to the present invention, there are also the following several technical features to highlight the essential features of the present invention:

[0105] Before synchronously inputting the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer into the time-sharing prediction model, respectively performing numerical normalization processing on the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer, including: before synchronously inputting the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer into the time-sharing prediction model, respectively performing hexadecimal numerical conversion processing on the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer;

[0106] For example, a programmable logic device may be selected to perform hexadecimal value conversion processing on each reference data corresponding to a plurality of historical time segments before the current time segment, each configuration information of the target transformer, and the input AC voltage value of the target transformer;

[0107] Specifically, a programmable logic device can be selected to perform hexadecimal value conversion processing on each reference data corresponding to a plurality of historical time segments before the current time segment, each configuration information of the target transformer, and the input AC voltage value of the target transformer, including: the programmable logic device is a CPLD chip programmed and designed in VHDL language;

[0108] And wherein, the winding AC loss of the target transformer in the current time segment output by the time-sharing prediction model is a representation after numerical normalization, including: the winding AC loss of the target transformer in the current time segment output by the time-sharing prediction model is a hexadecimal numerical representation.

[0109] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0110] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0111] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A method for determining AC loss of transformer windings, characterized in that: The method comprises: Measure reference data corresponding to a plurality of historical time segments before a current time segment respectively of a target transformer, wherein the reference data is a single temperature data corresponding to each historical time segment of the target transformer; Acquire various configuration information of the target transformer, wherein the various configuration information of the target transformer is the number of primary winding turns, the zero-degree resistance of the primary winding, the zero-degree inductance of the primary winding, the number of secondary winding turns, the zero-degree resistance of the secondary winding, the zero-degree inductance of the secondary winding, the volume and weight of transformer oil; Performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after the multiple learning operations, outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, wherein the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer; The time-sharing prediction model is used to predict the winding AC loss of the target transformer in the current time segment based on the reference data, the configuration information of the target transformer and the input AC voltage value of the target transformer; Packing the predicted AC loss of the winding of the target transformer in the current time segment with the transformer number of the target transformer to obtain a data packet, and wirelessly transmitting the data packet to the power facility server via a wireless communication link; The single temperature data is the primary winding wire temperature, secondary winding wire temperature, primary winding current value, secondary winding current value and transformer oil temperature of the target transformer at each evenly spaced time in the historical time segment; The current time segment takes the current moment as the starting time point, the current time segment and multiple historical time segments before the current time segment form a complete time interval, and the current time segment has the same duration as each of the multiple historical time segments before the current time segment; Among them, obtaining various configuration information of the target transformer, the various configuration information of the target transformer is the primary winding turns, the primary winding zero-degree resistance, the primary winding zero-degree inductance, the secondary winding turns, the secondary winding zero-degree resistance, the secondary winding zero-degree inductance, the transformer oil volume and weight of the target transformer, including: the primary winding zero-degree resistance, the primary winding zero-degree inductance, the secondary winding zero-degree resistance and the secondary winding zero-degree inductance of the target transformer are the primary winding resistance, the primary winding inductance, the secondary winding resistance and the secondary winding inductance of the target transformer measured at zero degrees respectively; Among them, performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after completing the multiple learning operations, and outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, and the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer, including: in each learning operation performed on the radial basis neural network, the AC loss of the winding of the target transformer obtained by measurement in a certain historical time segment is used as a single output content of the radial basis neural network, and each reference data corresponding to multiple historical time segments before the certain historical time segment of the target transformer, each configuration information of the target transformer and the input AC voltage value of the target transformer are used as multiple input contents of the radial basis neural network to complete this learning operation; Wherein, the reference data corresponding to the target transformer in the multiple historical time segments before the current time segment are measured, and the single temperature data corresponding to the target transformer in each historical time segment is the primary winding wire temperature, secondary winding wire temperature, primary winding current value, secondary winding current value and transformer oil body temperature of the target transformer at each evenly spaced moment in the historical time segment, including: the number of each evenly spaced moment in each historical time segment is proportional to the input AC voltage value of the target transformer; Among them, before synchronously inputting the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer into the time-sharing prediction model, numerical normalization processing is performed on the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer, including: before synchronously inputting the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer into the time-sharing prediction model, hexadecimal numerical conversion processing is performed on the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the various configuration information of the target transformer and the input AC voltage value of the target transformer, and the AC loss of the winding of the target transformer in the current time segment output by the time-sharing prediction model is in hexadecimal numerical representation.

2. The method for determining the AC loss of transformer windings according to claim 1, characterized in that: After packaging the AC loss of the winding of the target transformer obtained by intelligent prediction in the current time segment together with the transformer number of the target transformer to obtain a data packet, and wirelessly transmitting the data packet to a remote power facility server through a wireless communication link, the method further includes: Receive the AC loss of the winding of the target transformer obtained by intelligent prediction in the current time segment. If the predicted AC loss of the winding of the target transformer in the current time segment is greater than or equal to the set loss threshold, perform an early alarm action for the excessive AC loss of the winding in the current time segment.

3. The method for determining the AC loss of transformer windings according to claim 2, characterized in that: After using the time-sharing prediction model to intelligently predict the winding AC loss of the target transformer in the current time segment based on the reference data corresponding to multiple historical time segments before the current time segment, the configuration information of the target transformer, and the input AC voltage value of the target transformer, the method further includes: Receive the AC loss of the winding of the target transformer in the current time segment obtained by intelligent prediction, and synchronously display the AC loss of the winding of the target transformer in the current time segment obtained by intelligent prediction and the transformer number of the target transformer.

4. The method for determining the AC loss of transformer windings according to claim 2, characterized in that: After performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after the multiple learning operations, outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, and the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer, the method further includes: The time-sharing prediction model is received, and the storage of the time-sharing prediction model is completed by storing various model parameters of the time-sharing prediction model.

5. The method for determining the AC loss of transformer windings according to claim 2, characterized in that: After measuring the reference data of the target transformer corresponding to a plurality of historical time segments before the current time segment, the single temperature data corresponding to each historical time segment of the target transformer is the primary winding wire temperature, the secondary winding wire temperature, the primary winding current value, the secondary winding current value and the transformer oil body temperature of the target transformer at each evenly spaced moment in the historical time segment, the method further includes: If the average temperature of the primary winding wire or the average temperature of the secondary winding wire of the target transformer at each evenly spaced time in a certain historical time segment is greater than or equal to the set temperature limit, an alarm action corresponding to the winding wire temperature exceeding the limit is executed.

6. The method for determining the AC loss of a transformer winding according to any one of claims 2 to 5, characterized in that: The time-sharing prediction model is used to intelligently predict the winding AC loss of the target transformer in the current time segment based on the reference data corresponding to the multiple historical time segments before the current time segment, the configuration information of the target transformer and the input AC voltage value of the target transformer, including: synchronously inputting the reference data corresponding to the multiple historical time segments before the current time segment, the configuration information of the target transformer and the input AC voltage value of the target transformer into the time-sharing prediction model; Wherein, the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the configuration information of the target transformer and the input AC voltage value of the target transformer are synchronously input into the time-sharing prediction model, including: before the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the configuration information of the target transformer and the input AC voltage value of the target transformer are synchronously input into the time-sharing prediction model, the reference data corresponding to the target transformer in multiple historical time segments before the current time segment, the configuration information of the target transformer and the input AC voltage value of the target transformer are respectively subjected to numerical normalization processing; Among them, the time-sharing prediction model is used to intelligently predict the winding AC loss of the target transformer in the current time segment based on the reference data corresponding to multiple historical time segments before the current time segment, the configuration information of the target transformer and the input AC voltage value of the target transformer, and also includes: executing the time-sharing prediction model to obtain the winding AC loss of the target transformer in the current time segment output by the time-sharing prediction model; Among them, executing the time-sharing prediction model to obtain the winding AC loss of the target transformer in the current time segment output by the time-sharing prediction model includes: the winding AC loss of the target transformer in the current time segment output by the time-sharing prediction model is a representation after numerical normalization.

7. The method for determining the AC loss of a transformer winding according to any one of claims 2 to 5, characterized in that: The AC loss of the winding of the target transformer obtained by intelligent prediction in the current time segment is packaged together with the transformer number of the target transformer to obtain a data packet, and the data packet is wirelessly transmitted to a remote power facility server through a wireless communication link, including: the remote power facility server is a big data server, a cloud computing server or a blockchain server; Among them, the winding AC loss of the target transformer obtained by intelligent prediction in the current time segment is packaged together with the transformer number of the target transformer to obtain a data packet, and the data packet is wirelessly transmitted to a remote power facility server through a wireless communication link. It also includes: the winding AC loss of the target transformer obtained by intelligent prediction in the current time segment is packaged together with the transformer number of the target transformer into an IP data packet, and the IP data packet is wirelessly transmitted to a remote power facility server through a wireless communication link.

8. The method for determining the AC loss of a transformer winding according to any one of claims 2 to 5, characterized in that: Performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after the multiple learning operations, outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, wherein the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer, further comprising: using a numerical mapping formula to represent a numerical mapping relationship of the positive correlation between the number of learning operations of the radial basis neural network and the weight of the target transformer; Among them, performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after completing the multiple learning operations, outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer, and also including: in the numerical mapping formula, the weight of the target transformer is an input parameter of the numerical mapping formula; Among them, performing multiple learning operations on the radial basis neural network to obtain the radial basis neural network after completing the multiple learning operations, outputting the radial basis neural network after the multiple learning operations as a time-sharing prediction model, and the number of learning operations of the radial basis neural network is positively correlated with the weight of the target transformer, and also includes: in the numerical mapping formula, the number of learning operations of the radial basis neural network corresponding to the weight of the target transformer is the output parameter of the numerical mapping formula; Wherein, the reference data corresponding to the target transformer in multiple historical time segments before the current time segment are measured, and the single temperature data corresponding to each historical time segment of the target transformer is the primary winding wire temperature, secondary winding wire temperature, primary winding current value, secondary winding current value and transformer oil body temperature of the target transformer at each evenly spaced moment in the historical time segment, including: the primary winding wire temperature, secondary winding wire temperature and transformer oil body temperature of the target transformer at any moment in any historical time segment are all measured in real time using a non-contact temperature sensor; Among them, the primary winding wire temperature, the secondary winding wire temperature and the transformer oil body temperature of the target transformer at any time in any historical time segment are all measured in real time using non-contact temperature sensors, including: using three non-contact temperature sensors to respectively perform real-time measurement of the primary winding wire temperature, the secondary winding wire temperature and the transformer oil body temperature of the target transformer at any time in any historical time segment.