Transformer load monitoring method and system based on deep learning, medium and equipment
By building a transformer load monitoring model based on deep learning, the complex and time-consuming problem of traditional detection processes is solved, fast and accurate load status monitoring is achieved, and the safety and efficiency of power grid operation is improved.
Patent Information
- Application Number
- CN202411654890.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-19
- Publication Date
- 2025-02-25
AI Technical Summary
The load state detection process of traditional transformers is complex and time-consuming, which affects the economical and response speed of the power grid operation.
Using a deep learning-based method, the transformer load current waveform and data is collected in real time, the objective function is defined, the gain function is determined, the tree structure is gradually added, and the final transformer load monitoring model is constructed.
Fast and accurate transformer load status monitoring is achieved, the safety and efficiency of power grid operation is improved, and the time and complexity of manual inspection is reduced.
Smart Images
Figure CN119514369A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of transformer state monitoring, and in particular to a transformer load monitoring method, system, medium and equipment based on deep learning. Background Art
[0002] A transformer is a device that uses the principle of electromagnetic induction to change AC voltage. It consists of a primary coil, a secondary coil, and an iron core. The main functions are: voltage conversion, current conversion, impedance conversion, electrical isolation, voltage stabilization, etc. The transformer is equivalent to a safety switch in the power system. Once the transformer is overloaded, the power system is prone to problems, which in turn leads to loss of personnel or property. The ability to detect the load state of the transformer has become an important indicator for detecting the safety performance of the transformer. The faster the detection speed of the transformer load state, the shorter the response time of the power dispatcher, and the faster and more accurate the dispatch of the power system. Therefore, it is necessary to quickly monitor the load state of the transformer online.
[0003] The traditional transformer load status detection process requires technicians to measure many parameters, the process is relatively complicated, and it takes a lot of time, which is not conducive to the economic efficiency of power grid operation. Summary of the invention
[0004] Based on this, it is necessary to propose a transformer load monitoring method, system, medium and equipment based on deep learning to address the above problems.
[0005] A transformer load monitoring method based on deep learning, the method comprising:
[0006] Collect transformer load current waveform and transformer load data in real time.
[0007] An objective function is defined according to the transformer load current waveform and transformer load data.
[0008] The gain function is determined by the first-order derivative and the second-order derivative of the objective function.
[0009] Based on the gain function, the transformer load data is used as the monitoring target, and the transformer load current waveform is used as the feature input for modeling, and the final transformer load monitoring model is constructed by gradually adding the latest tree.
[0010] When the transformer is connected to the grid and in operation, the transformer load current waveform measured by the current sensor is input into the final transformer load monitoring model to obtain the transformer load result.
[0011] The real-time collection of transformer load current waveform and transformer load data specifically includes:
[0012] Simulate different load states of the transformer to obtain the transformer load current waveform;
[0013] Acquiring transformer current data in the transformer load current waveform through a current sensor;
[0014] Transformer load data is acquired according to the transformer current data, and the transformer load data is preprocessed.
[0015] The step of defining the objective function according to the transformer load current waveform and transformer load data specifically includes:
[0016] according to Define the objective function, where Obj is the objective function, y i is the true value, is the monitoring value of the transformer load data corresponding to the i-th transformer load current waveform, l is the loss function, Ω(f k ) is the regularization term, and K is the number of trees.
[0017] The step of determining a gain function by using the first-order derivative and the second-order derivative of the objective function and adaptively updating the model parameters by using the gain function specifically includes:
[0018] according to Determine the gain function, where G is the gain function, L is the left child node, R is the right child node, G L is the first-order derivative of the left child node, G R is the first-order derivative of the right child node, H L is the second-order derivative of the left child node, H R is the second-order derivative of the right child node, γ is the leaf number regularization parameter, and λ is the leaf weight regularization parameter.
[0019] Among them, based on the gain function, the transformer load data is used as the monitoring target, the transformer load current waveform is used as the feature input for modeling, and the final transformer load monitoring model is constructed by gradually adding the latest tree, which specifically includes:
[0020] The transformer load data is used as a monitoring target, and the transformer load current waveform is used as a feature input for modeling to determine an initial transformer load monitoring model.
[0021] Based on the gain function, an optimal splitting point is selected according to the first-order derivative and the second-order derivative when the objective function is minimized, and a left child node and a right child node are determined according to the optimal splitting point.
[0022] Generate an updated tree through the left child node and the right child node.
[0023] The latest tree is fitted into the initial transformer load monitoring model, and the transformer load monitoring model of the current round is determined by gradually adding the latest tree.
[0024] The transformer load monitoring model of the current round is used for monitoring, the monitoring value of the current round is obtained, and the residual and the gradient are updated.
[0025] Repeat the above steps until the preset number of trees is reached to determine the final transformer load monitoring model.
[0026] The method of taking the transformer load as the monitoring target and taking the transformer load current waveform as the characteristic input for modeling to determine the initial transformer load monitoring model specifically includes:
[0027] according to Determine the initial transformer load monitoring model, where: is the monitoring value corresponding to the i-th input feature, x is the input feature, i.e., the transformer load current waveform, f k is the Kth tree.
[0028] The above steps are repeated until a preset number of trees is reached to determine the final transformer load monitoring model, which specifically includes:
[0029] according to Determine the final transformer load monitoring model, where: is the final monitoring value of the final transformer load monitoring model, x new is the input feature to be predicted, N is the preset number of trees, f k is the Kth tree.
[0030] Among them, based on the gain function, taking the transformer load as the monitoring target, taking the transformer load current waveform as the feature input for modeling, and constructing the final transformer load monitoring model by gradually adding the latest tree, it also specifically includes:
[0031] The model parameters of the final transformer load monitoring model are updated using an adaptive learning rate method.
[0032] A transformer load monitoring system based on deep learning, the system comprising:
[0033] The acquisition module is used to acquire the transformer load current waveform and transformer load data in real time.
[0034] The objective function definition module is used to define the objective function according to the transformer load current waveform and the transformer load data.
[0035] The gain function determination module is used to determine the gain function through the first-order derivative and the second-order derivative of the objective function.
[0036] The final transformer load monitoring model determination module is used to build the final transformer load monitoring model by gradually adding the latest tree based on the gain function, taking the transformer load data as the monitoring target and the transformer load current waveform as the feature input for modeling.
[0037] The transformer load monitoring module is used to input the transformer load current waveform measured by the current sensor into the final transformer load monitoring model when the transformer is connected to the grid and operate, so as to obtain the transformer load result.
[0038] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the steps of the method described above.
[0039] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.
[0040] The embodiments of the present invention have the following beneficial effects:
[0041] The present invention defines an objective function according to the transformer load current waveform and transformer load data, determines a gain function through the first-order derivative and the second-order derivative of the objective function, and can guide the update direction of the model parameters through the gain function, ensuring that the model gradually approaches the optimal solution during the training process. Furthermore, based on the gain function, with the load data as the monitoring target and the load current waveform as the feature input, the complexity of the final transformer load monitoring model is gradually enhanced by gradually adding the latest tree, thereby improving the stability and reliability of the model. According to the final transformer load monitoring model, the future load trend of the transformer is accurately predicted to obtain the transformer load result, and potential load anomalies are discovered in time, thereby improving the safety and efficiency of power grid operation. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0043] in:
[0044] Figure 1 A flowchart of an embodiment of a transformer load monitoring method based on deep learning provided by the present invention;
[0045] Figure 2 A flowchart of another embodiment of a transformer load monitoring method based on deep learning provided by the present invention;
[0046] Figure 3 A schematic structural diagram of an embodiment of a transformer load monitoring system based on deep learning provided by the present invention;
[0047] Figure 4 A schematic structural diagram of an embodiment of a medium provided by the present invention;
[0048] Figure 5 A schematic structural diagram of an embodiment of the device provided by the present invention. DETAILED DESCRIPTION
[0049] 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.
[0050] like Figure 1 As shown, Figure 1 A flowchart of an embodiment of a transformer load monitoring method based on deep learning provided by the present invention. A transformer load monitoring method based on deep learning, the method comprising:
[0051] S101: Collect transformer load current waveform and transformer load data in real time.
[0052] Exemplarily, different load states of the transformer are simulated to obtain the transformer load current waveform; transformer current data in the transformer load current waveform is obtained through a current sensor; transformer load data is obtained based on the transformer current data, and the transformer load data is preprocessed.
[0053] S102: defining an objective function according to the transformer load current waveform and transformer load data.
[0054] Exemplarily, the objective function is determined according to the following formula:
[0055]
[0056] Among them, Obj is the objective function, y i is the true value, is the monitoring value of the transformer load data corresponding to the i-th transformer load current waveform, l is the loss function, Ω(f k ) is the regularization term, and K is the number of trees.
[0057] S103: determining a gain function through the first-order derivative and the second-order derivative of the objective function, and adaptively updating the model parameters through the gain function.
[0058] Exemplarily, the gain function is determined according to the following formula:
[0059]
[0060] Among them, G is the gain function, L is the left child node, R is the right child node, G L is the first-order derivative of the left child node, G R is the first-order derivative of the right child node, H L is the second-order derivative of the left child node, H R is the second-order derivative of the right child node, γ is the leaf number regularization parameter, and λ is the leaf weight regularization parameter.
[0061] S104: Based on the gain function, the transformer load data is used as the monitoring target, and the transformer load current waveform is used as the feature input for modeling. The final transformer load monitoring model is constructed by gradually adding the latest tree, and the model parameters are adaptively updated.
[0062] Exemplarily, transformer load data is used as the monitoring target, and transformer load current waveform is used as the feature input for modeling to determine the initial transformer load monitoring model; based on the gain function, the optimal splitting point is selected according to the first-order derivative and the second-order derivative when the objective function is minimized, and the left child node and the right child node are determined according to the optimal splitting point; the latest tree is generated through the left child node and the right child node; the latest tree is fitted to the initial transformer load monitoring model, and the transformer load monitoring model of the current round is determined by gradually adding the latest tree; monitoring is performed through the transformer load monitoring model of the current round, the monitoring value of the current round is obtained, and the residual and gradient are updated; the above steps are repeated until the preset number of trees is reached to determine the final transformer load monitoring model.
[0063] S105: When the transformer is connected to the grid and in operation, the transformer load current waveform measured by the current sensor is input into a final transformer load monitoring model to obtain a transformer load result.
[0064] For example, when the transformer is connected to the grid and operated, the current waveform measured by the current sensor is fed back to the final transformer load monitoring model to predict the transformer load result. The transformer load is then intelligently optimized and managed in real time based on the transformer load result, including adjusting load distribution, preventing overload or underload, and other measures.
[0065] It can be seen from the above description that the present invention defines the objective function according to the transformer load current waveform and the transformer load data, determines the gain function through the first-order derivative and the second-order derivative of the objective function, and can guide the update direction of the model parameters through the gain function to ensure that the model gradually approaches the optimal solution during the training process. Furthermore, based on the gain function, with the load data as the monitoring target and the load current waveform as the feature input, the complexity of the final transformer load monitoring model is gradually enhanced by gradually adding the latest tree, thereby improving the stability and reliability of the model. According to the final transformer load monitoring model, the future load trend of the transformer is accurately predicted to obtain the transformer load result, and potential load anomalies are discovered in time, thereby improving the safety and efficiency of power grid operation.
[0066] like Figure 2 As shown, Figure 2 A flowchart of another embodiment of a transformer load monitoring method based on deep learning provided by the present invention. A transformer load monitoring method based on deep learning, the method comprising:
[0067] S201: Simulate different load states of the transformer to obtain a transformer load current waveform.
[0068] S202: Obtain transformer current data in the transformer load current waveform through a current sensor.
[0069] Exemplarily, a transformer load model is built in PSCAD, a large number of different load states (underload, overload, rated state) are simulated, the transformer load current waveform is obtained, and the transformer current data in the transformer load current waveform is obtained through a current sensor.
[0070] S203: Obtain transformer load data according to the transformer current data, and pre-process the transformer load data.
[0071] For example, an advanced current sensor is designed to ensure that it can accurately capture the current data of the transformer for subsequent monitoring and analysis. The current sensor adopts a design that combines a non-contact Hall effect sensor with optical fiber sensing technology. The specific design is as follows:
[0072] (1) Using high-sensitivity CMOS Hall sensors, which can provide good output signals at low currents.
[0073] (2) Use a multi-stage sawtooth ring-shaped superimposed magnetic core to increase the coupling with the current, making the magnetic field distribution more uniform and improving the sensitivity.
[0074] (3) Integrated fiber optic sensor, based on the Brillouin scattering principle, uses optical fiber to measure the temperature change caused by current, achieving high-precision current sensing.
[0075] S204: defining an objective function according to the transformer load current waveform and the transformer load data.
[0076] Exemplarily, the objective function is determined according to the following formula:
[0077]
[0078] Among them, Obj is the objective function, y i is the true value, is the monitoring value of the transformer load data corresponding to the i-th transformer load current waveform, l is the loss function, Ω(f k ) is the regularization term, and K is the number of trees.
[0079] The regularization term is determined according to the following formula:
[0080]
[0081] Among them, ω j is the jth leaf node score, γ is the leaf number regularization parameter, λ is the leaf weight regularization parameter, and j is a leaf node.
[0082] S205: Determine a gain function through the first-order derivative and the second-order derivative of the objective function.
[0083] Exemplarily, the gain function is determined according to the following formula:
[0084]
[0085] Among them, G is the gain function, L is the left child node, R is the right child node, G L is the first-order derivative of the left child node, G R is the first-order derivative of the right child node, H L is the second-order derivative of the left child node, H R is the second-order derivative of the right child node, γ is the leaf number regularization parameter, and λ is the leaf weight regularization parameter.
[0086] S206: Using transformer load data as a monitoring target and using the transformer load current waveform as a feature input for modeling, and determining an initial transformer load monitoring model.
[0087] Exemplarily, the initial transformer load monitoring model is determined according to the following formula:
[0088]
[0089] in, is the monitoring value corresponding to the i-th input feature, x is the input feature, i.e., the transformer load current waveform, f kis the Kth tree.
[0090] S207: Based on the gain function, select the optimal splitting point according to the first-order derivative and the second-order derivative when the objective function is minimized, and determine the left child node and the right child node according to the optimal splitting point.
[0091] S208: Generate the latest tree through the left child node and the right child node.
[0092] S209: Fit the latest tree to the initial transformer load monitoring model, and determine the transformer load monitoring model of the current round by gradually adding the latest tree.
[0093] S210: Monitor through the current round of transformer load monitoring model, obtain the current round of monitoring values, and update the residual and gradient.
[0094] S211: Repeat steps S206-S210 until a preset number of trees is reached to determine a final transformer load monitoring model.
[0095] Exemplarily, the final transformer load monitoring model is determined according to the following formula:
[0096]
[0097] in, is the final monitoring value of the final transformer load monitoring model, x new is the input feature to be predicted, N is the preset number of trees, f k is the Kth tree.
[0098] S212: Adopting an adaptive learning rate method to update the model parameters of the final transformer load monitoring model.
[0099] For example, the adaptive learning rate method can dynamically adjust the learning rate according to the changes in data to maintain the stability and convergence speed of the model. The entire adaptive process is as follows:
[0100]
[0101] Among them, η is the learning rate.
[0102] S213: When the transformer is connected to the grid and in operation, the transformer load current waveform measured by the current sensor is input into the final transformer load monitoring model to obtain the transformer load result.
[0103] It should be noted that step S213 Figure 1 This has been discussed in detail in the implementation scenario shown and will not be repeated here.
[0104] From the above description, it can be seen that the present invention also enhances the generalization ability of the model by adaptively updating model parameters, so that it can adapt to different operating environments and load conditions, thereby improving the model's ability to predict load changes.
[0105] like Figure 3 As shown, Figure 3 A schematic diagram of a transformer load monitoring system based on deep learning according to an embodiment of the present invention is provided. A transformer load monitoring system based on deep learning 10, the system comprises:
[0106] The acquisition module 11 is used to acquire the transformer load current waveform and transformer load data in real time.
[0107] The objective function definition module 12 is used to define the objective function according to the transformer load current waveform and the transformer load data.
[0108] The gain function determination module 13 is used to determine the gain function through the first-order derivative and the second-order derivative of the objective function.
[0109] The final transformer load monitoring model determination module 14 is used to build the final transformer load monitoring model by gradually adding the latest tree based on the gain function, taking the transformer load as the monitoring target and the transformer load current waveform as the feature input for modeling.
[0110] The transformer load monitoring module 15 is used to input the transformer load current waveform measured by the current sensor into the final transformer load monitoring model to obtain the transformer load result when the transformer is connected to the grid and in operation.
[0111] Exemplarily, in the acquisition module 11, different load states of the transformer are simulated to obtain the transformer load current waveform; the transformer current data in the transformer load current waveform is obtained by the current sensor; the transformer load data is obtained according to the transformer current data, and the transformer load data is preprocessed. In the objective function definition module 12, the objective function is defined according to the transformer load current waveform and the transformer load data. In the gain function determination module 13, the gain function is determined by the first-order derivative and the second-order derivative of the objective function. In the final transformer load monitoring model determination module 14, the transformer load data is used as the monitoring target, and the transformer load current waveform is used as the feature input for modeling to determine the initial transformer load monitoring model; based on the gain function, the optimal split point is selected according to the first-order derivative and the second-order derivative when the objective function is minimized, and the left child node and the right child node are determined according to the optimal split point; the latest tree is generated through the left child node and the right child node; the latest tree is fitted to the initial transformer load monitoring model, and the transformer load monitoring model of the current round is determined by gradually adding the latest tree; the transformer load monitoring model of the current round is used for monitoring, the monitoring value of the current round is obtained, and the residual and gradient are updated; the above steps are repeated until the preset number of trees is reached to determine the final transformer load monitoring model. In the transformer load monitoring module 15, when the transformer is connected to the grid, the transformer load current waveform measured by the current sensor is input into the final transformer load monitoring model to obtain the transformer load result.
[0112] like Figure 4 As shown, Figure 4 The device 20 includes a memory 21 and a processor 22. The memory 21 stores a computer program, and the processor 22 executes the computer program when working to implement the following. Figure 1 and Figure 2 The method shown.
[0113] The specific technical details of a deep learning-based transformer load monitoring method implemented when the above-mentioned device 20 executes a computer program have been discussed in detail in the aforementioned method steps, so they will not be repeated here.
[0114] like Figure 5 As shown, Figure 5 The structure diagram of an embodiment of the medium provided by the present invention is shown in FIG. The medium 30 stores at least one computer program 31, and the computer program 31 is executed by the processor 22 to implement the following Figure 1 and Figure 2 In one embodiment, the medium 30 may be a storage chip, a hard disk, a mobile hard disk, a USB flash drive, an optical disk, or other readable and writable storage tools, or a server, etc.
[0115] The above describes specific embodiments of the present specification, and other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily have to be performed in the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0116] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer-readable storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.
[0117] The apparatus, device, non-volatile computer-readable storage medium and method provided in the embodiments of this specification correspond to each other, and therefore, the apparatus, device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the corresponding apparatus, device, and non-volatile computer storage medium will not be repeated here.
[0118] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
[0119] For the convenience of description, the above device is described by being divided into various units according to their functions and described separately. Of course, when implementing this specification, the functions of each unit can be implemented in one or more software and / or hardware. It should be understood by those skilled in the art that this specification embodiment can be provided as a method, system, or computer program product. Therefore, this specification embodiment can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification embodiment can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0120] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0121] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0122] The above disclosure is only the preferred embodiment of the present invention, which certainly cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.
Claims
1. A transformer load monitoring method based on deep learning, characterized in that: The method comprises: Real-time collection of transformer load current waveform and transformer load data; Defining an objective function according to the transformer load current waveform and transformer load data; Determining a gain function by first-order derivative and second-order derivative of the objective function; Based on the gain function, the transformer load data is used as the monitoring target, the transformer load current waveform is used as the feature input for modeling, and the final transformer load monitoring model is constructed by gradually adding the latest tree; When the transformer is connected to the grid and in operation, the transformer load current waveform measured by the current sensor is input into the final transformer load monitoring model to obtain the transformer load result.
2. A transformer load monitoring method based on deep learning according to claim 1, characterized in that: The real-time collection of transformer load current waveform and transformer load data specifically includes: Simulate different load states of the transformer to obtain the transformer load current waveform; Acquiring transformer current data in the transformer load current waveform through a current sensor; Transformer load data is acquired according to the transformer current data, and the transformer load data is preprocessed.
3. A transformer load monitoring method based on deep learning according to claim 2, characterized in that: Defining the objective function according to the transformer load current waveform and transformer load data specifically includes: according to Define the objective function, where Obj is the objective function, y i is the true value, is the monitoring value of the transformer load data corresponding to the i-th transformer load current waveform, l is the loss function, Ω(f k ) is the regularization term, and K is the number of trees.
4. A transformer load monitoring method based on deep learning according to claim 3, characterized in that: The determining of the gain function by the first-order derivative and the second-order derivative of the objective function specifically includes: according to Determine the gain function, where G is the gain function, L is the left child node, R is the right child node, G L is the first-order derivative of the left child node, G R is the first-order derivative of the right child node, H L is the second-order derivative of the left child node, H R is the second-order derivative of the right child node, γ is the leaf number regularization parameter, and λ is the leaf weight regularization parameter.
5. A transformer load monitoring method based on deep learning according to claim 2, characterized in that: Based on the gain function, the transformer load data is used as the monitoring target, the transformer load current waveform is used as the feature input for modeling, and the final transformer load monitoring model is constructed by gradually adding the latest tree, specifically including: Taking the transformer load data as the monitoring target and the transformer load current waveform as the characteristic input for modeling, an initial transformer load monitoring model is determined; Based on the gain function, selecting an optimal splitting point according to the first-order derivative and the second-order derivative when the objective function is minimized, and determining a left child node and a right child node according to the optimal splitting point; Generate a latest tree through the left child node and the right child node; Fitting the latest tree to the initial transformer load monitoring model, and determining the transformer load monitoring model of the current round by gradually adding the latest tree; Monitor through the transformer load monitoring model of the current round, obtain the monitoring value of the current round, and update the residual and gradient; Repeat the above steps until the preset number of trees is reached to determine the final transformer load monitoring model.
6. A transformer load monitoring method based on deep learning according to claim 5, characterized in that: The method of taking the transformer load as the monitoring target and taking the transformer load current waveform as the characteristic input for modeling to determine the initial transformer load monitoring model specifically includes: according to Determine the initial transformer load monitoring model, where: is the monitoring value corresponding to the i-th input feature, x is the input feature, i.e., the transformer load current waveform, f k is the Kth tree.
7. The transformer load monitoring method based on deep learning according to claim 5 is characterized in that: Repeating the above steps until the preset number of trees is reached to determine the final transformer load monitoring model specifically includes: according to Determine the final transformer load monitoring model, where: is the final monitoring value of the final transformer load monitoring model, x new is the input feature to be predicted, N is the preset number of trees, f k is the Kth tree.
8. The transformer load monitoring method based on deep learning according to claim 1 is characterized in that: Based on the gain function, taking the transformer load as the monitoring target and taking the transformer load current waveform as the feature input for modeling, and constructing the final transformer load monitoring model by gradually adding the latest tree, the method further specifically includes: The model parameters of the final transformer load monitoring model are updated using an adaptive learning rate method.
9. A transformer load monitoring system based on deep learning, characterized in that: The system comprises: An acquisition module, used for real-time acquisition of transformer load current waveform and transformer load data; An objective function definition module, used for defining an objective function according to a transformer load current waveform and transformer load data; A gain function determination module, used to determine the gain function through the first-order derivative and the second-order derivative of the objective function; A final transformer load monitoring model determination module is used to construct a final transformer load monitoring model by gradually adding the latest tree based on the gain function, taking the transformer load data as the monitoring target and the transformer load current waveform as the feature input for modeling; The transformer load monitoring module is used to input the transformer load current waveform measured by the current sensor into the final transformer load monitoring model when the transformer is connected to the grid and operate, so as to obtain the transformer load result.
10. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 8.
11. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 8.