Transformer operation parameter prediction method and device, electronic equipment and storage medium

Abstract

The application discloses a transformer operation parameter prediction method and device, electronic equipment and a storage medium. According to the technical scheme, at least one current operation parameter group of a target transformer is determined, and a sequence of operation parameter groups is determined according to the at least one current operation parameter group. The current operation parameter group contains at least one operation parameter, the sequence of operation parameters is sorted according to the time sequence of the corresponding collection time of the current operation parameter group, and the sequence of operation parameter groups is input into a transformer operation parameter prediction model constructed in advance. At least one predicted operation parameter group of the target transformer is obtained through the transformer operation parameter prediction model, thereby realizing prediction of multiple operation parameters of the target transformer, improving the prediction efficiency, and realizing accurate prediction of the multiple operation parameters of the target transformer by means of a dynamic specific layer and a dynamic shared layer in the transformer operation parameter prediction model, and improving the prediction accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data prediction, and in particular to a transformer operating parameter prediction method and device, electronic equipment and a storage medium. BACKGROUND

[0002] In power transmission, transformers are a common and important electrical device that can raise or lower input AC voltage to adapt to different power needs.

[0003] During the operation of the transformer, there are various types of transformer faults, including but not limited to insulation aging, overheating, electrical breakdown, partial discharge, mechanical damage, and cooling system failure, etc. The occurrence of these faults is often closely related to the operating parameters of the device itself. How to accurately predict the transformer parameters is important for the maintenance of the transformer. SUMMARY

[0004] The present application provides a transformer operating parameter prediction method, device, electronic equipment and storage medium to solve the problem of accurately predicting multiple operating parameters of the transformer.

[0005] According to an aspect of the present application, a transformer operating parameter prediction method is provided, which comprises:

[0006] determining at least one current operating parameter group of the target transformer, and determining an operating parameter group sequence according to the at least one current operating parameter group, wherein the current operating parameter group contains at least one operating parameter, and the operating parameter sequence is sorted according to the time sequence of the collection time corresponding to the current operating parameter group;

[0007] inputting the operating parameter group sequence into a pre-constructed transformer operating parameter prediction model, and obtaining at least one predicted operating parameter group of the target transformer through the transformer operating parameter prediction model, wherein the number of predicted operating parameter groups in the at least one predicted operating parameter group is less than the number of current operating parameter groups in the at least one current operating parameter group, the time interval of the collection time of each current operating parameter group in the at least one current operating parameter group is a preset time interval, the time interval of the prediction time of each predicted operating parameter group in the at least one predicted operating parameter group is a preset time interval, and the time interval between the prediction time of the first predicted operating parameter group in the at least one predicted operating parameter group and the collection time of the first current operating parameter group in the at least one current operating parameter group is a preset time interval; the first predicted operating parameter group is the predicted operating parameter group with the earliest prediction time in the at least one predicted operating parameter group, and the first current operating parameter group is the current operating parameter group with the latest collection time in the at least one current operating parameter group;

[0008] The transformer operation parameter prediction model comprises a dynamic shared layer and a dynamic specific layer, the dynamic shared layer is used for performing orthogonal transformation on the operation parameter group sequence to generate a dynamic shared space, and the dynamic specific layer is used for performing random nonlinear mapping on the dynamic shared space to generate a dynamic specific space. The transformer operation parameter prediction model determines the common variation trend of each operation parameter in the operation parameter group through the dynamic shared layer, and determines the specific variation trend of each operation parameter in the operation parameter group according to the dynamic specific layer, so as to realize the prediction of the corresponding operation parameter group of the target transformer.

[0009] According to another aspect of the present application, a transformer operation parameter prediction device is provided, which comprises:

[0010] A current parameter determination module is configured to determine at least one current operation parameter group of a target transformer, and determine an operation parameter group sequence according to the at least one current operation parameter group. The current operation parameter group comprises at least one operation parameter, and the operation parameter sequence is sorted according to the time sequence of the corresponding acquisition time of the current operation parameter group.

[0011] A parameter prediction module is configured to input the operation parameter group sequence into a pre-constructed transformer operation parameter prediction model, and obtain at least one predicted operation parameter group of the target transformer through the transformer operation parameter prediction model. The number of predicted operation parameter groups in the at least one predicted operation parameter group is less than the number of current operation parameter groups in the at least one current operation parameter group. The time interval of the acquisition time of each current operation parameter group in the at least one current operation parameter group is a preset time interval. The time interval of the prediction time of each predicted operation parameter group in the at least one predicted operation parameter group is a preset time interval. The time interval between the prediction time of the first predicted operation parameter group in the at least one predicted operation parameter group and the acquisition time of the first current operation parameter group in the at least one current operation parameter group is a preset time interval. The first predicted operation parameter group is the predicted operation parameter group with the earliest prediction time in the at least one predicted operation parameter group, and the first current operation parameter group is the current operation parameter group with the latest acquisition time in the at least one current operation parameter group.

[0012] The transformer operation parameter prediction model comprises a dynamic shared layer and a dynamic specific layer, the dynamic shared layer is used for performing orthogonal transformation on the operation parameter group sequence to generate a dynamic shared space, and the dynamic specific layer is used for performing random nonlinear mapping on the dynamic shared space to generate a dynamic specific space. The transformer operation parameter prediction model determines the common variation trend of each operation parameter in the operation parameter group through the dynamic shared layer, and determines the specific variation trend of each operation parameter in the operation parameter group according to the dynamic specific layer, so as to realize the prediction of the corresponding operation parameter group of the target transformer.

[0013] According to another aspect of the present application, an electronic device is provided, which comprises:

[0014] at least one processor; and

[0015] a memory communicatively connected with the at least one processor; wherein,

[0016] the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the transformer operating parameter prediction method of any one of the embodiments of the present application.

[0017] According to another aspect of the present application, a computer readable storage medium is provided, which stores computer instructions for enabling a processor to implement the transformer operating parameter prediction method of any one of the embodiments of the present application when executed by the processor.

[0018] The technical solution of the embodiments of the present application determines at least one current operating parameter group of the target transformer, and determines an operating parameter group sequence according to the at least one current operating parameter group, wherein the current operating parameter group contains at least one operating parameter, the operating parameter sequence is sorted according to the time sequence of the corresponding collection time of the current operating parameter group, and the operating parameter group sequence is input into a pre-constructed transformer operating parameter prediction model, so as to obtain at least one predicted operating parameter group of the target transformer through the transformer operating parameter prediction model, thereby realizing the prediction of multiple operating parameters of the target transformer, improving the prediction efficiency, and realizing the accurate prediction of multiple operating parameters of the target transformer by means of the dynamic specific layer and the dynamic shared layer in the transformer operating parameter prediction model, and improving the prediction accuracy.

[0019] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0021] Figure 1 is a flowchart of a transformer operating parameter prediction method according to the first embodiment of the present application;

[0022] Figure 2 is a flowchart of another transformer operating parameter prediction method according to the second embodiment of the present application;

[0023] Figure 3 is a structural schematic diagram of a transformer operating parameter prediction device according to an embodiment three of the present application;

[0024] Figure 4 is a structural schematic diagram of an electronic device for implementing a transformer operating parameter prediction method of the present application. DETAILED DESCRIPTION

[0025] In order to enable persons skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by persons skilled in the art without creative labor should fall within the scope of protection of the present application.

[0026] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units does not have to be limited to only those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] Embodiment one

[0028] Figure 1 A flowchart of a transformer operating parameter prediction method is provided for the first embodiment of the present application. The present embodiment can be applicable to the prediction of multiple operating parameters of a target transformer. The method can be performed by a transformer operating parameter prediction device, which can be implemented in the form of hardware and / or software, and can be configured in an electronic device having data processing capability. As shown in the figure, the method comprises: Figure 1

[0029] S110, at least one current operating parameter group of the target transformer is determined, and a sequence of operating parameter groups is determined according to the at least one current operating parameter group. The current operating parameter group contains at least one operating parameter, and the sequence of operating parameters is sorted according to the time sequence of the corresponding collection time of the current operating parameter group.

[0030] ​The current operating parameter group is composed of at least one current operating parameter, and each current operating parameter group is composed of all current operating parameters collected at a collection time after at least one current operating parameter is collected.

[0031] The operating parameter group includes winding temperature, dissolved gas H2 in oil, dissolved gas C2H2 in oil, dissolved gas CO in oil, dissolved gas CO2 in oil, dissolved gas C2H4 in oil, dissolved gas CH4 in oil, oil temperature, oil pressure, ultrasonic partial discharge, high-frequency partial discharge, core grounding current, clamp grounding current, and transformer operating voiceprint, and the present application is not limited in this regard.

[0032] In the acquisition of the at least one current operating parameter group of the target transformer, the first preset number of current operating parameter groups collected at the current collection time are acquired as the at least one current operating parameter group of the target transformer, and each current operating parameter group is sorted according to the time sequence of the corresponding collection time, and the running parameter group sequence is generated according to the sorting result.

[0033] For example, at 2 o'clock, the current operating parameter group is collected every minute, and the preset number is 50. At this time, the current operating parameter group collected at 1 minute 59 seconds, the current operating parameter group collected at 1 minute 58 seconds, and the current operating parameter group collected at 1 minute 10 seconds are acquired as the at least one current operating parameter group.

[0034] S120, input the running parameter group sequence into the transformer operating parameter prediction model constructed in advance, and obtain at least one predicted operating parameter group of the target transformer through the transformer operating parameter prediction model, the number of predicted operating parameter groups in the at least one predicted operating parameter group is less than the number of current operating parameter groups in the at least one current operating parameter group, the time interval of the collection time of each current operating parameter group in the at least one current operating parameter group is a preset time interval, the time interval of the prediction time of each predicted operating parameter group in the at least one predicted operating parameter group is a preset time interval, and the time interval between the prediction time of the first predicted operating parameter group in the at least one predicted operating parameter group and the collection time of the first current operating parameter group in the at least one current operating parameter group is a preset time interval; the first predicted operating parameter group is the predicted operating parameter group with the earliest prediction time in the at least one predicted operating parameter group, and the first current operating parameter group is the current operating parameter group with the latest collection time in the at least one current operating parameter group.

[0035] The transformer operation parameter prediction model comprises a dynamic shared layer and a dynamic specific layer. The dynamic shared layer is configured to perform orthogonal transformation on the operation parameter group sequence to generate a dynamic shared space. The dynamic specific layer is configured to perform random nonlinear mapping on the dynamic shared space to generate a dynamic specific space. The transformer operation parameter prediction model determines the common variation trend of each operation parameter in the operation parameter group through the dynamic shared layer, and determines the specific variation trend of each operation parameter in the operation parameter group according to the dynamic specific layer, thereby realizing prediction of the corresponding operation parameter group of the target transformer.

[0036] The predicted operation parameter group can be an operation parameter group that will be collected at a subsequent plurality of collection times of the target transformer through the operation parameter group sequence.

[0037] Since the pre-constructed transformer operation parameter prediction model comprises a dynamic shared layer and a dynamic specific layer, the common variation trend of each operation parameter in the operation parameter group can be determined through the dynamic shared layer, and the specific variation trend of each operation parameter in the operation parameter group can be determined according to the dynamic specific layer, thereby realizing prediction of the corresponding operation parameter group of the target transformer.

[0038] Therefore, the operation parameter group sequence will be input into the pre-constructed transformer operation parameter prediction model, and the input operation parameter group sequence will be processed by the transformer operation parameter prediction model, thereby obtaining the predicted operation parameter group corresponding to each collection time in at least one collection time after the current collection time.

[0039] In order to ensure the accuracy of the prediction result and the stability of the overall system, too many current operation parameter groups will not be predicted. Therefore, the number of predicted operation parameter groups in at least one predicted operation parameter group is required to be less than the number of current operation parameter groups in at least one current operation parameter group.

[0040] For example, after obtaining 50 current operation parameter groups, the corresponding operation parameter group sequence will be determined according to the 50 current operation parameter groups, and the operation parameter group sequence will be input into the transformer operation parameter prediction model. The transformer operation parameter prediction model will predict the operation parameter groups that will be collected in the future 20 collection times, thereby obtaining the predicted operation parameter group corresponding to 2:01, the predicted operation parameter group corresponding to 2:02, and the predicted operation parameter group corresponding to 2:20.

[0041] By inputting the sequence of operating parameter groups into the pre-constructed transformer operating parameter prediction model, at least one predicted operating parameter group of the target transformer is obtained through the transformer operating parameter prediction model. The transformer operating parameter prediction model can calculate the characteristics of a large amount of data with fast computing efficiency, realize fast prediction of the at least one predicted operating parameter group, and realize accurate prediction of the at least one predicted operating parameter group by means of the dynamic shared layer and the dynamic specific layer of the transformer operating parameter prediction model.

[0042] In an alternative, the training process of the transformer operating parameter prediction model can include steps A1-A6:

[0043] Step A1, obtaining a training sample data group of a target transformer, the training sample data group containing at least one training sample operating parameter group of the target transformer and at least one first training sample operating parameter group corresponding to the at least one training sample operating parameter group, the first training sample operating parameter group being an actual training sample operating parameter group corresponding to the training sample operating parameter group.

[0044] Step A2, determining a sequence of training sample operating parameter groups according to the at least one training sample operating parameter group.

[0045] Step A3, performing orthogonal transformation on the sequence of training sample operating parameter groups to determine a sample dynamic shared space.

[0046] Step A4, performing random nonlinear mapping on the dynamic shared space to obtain a sample dynamic specific space.

[0047] Step A5, determining at least one second training sample operating parameter group of the target transformer according to the pre-set initial shared weight of the dynamic shared space, the pre-set initial specific weight of the dynamic specific space, the sample dynamic shared space, and the sample dynamic specific space, the second training sample operating parameter group being a predicted training sample operating parameter group corresponding to the training sample operating parameter group.

[0048] Step A6, adjusting the pre-set initial shared weight of the dynamic shared space and the pre-set initial specific weight of the dynamic specific space according to the at least one second training sample operating parameter group and the at least one first training sample operating parameter group of the sample transformer, obtaining again the training sample data group of the target transformer, and adjusting again the initial specific weight and the initial shared weight until the adjustment times reach a preset number of times to obtain the transformer operating parameter prediction model.

[0049] When training the transformer operating parameter prediction model, the sample data required for training needs to be obtained first. For this purpose, the training sample data group of the target transformer is obtained.

[0050] At least one training sample operation parameter group containing the target transformer and at least one first training sample operation parameter group corresponding to the at least one training sample operation parameter group are included in the training sample data set. The first training sample operation parameter group can be an operation parameter group collected at each collection time after the collection time corresponding to the training sample operation parameter.

[0051] For example, the collection time of the training sample operation parameter group with the latest collection time in the at least one training sample operation parameter group is 2:10, and the collection time of the first training sample operation parameter group with the earliest collection time in the corresponding at least one first training sample operation parameter group is 2:11.

[0052] Before inputting into the initial transformer operation parameter prediction model, in order to ensure that the initial transformer operation parameter prediction model can accurately understand the meaning of the input data, a training sample operation parameter group sequence that the initial transformer operation parameter prediction model can accurately understand is generated according to the at least one training sample operation parameter group.

[0053] In the initial transformer operation parameter prediction model, the input training sample operation parameter group sequence is subjected to orthogonal transformation in the dynamic sharing layer, so as to obtain a sample dynamic sharing space. And random nonlinear mapping is performed in the dynamic specific layer, so as to obtain a sample dynamic specific space.

[0054] After obtaining the sample dynamic specific space and the sample dynamic sharing space, at least one second training sample operation parameter group of the target transformer is determined according to the pre-set initial sharing weight of the dynamic sharing space, the pre-set initial specific weight of the dynamic specific space, the sample dynamic sharing space, and the sample dynamic specific space. The second training sample operation parameter group is an operation parameter group obtained after prediction according to the training sample data set.

[0055] According to the at least one first training sample operation parameter group of the sample transformer and the at least one second training sample operation parameter group, the initial specific weight and the initial sharing weight in the initial transformer operation parameter prediction model can be adjusted.

[0056] The above steps are repeated until the number of repetitions reaches a preset number, so as to obtain a transformer operation parameter prediction model.

[0057] In an optional solution, adjusting the pre-set initial sharing weight of the dynamic sharing space and the pre-set initial specific weight of the dynamic specific space can include steps B1-B2:

[0058] Step B1, determining a weight objective function of the initial specific weight.

[0059] Step B2, Crank-Nicolson type parameter solving optimization algorithm based on alternating direction multiplier method, solves the weight objective function to obtain the adjusted initial specific weight.

[0060] wherein the calculation of the transformer operating parameter prediction model on the at least one predicted operating parameter group can be represented by the following formula:

[0061] Y = XΘW m + ξ(XW e )W n

[0062] wherein Y∈R s×r is the predicted operating parameter group, X∈R s×r is the operating parameter group sequence of the target transformer; r is the output input dimension, i.e., the number of operating parameters; s is the sampling number; W n ∈R t×r , W n ∈R d×r are the weights of the dynamic shared layer and the dynamic specific layer, respectively; t is the dimension of the dynamic shared subspace; d is the dimension of the dynamic specific subspace; Θ∈R D×t is an orthogonal decoupling matrix; W e ∈R r×d is a random matrix, and the numerical value range is [-1, 1]; ξ is a nonlinear activation function, and ξ(XW e ) is the dynamic specific space.

[0063] A Lagrange multiplier factor is introduced, and an objective function for training the network until convergence is defined:

[0064]

[0065] wherein Δ is the Lagrange multiplier factor; wherein α, ρ and β are balance parameters; E = Y - XΘW m - ξ(XW e )W n is a residual term; is the F norm

[0066] Since the above objective function is a non-stationary objective function, it needs to be transformed to obtain the weight objective function, as shown in the following formula:

[0067]

[0068] wherein, ε is a set tolerance threshold, ε > 0, and wi is the i-th row of Wn.

[0069] Then, the weight target function is solved by using a Crank-Nicolson type parameter solving optimization algorithm based on an alternating direction multiplier method to obtain the adjusted initial specific weight.

[0070] Optionally, the process of solving the weight target function by using the Crank-Nicolson type parameter solving optimization algorithm based on the alternating direction multiplier method comprises:

[0071] (1) fixing W n , Θ, E and other parameters, updating the dynamic shared layer weight coefficient Wm

[0072] First, the derivative of the target function with respect to Wm is calculated:

[0073]

[0074] Approaching it to 0, the optimization solution of Wm is

[0075]

[0076] (2) fixing W m , Θ, E and other parameters, updating the dynamic specific layer weight coefficient Wn

[0077] First, the derivative of the target function with respect to Wn is calculated:

[0078]

[0079] Approaching it to 0, the optimization solution of Wn is

[0080]

[0081] (3) fixing W m , Θ, W n and other parameters, updating the residual term E

[0082] First, the derivative of the target function with respect to E is calculated:

[0083]

[0084] Approaching it to 0, the optimization solution of E is E k+1 :

[0085]

[0086] Fixing W m , E, W n and other parameters, updating the orthogonal decoupling matrix Θ

[0087] The sub-target function for solving the orthogonal decoupling matrix is:

[0088]

[0089] Since solving this function can be regarded as an orthogonal constraint problem, it is also called Stiefel manifold optimization problem. The derivative of the objective function with respect to Θ is calculated:

[0090]

[0091] By setting the skew-symmetric matrix A, the calculated Θ optimization solution Θ k+1 :

[0092] Δ k+1 =Δ k +ρ(Y-XΘW m -ξ(XW e )W n -E);

[0093] where A=GΘ T -ΘG T ; I is the unit matrix; τ is the accelerated gradient parameter. In addition, τ is updated in an inner loop for Stiefel manifold optimization solution.

[0094] Update the Lagrange multiplier factor Δ. Update the solution by the mode of double ascent iteration:

[0095] Δ k+1 =Δ k +ρ(Y-XΘW m -ξ(XW e )W n -E);

[0096] The final model parameters are solved by iteratively updating (1)-(5) to output the final weight coefficient

[0097] In an optional solution, adjusting the initial shared weight of the pre-set dynamic shared space and the initial specific weight of the pre-set dynamic specific space according to at least one second training sample running parameter group and at least one first training sample running parameter group of the sample transformer can include steps C1-C2:

[0098] Step C1, constraint the initial specific weight based on the row sparsity constraint to obtain the constrained initial specific weight.

[0099] Step C2, adjust the constrained initial specific weight of the pre-set dynamic specific space according to at least one predicted training sample running parameter group and at least one first training sample running parameter group of the sample transformer.

[0100] Since the high-dimensional random nonlinear features obtained in the dynamic specific layer can be redundant and various strategies of row sparsity constraint feature selection show effectiveness in potential shared feature extraction, we introduce row sparsity constraint on the initial specific weight of the dynamic specific layer.

[0101] And subsequently, by running at least one predicted training sample operation parameter group and at least one first training sample operation parameter group of the sample transformer, the constraint initial specific weight of the pre-set dynamic specific space is adjusted.

[0102] Optionally, the initial specific weight is constrained based on the row sparsity constraint to obtain the initial specific weight, including:

[0103] The initial specific weight is constrained based on the L1 norm and the L2 norm to obtain the initial specific weight.

[0104] After the initial specific weight is constrained based on the L1 norm and the L2 norm, the following expression can be obtained:

[0105]

[0106] d is the dimension of the dynamic specific subspace; W n represents the initial specific weight.

[0107] According to the technical scheme of the embodiment of the present application, by determining at least one current operation parameter group of the target transformer, and determining the operation parameter group sequence according to the at least one current operation parameter group, the current operation parameter group contains at least one operation parameter, the operation parameter sequence is sorted according to the time sequence of the corresponding collection time of the current operation parameter group, and the operation parameter group sequence is input into the transformer operation parameter prediction model constructed in advance, at least one predicted operation parameter group of the target transformer is obtained through the transformer operation parameter prediction model, thereby realizing the prediction of multiple operation parameters of the target transformer, improving the prediction efficiency, and realizing the accurate prediction of multiple operation parameters of the target transformer by means of the dynamic specific layer and the dynamic shared layer in the transformer operation parameter prediction model, and improving the prediction accuracy.

[0108] Embodiment two

[0109] Figure 2 Another flowchart of the transformer operation parameter prediction method is provided for the embodiment of the present application, and the process of obtaining at least one predicted operation parameter group of the target transformer through the transformer operation parameter prediction model in the foregoing embodiment is further optimized based on the foregoing embodiment, and the embodiment can be combined with each optional scheme in one or more of the foregoing embodiments. As shown in the figure, the transformer operation parameter prediction method of the embodiment can include the following steps: Figure 2 ​

[0110] S210, determine at least one current operation parameter group of the target transformer, and determine an operation parameter group sequence according to the at least one current operation parameter group, the current operation parameter group containing at least one operation parameter, and the operation parameter sequence being sorted according to time sequence of corresponding collection time of the current operation parameter group.

[0111] S220, perform orthogonal transformation on the operation parameter group sequence to generate a current dynamic shared space.

[0112] S230, perform random nonlinear mapping on the dynamic shared space to generate a current dynamic specific space.

[0113] S240, determine at least one predicted operation parameter group of the target transformer according to the current dynamic shared space, the current dynamic specific space, a predetermined shared weight of the current dynamic shared space, and a predetermined specific weight of the current dynamic specific space, the sum of the shared weight and the specific weight being 1.

[0114] After inputting the at least one current operation parameter group into the transformer operation parameter prediction model, orthogonal transformation is performed on a dynamic shared layer in the transformer operation parameter prediction model to generate a current dynamic shared space, and then random nonlinear mapping is performed on the dynamic shared space by a dynamic specific layer to generate a current dynamic specific space.

[0115] Then, at least one predicted operation parameter group of the target transformer is determined according to the current dynamic shared space, the current dynamic specific space, a predetermined shared weight of the current dynamic shared space, and a predetermined specific weight of the current dynamic specific space.

[0116] Wherein, the calculation of the transformer operation parameter prediction model on the at least one predicted operation parameter group can be represented by the following formula:

[0117] Y = X Θ W n + ξ (X W e ) W n

[0118] Wherein Y ∈ R s×r is the predicted operation parameter group, X ∈ R s×r is the operation parameter group sequence of the target transformer; r is the output input dimension, i.e., the number of operation parameters; s is the number of samples; W m ∈ R t×r , W n ∈ R d×r are the weights of the dynamic shared layer and the dynamic specific layer respectively; t is the dimension of the dynamic shared subspace; d is the dimension of the dynamic specific subspace; Θ ∈ R D×t is an orthogonal decoupling matrix; W e ∈ R r×dis a random matrix, the numerical value range is [-1, 1]; is a nonlinear activation function and is (XW e ) is a dynamic specific space.

[0119] According to the technical scheme of the embodiment of the application, the dynamic sharing characteristics of the multiple operating parameters of the target transformer can be determined by performing orthogonal transformation on the operating parameter group sequence to generate the current dynamic sharing space, and the dynamic specific characteristics of the multiple operating parameters of the target transformer can be determined by performing random nonlinear mapping on the dynamic sharing space to generate the current dynamic specific space, the dynamic analysis of the multiple parameter sequences is realized, the complex relationship between the multivariate chaotic time sequences is effectively separated, and thus the accuracy of the final prediction result is ensured.

[0120] Embodiment three

[0121] Figure 3 A structural block diagram of a transformer operating parameter prediction device is provided in the embodiment of the application, and the embodiment can be applicable to the case of predicting multiple operating parameters of a target transformer. The transformer operating parameter prediction device can be realized in the form of hardware and / or software, and the transformer operating parameter prediction device can be configured in an electronic device with data processing capability. As shown in the figure, the transformer operating parameter prediction device of the embodiment can include a current parameter determination module 310 and a parameter prediction module 320. Among them: Figure 3

[0122] The current parameter determination module 310 is configured to determine at least one current operating parameter group of a target transformer, and determine an operating parameter group sequence according to the at least one current operating parameter group, the current operating parameter group containing at least one operating parameter, and the operating parameter sequence being sorted according to the time sequence of the corresponding collection time of the current operating parameter group;

[0123] The parameter prediction module 320 is configured to input the operating parameter group sequence into a pre-constructed transformer operating parameter prediction model, and obtain at least one predicted operating parameter group of the target transformer through the transformer operating parameter prediction model, the number of predicted operating parameter groups in the at least one predicted operating parameter group being less than the number of current operating parameter groups in the at least one current operating parameter group, the time interval of the collection time of each current operating parameter group in the at least one current operating parameter group being a preset time interval, the time interval of the prediction time of each predicted operating parameter group in the at least one predicted operating parameter group being the preset time interval, and the time interval between the prediction time of the first predicted operating parameter group in the at least one predicted operating parameter group and the collection time of the first current operating parameter group in the at least one current operating parameter group being the preset time interval; the first predicted operating parameter group being the predicted operating parameter group with the earliest prediction time in the at least one predicted operating parameter group, and the first current operating parameter group being the current operating parameter group with the latest collection time in the at least one current operating parameter group.​

[0124] The transformer operation parameter prediction model comprises a dynamic shared layer and a dynamic specific layer, the dynamic shared layer is used for performing orthogonal transformation on the operation parameter group sequence to generate a dynamic shared space, and the dynamic specific layer is used for performing random nonlinear mapping on the dynamic shared space to generate a dynamic specific space, the transformer operation parameter prediction model determines the common variation trend of each operation parameter in the operation parameter group through the dynamic shared layer, and determines the specific variation trend of each operation parameter in the operation parameter group according to the dynamic specific layer, so as to realize the prediction of the corresponding operation parameter group of the target transformer.

[0125] On the basis of the above-mentioned embodiments, the parameter prediction module 320 comprises:

[0126] performing orthogonal transformation on the operation parameter group sequence to generate a current dynamic shared space;

[0127] performing random nonlinear mapping on the dynamic shared space to generate a current dynamic specific space;

[0128] determining at least one predicted operation parameter group of the target transformer according to the current dynamic shared space, the current dynamic specific space, a pre-determined shared weight of the current dynamic shared space, and a pre-determined specific weight of the current dynamic specific space, and the sum of the shared weight and the specific weight is 1.

[0129] The transformer operation parameter prediction device provided in the embodiments of the present application can execute the transformer operation parameter prediction method provided in any of the embodiments of the present application, and has the corresponding function modules and beneficial effects of the execution method.

[0130] Embodiment four

[0131] Figure 4 A structural schematic diagram of an electronic device 10 that can be used to implement embodiments of the present application is shown. The electronic device is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices (e.g., headsets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit the implementations of the present application described and / or claimed in this document.

[0132] As Figure 4As shown, the electronic device 10 includes at least one processor 11, and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., communicatively connected to the at least one processor 11, where the memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or loaded into the random access memory (RAM) 13 from the storage unit 18. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0133] Various components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc., an output unit 17, such as various types of displays, a speaker, etc., a storage unit 18, such as a magnetic disk, an optical disk, etc., and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0134] The processor 11 can be various general and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 performs various methods and processes described above, such as the transformer operating parameter prediction method.

[0135] In some embodiments, the transformer operating parameter prediction method can be implemented as a computer program tangibly embodied in a computer readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the transformer operating parameter prediction method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the transformer operating parameter prediction method by any other appropriate means, such as by means of firmware.

[0136] The various embodiments of the systems and techniques described above can be implemented in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a load programmable logic device (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0137] Computer programs used to implement the processes of the application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the computer program

[0138] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store computer programs for use by or in connection with an instruction execution system, apparatus, or device. Computer-readable storage media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium will include one or more lines of electrical connections, portable computer disks, hard disk drives, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0139] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0140] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0141] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. A server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0142] It should be understood that the various forms of flow shown above can be re-ordered, added to, or deleted from without departing from the scope of the present disclosure. For example, the steps recited in the present disclosure can be executed in parallel, executed in sequence, or executed in a different order, as long as the desired results of the present disclosure are achieved, and the present disclosure is not limited herein.

[0143] The specific embodiments described above are not intended to be limiting, and persons skilled in the art will appreciate that various modifications, combinations, sub-combinations and alternatives can be made to the specific embodiments without departing from the spirit and principles of the disclosure. Accordingly, the disclosure is not limited to the specific embodiments described above, but only by the scope of the appended claims.

Claims

1. A method for predicting transformer operating parameters, characterized in that, include: Determine at least one current operating parameter group of the target transformer, and determine the operating parameter group sequence based on the at least one current operating parameter group. The current operating parameter group contains at least one operating parameter, and the operating parameter group sequence is sorted according to the time order of the acquisition time corresponding to the current operating parameter group. The sequence of operating parameter sets is input into a pre-constructed transformer operating parameter prediction model. At least one predicted operating parameter set for the target transformer is obtained through the model. The number of predicted operating parameter sets in the at least one predicted operating parameter set is less than the number of current operating parameter sets in the at least one current operating parameter set. The time interval between the acquisition times of each current operating parameter set in the at least one current operating parameter set is a preset time interval. The time interval between the prediction times of each predicted operating parameter set in the at least one predicted operating parameter set is a preset time interval. The time interval between the prediction time of the first predicted operating parameter set in the at least one predicted operating parameter set and the acquisition time of the first current operating parameter set in the at least one current operating parameter set is a preset time interval. The first predicted operating parameter set is the predicted operating parameter set with the earliest prediction time among the at least one predicted operating parameter sets, and the first current operating parameter set is the current operating parameter set with the latest acquisition time among the at least one current operating parameter sets. The transformer operating parameter prediction model includes a dynamic shared layer and a dynamic specific layer. The dynamic shared layer is used to perform orthogonal transformation on the operating parameter group sequence to generate a dynamic shared space. The dynamic specific layer is used to perform random nonlinear mapping on the dynamic shared space to generate a dynamic specific space. The transformer operating parameter prediction model determines the common change trend of each operating parameter in the operating parameter group through the dynamic shared layer, and determines the specific change trend of each operating parameter in the operating parameter group according to the dynamic specific layer, thereby realizing the prediction of the operating parameter group corresponding to the target transformer. Specifically, at least one set of predicted operating parameters for the target transformer is obtained through the transformer operating parameter prediction model, including: Perform an orthogonal transformation on the sequence of operating parameters to generate the current dynamic shared space; The dynamic shared space is subjected to a random nonlinear mapping to generate the current dynamic specific space; Based on the current dynamic shared space, the current dynamic specific space, the pre-determined shared weight of the current dynamic shared space, and the pre-determined specific weight of the current dynamic specific space, at least one set of predicted operating parameters for the target transformer is determined, wherein the sum of the shared weight and the specific weight is 1.

2. The method according to claim 1, characterized in that, The training process of the transformer operating parameter prediction model includes: Obtain a training sample data set of the target transformer. The training sample data set includes at least one training sample operating parameter set of the target transformer and at least one first training sample operating parameter set corresponding to the at least one training sample operating parameter set. The first training sample operating parameter set is the actual training sample operating parameter set corresponding to the training sample operating parameter set. Determine the sequence of training sample running parameter groups based on the at least one training sample running parameter group; The sequence of training sample running parameter groups is orthogonally transformed to determine the dynamic shared space of the samples. By performing a random nonlinear mapping on the dynamic shared space, a sample dynamic specific space is obtained; Based on the initial shared weight of the pre-set dynamic shared space, the initial specific weight of the pre-set dynamic specific space, the sample dynamic shared space, and the sample dynamic specific space, at least one second training sample running parameter group of the target transformer is determined, and the second training sample running parameter group is the predicted training sample running parameter group corresponding to the training sample running parameter group. Based on the at least one second training sample operating parameter group and the at least one first training sample operating parameter group of the sample transformer, the initial shared weight of the preset dynamic shared space and the initial specific weight of the preset dynamic specific space are adjusted. The training sample data group of the target transformer is obtained again, and the initial specific weight and the initial shared weight are adjusted again until the number of adjustments reaches the preset number, so as to obtain the transformer operating parameter prediction model.

3. The method according to claim 2, characterized in that, Adjusting the initial sharing weight of the pre-set dynamic shared space and the initial specific weight of the pre-set dynamic specific space includes: Determine the objective function for the initial specific weights; The Crank-Nicolson class parameter optimization algorithm based on the alternating direction multiplier method is used to solve the weight objective function to obtain the adjusted initial specific weights.

4. The method according to claim 2, characterized in that, Based on the at least one second training sample running parameter set and the at least one first training sample running parameter set of the sample transformer, the initial shared weights of the pre-set dynamic shared space and the initial specific weights of the pre-set dynamic specific space are adjusted, including: Based on row sparsity constraints, the initial specific weights are constrained to obtain constrained initial specific weights; The initial specific weights of the constraints of the pre-set dynamic specific space are adjusted based on the at least one set of predicted training sample operating parameters and the at least one set of first training sample operating parameters of the sample transformer.

5. The method according to claim 4, characterized in that, Based on row sparsity constraints, the initial specific weights are constrained to obtain constrained initial specific weights, including: Based on the L1 norm and L2 norm, the initial specific weights are constrained to obtain the constrained initial specific weights.

6. The method according to claim 3, characterized in that, The expression for the weighted objective function is: ； In the formula, It is a Lagrange multiplicative factor; For the target transformer, at least one set of predicted operating parameters is used in the second number of adjacent predicted acquisition frames. The sequence of operating parameters; r is the output / input dimension, i.e., the number of operating parameters; s is the number of samples; These are shared weights and specific weights, respectively. t is the dimension of the dynamically shared subspace; d is the dimension of the dynamically specific subspace; It is an orthogonal decoupling matrix; Let be a random matrix; ξ be a nonlinear activation function; where α, β and β are equilibrium parameters; It is the F-norm; For residual terms; ; ; ε is the set tolerance threshold; ε > 0. yes The i-th row.

7. A transformer operating parameter prediction device, characterized in that, include: The current parameter determination module is used to determine at least one current operating parameter group of the target transformer, and to determine the operating parameter group sequence based on the at least one current operating parameter group. The current operating parameter group contains at least one operating parameter, and the operating parameter group sequence is sorted according to the time order of the acquisition time corresponding to the current operating parameter group. The parameter prediction module is used to input the sequence of operating parameter groups into a pre-constructed transformer operating parameter prediction model, and obtain at least one predicted operating parameter group for the target transformer through the transformer operating parameter prediction model. The number of predicted operating parameter groups in the at least one predicted operating parameter group is less than the number of current operating parameter groups in the at least one current operating parameter group. The time interval between the acquisition times of each current operating parameter group in the at least one current operating parameter group is a preset time interval. The time interval between the prediction times of each predicted operating parameter group in the at least one predicted operating parameter group is a preset time interval. The time interval between the prediction time of the first predicted operating parameter group in the at least one predicted operating parameter group and the acquisition time of the first current operating parameter group in the at least one current operating parameter group is a preset time interval. The first predicted operating parameter group is the predicted operating parameter group with the earliest prediction time in the at least one predicted operating parameter group, and the first current operating parameter group is the current operating parameter group with the latest acquisition time in the at least one current operating parameter group. The transformer operating parameter prediction model includes a dynamic shared layer and a dynamic specific layer. The dynamic shared layer is used to perform orthogonal transformation on the operating parameter group sequence to generate a dynamic shared space. The dynamic specific layer is used to perform random nonlinear mapping on the dynamic shared space to generate a dynamic specific space. The transformer operating parameter prediction model determines the common change trend of each operating parameter in the operating parameter group through the dynamic shared layer, and determines the specific change trend of each operating parameter in the operating parameter group according to the dynamic specific layer, thereby realizing the prediction of the operating parameter group corresponding to the target transformer. The parameter prediction module includes: Perform an orthogonal transformation on the sequence of operating parameters to generate the current dynamic shared space; The dynamic shared space is subjected to a random nonlinear mapping to generate the current dynamic specific space; Based on the current dynamic shared space, the current dynamic specific space, the pre-determined shared weight of the current dynamic shared space, and the pre-determined specific weight of the current dynamic specific space, at least one set of predicted operating parameters for the target transformer is determined, wherein the sum of the shared weight and the specific weight is 1.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the transformer operating parameter prediction method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the transformer operating parameter prediction method according to any one of claims 1-6.

Images