A power distribution transformer on-load voltage regulation method, device, equipment and medium

Abstract

Embodiments of the present application provide a power distribution transformer on-load voltage regulating method, device, equipment and medium, and relate to the technical field of power distribution transformers in electrical engineering, which comprises: clustering load historical data corresponding to a target power distribution transformer to obtain a plurality of cluster centers; selecting a corresponding target cluster center from the plurality of cluster centers according to the category corresponding to the target power distribution transformer, and selecting a candidate power distribution transformer with the closest Euclidean distance to the target power distribution transformer through the target cluster center; performing principal component dimension reduction on the operation data corresponding to the candidate power distribution transformer to obtain reduced dimension data, inputting the reduced dimension data into a preset voltage prediction model, and outputting a voltage prediction value corresponding to a target period; and predicting a target voltage gear corresponding to the target period according to the voltage prediction value and an actual voltage value. The present application can predict the future voltage of the power distribution transformer, improve the stability of the equipment, reduce the action of the on-load voltage regulating switch, and improve the intelligent level of the power distribution transformer.

Description

Technical Field

[0001] This invention relates to the field of distribution transformer technology in electrical engineering, and more specifically, to an on-load tap changer method, apparatus, equipment, and medium for distribution transformers. Background Technology

[0002] On-load tap changer technology is mostly used in main transformers of substations above 35kV. In recent years, it has begun to be applied in 10kV distribution transformers. Its conventional control method is to collect the current low-voltage side voltage value of the transformer. If the voltage value is continuously lower than the step-up threshold for a certain delay, a step-up command is issued and the transformer steps up one level. Conversely, if the voltage value is continuously higher than the step-down threshold for a certain delay, a step-down command is issued and the transformer steps down one level. If the voltage is between the step-up and step-down thresholds, the current level is maintained.

[0003] With the large-scale integration of distributed photovoltaic power generation equipment and electric vehicle charging equipment into the power distribution network, the voltage fluctuation amplitude and speed of the power distribution network have increased. Therefore, conventional on-load tap changer control methods applied to 10kV distribution transformers have the following problems: if the set delay time is too long, it cannot quickly adjust the voltage to maintain the stability of the low-voltage side of the transformer; if the delay time is too short, the on-load tap changer of the transformer will frequently operate with the fluctuating voltage, reducing the equipment life; the changing patterns of similar transformers cannot be fully accumulated and effectively identified, so voltage regulation control can only be performed based on the current voltage situation, and it is not possible to predict voltage changes in advance and respond accordingly; current voltage regulation control methods can only adjust the voltage step by step, and cannot directly adjust to the ideal voltage level so that the load can operate under the ideal voltage. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide an on-load voltage regulation method, apparatus, equipment and medium for distribution transformers, so as to solve the problem that the existing voltage regulation methods are not intelligent enough.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] In a first aspect, the present invention provides an on-load tap changing method for a distribution transformer, comprising:

[0007] Multiple cluster centers are obtained by clustering the historical load data corresponding to the target distribution transformer;

[0008] Based on the type of the target distribution transformer, a corresponding target cluster center is selected from multiple cluster centers, and a candidate distribution transformer with the closest Euclidean distance to the target distribution transformer is selected through the target cluster center;

[0009] Principal component dimensionality reduction is performed on the operating data corresponding to the candidate distribution transformers to obtain dimensionality-reduced data. The dimensionality-reduced data is then input into a preset voltage prediction model, and the voltage prediction value corresponding to the target time period is output.

[0010] The target voltage level corresponding to the target time period is predicted based on the predicted voltage value and the actual voltage value.

[0011] In an optional implementation, the step of clustering the historical load data corresponding to the target distribution transformer to obtain multiple cluster centers includes:

[0012] Obtain the same type of distribution transformers in the jurisdiction where the target distribution transformer is located, and obtain the load history data of the same type of distribution transformers;

[0013] Each of the same type of distribution transformers is treated as a data unit to obtain multiple data units, and multi-dimensional parameters corresponding to each data unit are generated.

[0014] A preset number of initial cluster centers are randomly selected from the plurality of data units. The Euclidean distance between each data unit and the initial cluster center is calculated. The initial cluster center to which each data unit belongs is determined based on the Euclidean distance.

[0015] Calculate the average value of the multi-dimensional parameters for each initial cluster center based on the data units contained in each initial cluster center;

[0016] The average value of the multi-dimensional parameters is used as the new cluster center, and the step of calculating the Euclidean distance between each data unit and the initial cluster center is repeated until the new cluster center no longer changes, thereby obtaining multiple cluster centers.

[0017] In an optional implementation, the step of performing principal component dimensionality reduction on the operating data corresponding to the candidate distribution transformer to obtain dimensionality-reduced data includes:

[0018] Obtain the multi-dimensional variables corresponding to the operational data, including digitized variables and non-digitized variables;

[0019] The non-digital variables are converted into digital variables, and the digital variables are converted into a data matrix;

[0020] The covariance matrix is ​​obtained by removing the mean from the data matrix, and the eigenvalues ​​are obtained by performing eigenvalue decomposition on the covariance matrix.

[0021] Obtain a preset number of feature vector matrices corresponding to the feature values, and fuse the feature vector matrices to obtain the dimensionality-reduced data.

[0022] In an optional implementation, the step of inputting the dimensionality-reduced data into a preset voltage prediction model and outputting the voltage prediction value corresponding to the target time period includes:

[0023] Select target dimensionality-reduced data corresponding to a preset time period from the dimensionality-reduced data, and generate time series data based on the target dimensionality-reduced data;

[0024] The time series data is input into the preset voltage prediction model, and the preset voltage prediction model is used to predict the target time series data corresponding to the target time period; the target time period is the period after the preset time period.

[0025] Obtain the regression average value of the target time series data, and output the regression average value as the voltage prediction value.

[0026] In an optional implementation, the step of predicting the target voltage level corresponding to the target time period based on the predicted voltage value and the actual voltage value includes:

[0027] Obtain the operating voltage value corresponding to the operating data, and construct a comprehensive regression voltage calculation formula based on the operating voltage value and the voltage prediction value;

[0028] Obtain the actual voltage value corresponding to the predicted voltage value, wherein the predicted voltage value and the actual voltage value correspond to the same time period;

[0029] The comprehensive regression voltage calculation formula is adjusted using the actual voltage value to obtain the target voltage calculation formula;

[0030] The final voltage prediction value is calculated using the target voltage calculation formula, and the target voltage level is determined based on the final voltage prediction value.

[0031] In an optional implementation, after the step of determining the target voltage level based on the final voltage prediction value, the method further includes:

[0032] Determine whether the target voltage level is the same as the current voltage level;

[0033] If they are different, when the tap changer has completed energy storage, there is no short circuit or overcurrent on the load side, the daily voltage adjustment frequency has not reached the limit, and the switch has not malfunctioned, the voltage level will be adjusted from the current voltage level to the target voltage level.

[0034] In an optional implementation, the step of adjusting the voltage range from the current voltage range to the target voltage range includes:

[0035] If the on-load tap changer supports cross-range voltage adjustment, then the voltage range is directly adjusted to the target voltage range;

[0036] If the on-load tap changer does not support cross-level voltage regulation, then the voltage level is adjusted to a level close to the target voltage level.

[0037] In a second aspect, the present invention provides an on-load tap changing device for a distribution transformer, comprising:

[0038] The data clustering module is used to cluster the historical load data corresponding to the target distribution transformer to obtain multiple cluster centers.

[0039] The transformer selection module is used to select a corresponding target cluster center from multiple cluster centers according to the type of the target distribution transformer, and select a candidate distribution transformer with the closest Euclidean distance to the target distribution transformer through the target cluster center;

[0040] The voltage prediction module is used to perform principal component dimensionality reduction on the operating data corresponding to the candidate distribution transformer to obtain dimensionality-reduced data, input the dimensionality-reduced data into a preset voltage prediction model, and output the voltage prediction value corresponding to the target time period.

[0041] The voltage level prediction module is used to predict the target voltage level corresponding to the target time period based on the predicted voltage value and the actual voltage value.

[0042] Thirdly, the present invention provides an electronic device including a processor and a memory, the memory storing machine-executable instructions that can be executed by the processor, the processor executing the machine-executable instructions to implement the on-load tap-changing method for a distribution transformer as described in the first aspect.

[0043] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the on-load tap changing method for a distribution transformer as described in the first aspect.

[0044] The present invention provides an on-load tap changer method, apparatus, equipment and medium for distribution transformers. By utilizing a large amount of existing transformer operating data, the future voltage of the distribution transformer can be predicted, thereby improving the stability of the equipment, reducing the operation of on-load tap changers, and improving the intelligence level of the distribution transformer. Attached Figure Description

[0045] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 A block diagram of an electronic device provided by an embodiment of the present invention is shown;

[0047] Figure 2 A schematic flowchart of an on-load tap changing method for a distribution transformer provided by an embodiment of the present invention is shown.

[0048] Figure 3 A schematic flowchart of a principal component dimensionality reduction method provided in an embodiment of the present invention is shown;

[0049] Figure 4 This diagram illustrates the principle of a preset voltage prediction model training method provided by an embodiment of the present invention.

[0050] Figure 5 The diagram shows a functional block diagram of an on-load tap changer for a distribution transformer provided in an embodiment of the present invention.

[0051] icon:

[0052] 100 - Electronic equipment; 110 - Memory; 120 - Processor; 130 - Communication module; 500 - On-load tap changer for distribution transformer; 510 - Data clustering module; 520 - Transformer selection module; 530 - Voltage prediction module; 540 - Gear position prediction module. Detailed Implementation

[0053] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0054] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0055] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0056] Please refer to Figure 1 , Figure 1 This is a block diagram of an electronic device 100 provided in this embodiment. The electronic device 100 includes a memory 110, a processor 120, and a communication module 130. The memory 110, processor 120, and communication module 130 are electrically connected to each other directly or indirectly to realize data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines.

[0057] The memory 110 is used to store programs or data. The memory 110 may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0058] The processor 120 is used to read / write data or programs stored in the memory 110 and to perform corresponding functions.

[0059] The communication module 130 is used to establish a communication connection between the electronic device 100 and other communication terminals through the network, and to send and receive data through the network.

[0060] It should be understood that, Figure 1 The structure shown is only a schematic diagram of the electronic device 100. The electronic device 100 may also include components that are larger than... Figure 1The more or fewer components shown, or having the same Figure 1 The different configurations shown. Figure 1 The components shown can be implemented using hardware, software, or a combination thereof.

[0061] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating an on-load tap changing method for a distribution transformer provided in this embodiment. The method includes:

[0062] S210. Cluster the historical load data corresponding to the target distribution transformer to obtain multiple cluster centers.

[0063] Because the low-voltage side voltage of a distribution transformer is related to the transformer ratio, but mainly to the high-voltage side voltage, the transformer's short-circuit impedance, and the load size, this embodiment calculates the low-voltage side voltage to the rated transformer ratio, thus disregarding the influence of the transformer ratio. The high-voltage side voltage, originating from the system and unaffected by a single distribution transformer, is also disregarded. The transformer's short-circuit impedance is proportional to its capacity; transformers of different efficiency levels have the same short-circuit impedance. Therefore, historical data from transformers of the same capacity can be used as a reference, without considering the specific efficiency level.

[0064] If the target distribution transformer does not have historical load data, the historical load data of other transformers of the same capacity can be used as a reference, and then clustering can be performed to obtain multiple cluster centers. K-Means clustering can be used as the clustering method.

[0065] S220. Select a corresponding target cluster center from multiple cluster centers according to the type of the target distribution transformer, and select a candidate distribution transformer with the closest Euclidean distance to the target distribution transformer through the target cluster center.

[0066] After obtaining multiple cluster centers through clustering, each cluster center represents a type of distribution transformer. The corresponding target cluster center is selected according to the type of the target distribution transformer. Then, in the target cluster center, the Euclidean distance between the target distribution transformer and other distribution transformers is calculated, and the other distribution transformers with the closest Euclidean distance to the target distribution transformer are selected as candidate distribution transformers.

[0067] S230. Perform principal component dimensionality reduction on the operating data corresponding to the candidate distribution transformer to obtain dimensionality-reduced data. Input the dimensionality-reduced data into a preset voltage prediction model and output the voltage prediction value corresponding to the target time period.

[0068] The operating data of the candidate distribution transformer is regarded as the operating data of the target distribution transformer. Since the operating data is affected by different factors and the dimensions of the data of different factors are relatively complex, in order to reduce the amount of calculation, the data of different factors can be processed by principal component dimensionality reduction to obtain dimensionality-reduced data. Then, the dimensionality-reduced data is processed by a preset voltage prediction model to predict the voltage prediction value corresponding to the target time period.

[0069] S240. Predict the target voltage level corresponding to the target time period based on the predicted voltage value and the actual voltage value.

[0070] After determining the predicted voltage value, the target voltage level corresponding to the predicted voltage can be determined. Then, the level can be adjusted in conjunction with the current voltage level corresponding to the current voltage, thereby achieving voltage prediction and advance level adjustment.

[0071] This embodiment utilizes a large amount of existing transformer operating data to predict the future voltage of the distribution transformer, thereby improving equipment stability, reducing the operation of on-load tap changers, and enhancing the intelligence level of the distribution transformer.

[0072] In one implementation, the step of clustering the historical load data corresponding to the target distribution transformer to obtain multiple cluster centers includes:

[0073] Obtain the same type of distribution transformers in the jurisdiction where the target distribution transformer is located, and obtain the load history data of the same type of distribution transformers;

[0074] Each of the same type of distribution transformers is treated as a data unit to obtain multiple data units, and multi-dimensional parameters corresponding to each data unit are generated.

[0075] A preset number of initial cluster centers are randomly selected from the plurality of data units. The Euclidean distance between each data unit and the initial cluster center is calculated. The initial cluster center to which each data unit belongs is determined based on the Euclidean distance.

[0076] Calculate the average value of the multi-dimensional parameters for each initial cluster center based on the data units contained in each initial cluster center;

[0077] The average value of the multi-dimensional parameters is used as the new cluster center, and the step of calculating the Euclidean distance between each data unit and the initial cluster center is repeated until the new cluster center no longer changes, thereby obtaining multiple cluster centers.

[0078] The operating conditions of distribution transformers in the same jurisdiction are usually the same. Therefore, when the load history data of the target distribution transformer cannot be obtained, such as when the target distribution transformer is a newly added distribution transformer in the current jurisdiction, in order to intelligently control the target distribution transformer, the load history data of other distribution transformers of the same type in the jurisdiction can be obtained first and clustered.

[0079] Using a single transformer as a data unit and the most recent year as the time limit, parameters affecting voltage changes, such as the daily average minimum and maximum load values, the average maximum demand over 15 minutes, and the daily average values, are selected as the dimensions of this data unit.

[0080] Specify K classes and randomly select K cluster centers Ci (1≤i≤K) for the data units. Calculate the Euclidean distance between the other data units and the cluster centers Ci, and assign the data units to the class represented by the nearest cluster center. Then calculate the average value of each dimension parameter of all data units in each class, and use this value as the new cluster center. Perform the next iteration until the cluster centers no longer change or the maximum number of iterations is reached.

[0081] Generally, the higher the computer configuration, the larger the value of K, and the greater the corresponding computational load.

[0082] The formula for calculating Euclidean distance is as follows: Assume there are two points in n-dimensional space:

[0083] a (x 11 ,x 12 ...,x 1n ), b(x 21 ,x 22 ...,x 2n If we consider a and b, then the formula for the distance between points a and b in n-dimensional space is:

[0084] d 12 =

[0085] By calculating the Euclidean distance from the target distribution transformer to each cluster center, the closest one belongs to the corresponding cluster. Then, the same method can be used to select the corresponding candidate distribution transformer based on the target cluster center.

[0086] This embodiment obtains historical load data of transformers of the same type as the target distribution transformer as a reference, performs clustering processing, selects the candidate transformer that is closest to the target distribution transformer based on the clustering results, and then uses the operating data of the candidate distribution transformer as the operating data of the target distribution transformer, making the obtained operating data of the target distribution transformer more accurate and scientific.

[0087] In one embodiment, the step of performing principal component dimensionality reduction on the operating data corresponding to the candidate distribution transformer to obtain dimensionality-reduced data includes:

[0088] Obtain the multi-dimensional variables corresponding to the operational data, including digitized variables and non-digitized variables;

[0089] The non-digital variables are converted into digital variables, and the digital variables are converted into a data matrix;

[0090] The covariance matrix is ​​obtained by removing the mean from the data matrix, and the eigenvalues ​​are obtained by performing eigenvalue decomposition on the covariance matrix.

[0091] Obtain a preset number of feature vector matrices corresponding to the feature values, and fuse the feature vector matrices to obtain the dimensionality-reduced data.

[0092] Please refer to Figure 3 , Figure 3 This is a flowchart illustrating a principal component dimensionality reduction method provided in this embodiment.

[0093] Principal component analysis (PCA) can be used to reduce dimensionality. After determining the candidate distribution transformers, the historical data of the candidate distribution transformers for the most recent year are obtained. The data includes season, weather, wind force, temperature, humidity, sampling time, regression voltage at sampling time, current at sampling time, power factor, etc.

[0094] First, digitize the non-digital variables: for example, the seasons spring, summer, autumn, and winter are transformed into 1, 2, 3, and 4, and the weather conditions such as sunny, cloudy, overcast, foggy, rainy, and snowy are transformed into 1, 2, 3, 4, 5, and 6.

[0095] Then, a data matrix is ​​constructed. Each row of the matrix corresponds to all the data of the above categories at a sampling time. For example, if sampling is usually performed once per minute, then one row of data is generated per minute, which is set to n rows. Each column represents the data of the same category at different sampling times, which is set to d columns.

[0096] Then, the data is averaged, which means subtracting the average value of each column from the data in all columns to obtain the matrix. .

[0097] Find the covariance matrix S:

[0098]

[0099] Perform eigenvalue decomposition on the covariance matrix S:

[0100]

[0101] where the matrix Λ=diag(λ1, λ2,…,λ d), λ1≥λ2≥…≥λ d Let V be the eigenvalues ​​of the covariance matrix S, and let V be the eigenvector matrix corresponding to each eigenvalue: V = [V1, V2, ..., V2]. d Based on the hardware specifications for running this calculation program, select the first k eigenvalues ​​to make the following equation true:

[0102]

[0103] A value of γ greater than 0.8 is acceptable; the better the hardware, the larger the γ should be. Utilize the eigenvector W corresponding to these k eigenvalues. k =[V1,…,V k This allows d-dimensional data to be reduced to k-dimensional data Z:

[0104]

[0105] The above regression voltage is the voltage sample value multiplied by (K) C / K0), where K C K0 represents the actual transformer ratio at the time of sampling, such as 10.5 / 0.23. K0 is the standard transformer ratio, which is 10 / 0.23 for distribution transformers.

[0106] This embodiment addresses the computation, storage, and modeling challenges posed by high-dimensional data by reducing the dimensionality of multidimensional operational data while preserving the core information of the data. Ultimately, this improves data processing efficiency and model performance.

[0107] In one implementation, the step of inputting the dimensionality-reduced data into a preset voltage prediction model and outputting the voltage prediction value corresponding to the target time period includes:

[0108] Select target dimensionality-reduced data corresponding to a preset time period from the dimensionality-reduced data, and generate time series data based on the target dimensionality-reduced data;

[0109] The time series data is input into the preset voltage prediction model, and the preset voltage prediction model is used to predict the target time series data corresponding to the target time period; the target time period is the period after the preset time period.

[0110] Obtain the regression average value of the target time series data, and output the regression average value as the voltage prediction value.

[0111] Please refer to Figure 4 , Figure 4 This is a schematic diagram illustrating the principle of a preset voltage prediction model training method provided in this embodiment.

[0112] The preset voltage prediction model can adopt the LSTM (Long Short-Term Memory) neural network model. First, initialize all weights in the input layer to the output layer of the LSTM neural network to 1, initialize all biases to 0, initialize all cell states C and hidden states h in the hidden layer to 0, and set the number of loops. Setting the number of loops too small may result in a large error, while setting it too large will make the program run for too long. Generally, it is set to about 500 times. If the final error is too large after the program is calculated, the number of loops can be increased, and if the error is too small, the number of loops can be reduced to shorten the calculation time.

[0113] Then, all data within every N minutes is selected as a time series data set, which serves as the training input X for the LSTM (Long Short-Term Memory) neural network. For example, if N is 5, then X consists of 5 rows and k columns, representing the k data points from the 1st minute to the 5th minute. The hidden layers of the LSTM neural network contain a self-loop structure. The expression for the forget gate output in this self-loop structure is as follows:

[0114] f t =σ(W f ×[h t-1 X t ]+b f )

[0115] Among them W f It is the weight matrix of the forget gate, h t-1 It is the hidden state from the previous moment, X t It is the input value at the current moment, b f σ is the bias term of the forget gate, and σ is the sigmoid activation function.

[0116] The expression for the input gate's output is as follows:

[0117] i t =σ(W i ×[h t-1 ,X t ]+b i )

[0118] Among them W i It is the weight matrix of the input gate, b i It is the bias term of the input gate.

[0119] The candidate cell state expression is as follows:

[0120] tanh(W C ×[h t-1 ,X t ]+b C )

[0121] Among them W C This is the weight matrix of the candidate cells, b cis the bias term for candidate cells, and tanh is the hyperbolic tangent activation function.

[0122] The cell state expression is as follows:

[0123] C t =f t ⊙C t-1 + i t ⊙

[0124] Where C t-1 The cell state at the previous moment is represented by ⊙, which indicates element-wise multiplication.

[0125] The output expression of the output gate is as follows:

[0126] O t =σ(W o ×[h t-1 ,X t ]+b o )

[0127] Where W0 is the weight matrix of the output gate, b o It is the bias term of the output gate.

[0128] The expression for the current hidden state is as follows:

[0129] h t = O t ⊙tanh(C t )

[0130] In the above expressions, the subscript 't' represents the value at the current time, and 't-1' represents the value at the previous time. The first time series is the k-dimensional data of the first minute. During calculation, the value at the 't-1' subscript is initialized to 0, and the value at the 't' subscript is the current value of the first minute. The second time series is the data of the second minute. During calculation, the value at the 't-1' subscript is the calculated value of the first minute, and the value at the 't' subscript is the current value of the second minute; and so on, to obtain the output value of the corresponding hidden layer for each time series.

[0131] After completing one iteration of the network's computation in the output layer of the LSTM neural network, the output value for each time series can be obtained. The average regression voltage over the following 15 minutes of each time series data is used as the training output value for that time series. For example, if the first 5 minutes are the input time series data, then the average regression voltage from the historical data from the 6th to the 20th minute is used as the training output value for each time series. The regression voltage value is calculated as described above; the average of the regression values ​​over the following 15 minutes of the sequence data is the 15-minute average regression voltage. The difference between the two output values ​​is calculated, which is the sample error. Training ends when the sample error meets the condition or reaches the required number of iterations; otherwise, the weights and biases from the input layer to the output layer are adjusted using gradient descent, and the calculation is repeated.

[0132] During the training process of the preset voltage prediction model, voltage prediction can be performed using it. Historical data can be selected using a windowing method during prediction. For example, in the first set of data: the input values ​​are time series data from 1 to 5 minutes, and the average regression voltage from 6 to 20 minutes is the training output value. With each windowing step being 1 minute, in the second set of data: the input values ​​are time series data from 2 to 6 minutes, and the average regression voltage from 7 to 21 minutes is the training output value, and so on.

[0133] Once the preset voltage prediction model has been trained, it can be used for voltage prediction. The target dimensionality-reduced data is used to generate time series data, which is then input into the preset voltage prediction model to predict the target time series data. For example, using the data from the first 5 minutes, the predicted voltage value for the next 15 minutes can be predicted.

[0134] This embodiment trains a preset voltage prediction model and uses the trained preset voltage prediction model to predict the target time series of the target period. Based on the target time series, the predicted voltage value is obtained, thereby realizing the advance prediction of voltage and providing a voltage basis for subsequent on-load voltage regulation.

[0135] In one embodiment, the step of predicting the target voltage level corresponding to the target time period based on the predicted voltage value and the actual voltage value includes:

[0136] Obtain the operating voltage value corresponding to the operating data, and construct a comprehensive regression voltage calculation formula based on the operating voltage value and the voltage prediction value;

[0137] Obtain the actual voltage value corresponding to the predicted voltage value, wherein the predicted voltage value and the actual voltage value correspond to the same time period;

[0138] The comprehensive regression voltage calculation formula is adjusted using the actual voltage value to obtain the target voltage calculation formula;

[0139] The final voltage prediction value is calculated using the target voltage calculation formula, and the target voltage level is determined based on the final voltage prediction value.

[0140] After obtaining the predicted voltage, the actual voltage value corresponding to the predicted voltage can be used for continuous optimization and adjustment.

[0141] For example, by inputting the target dimensionality-reduced data into a preset voltage prediction model, the predicted voltage value is Y. P Among them, the voltage data corresponding to the target dimensionality reduction data is Y. X Then Y P and Y X Multiplied by their respective weights ω P and ω X The comprehensive regression voltage Y is obtained Z Right now:

[0142] Y Z =Y P ×ω P + Y X ×ω X

[0143] Where the weight ω P and ω X The sum is always 1, initialized to 0.5. The accuracy of the comprehensive regression voltage is judged based on the actual voltage after 15 minutes; if accurate, ω is increased. P and reduce ω X If inaccurate, reduce ω. P and increase ω X For example, each increase or decrease is 0.01.

[0144] In weight ω P and ω X Once the voltage stabilizes, the above formula can be used to calculate the composite regression voltage, which can then be used as the final predicted voltage.

[0145] This embodiment uses the actual voltage to correct the calculation formula of the comprehensive regression voltage, making the difference between the predicted voltage and the actual voltage smaller and increasing the accuracy of voltage prediction.

[0146] In one embodiment, after the step of determining the target voltage level based on the final voltage prediction value, the method further includes:

[0147] Determine whether the target voltage level is the same as the current voltage level;

[0148] If they are different, when the tap changer has completed energy storage, there is no short circuit or overcurrent on the load side, the daily voltage adjustment frequency has not reached the limit, and the switch has not malfunctioned, the voltage level will be adjusted from the current voltage level to the target voltage level.

[0149] Calculate the values ​​of the regression voltage Yz adjusted to all taps of the on-load tap changer on the transformer:

[0150] Y T =[Y T1 ,…,Y TN ]

[0151] Where Y T1 =Yz× K0 / K1,…,Y TN = Yz× K0 / K n K0 is the standard turns ratio, K1…K n The strain ratio for all tap positions of the on-load tap changer of the transformer.

[0152] Calculate Y T The ideal target voltage value Y on the low-voltage side of the same transformer m The absolute value of Y, where the minimum value is Min(|Y). T -Y m The corresponding gear is the target gear for on-load tap regulation control. If the target gear is the same as the current gear, no gear adjustment is performed; otherwise, on-load tap regulation is performed.

[0153] Before performing on-load voltage regulation, a voltage regulation judgment is required.

[0154] If the on-load tap changer supports cross-range voltage adjustment, then the voltage range is directly adjusted to the target voltage range;

[0155] If the on-load tap changer does not support cross-level voltage regulation, then the voltage level is adjusted to a level close to the target voltage level.

[0156] For example: If the current gear is 5 and the target gear is 3, then if the on-load tap changer supports cross-gear voltage adjustment, adjust directly to gear 3; otherwise, adjust to gear 4.

[0157] This embodiment ensures the success rate of voltage regulation and equipment safety by performing voltage regulation only after all on-load voltage regulation conditions are met.

[0158] To perform the corresponding steps in the above embodiments and various possible methods, an implementation method of an on-load tap changer for a distribution transformer is given below. Please refer to [link / reference needed]. Figure 5 , Figure 5 This is a functional block diagram of an on-load tap changer for a distribution transformer provided in an embodiment of the present invention. It should be noted that the on-load tap changer for a distribution transformer provided in this embodiment has the same basic principle and technical effects as those in the above embodiments. For the sake of brevity, any parts not mentioned in this embodiment can be referred to the corresponding content in the above embodiments. The on-load tap changer 500 for the distribution transformer includes:

[0159] The data clustering module 510 is used to cluster the historical load data corresponding to the target distribution transformer to obtain multiple cluster centers.

[0160] The transformer selection module 520 is used to select a corresponding target cluster center from multiple cluster centers according to the type of the target distribution transformer, and select a candidate distribution transformer with the closest Euclidean distance to the target distribution transformer through the target cluster center;

[0161] The voltage prediction module 530 is used to perform principal component dimensionality reduction on the operating data corresponding to the candidate distribution transformer to obtain dimensionality-reduced data, input the dimensionality-reduced data into a preset voltage prediction model, and output the voltage prediction value corresponding to the target time period.

[0162] The voltage level prediction module 540 is used to predict the target voltage level corresponding to the target time period based on the predicted voltage value and the actual voltage value.

[0163] Optionally, the above modules can be stored in the form of software or firmware. Figure 1 The memory shown is either stored in or embedded in the operating system (OS) of the electronic device, and can be used by... Figure 1 The processor executes the commands. Meanwhile, the data and program code required to execute these modules can be stored in memory.

[0164] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0165] In addition, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0166] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0167] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for on-load voltage regulation of a distribution transformer, characterized in that, include: Multiple cluster centers are obtained by clustering the historical load data corresponding to the target distribution transformer; Based on the type of the target distribution transformer, a corresponding target cluster center is selected from multiple cluster centers, and a candidate distribution transformer with the closest Euclidean distance to the target distribution transformer is selected through the target cluster center; Principal component dimensionality reduction is performed on the operating data corresponding to the candidate distribution transformers to obtain dimensionality-reduced data. The dimensionality-reduced data is then input into a preset voltage prediction model, and the voltage prediction value corresponding to the target time period is output. Predict the target voltage level corresponding to the target time period based on the predicted voltage value and the actual voltage value; The step of clustering the historical load data corresponding to the target distribution transformer to obtain multiple cluster centers includes: Obtain the same type of distribution transformers in the jurisdiction where the target distribution transformer is located, and obtain the load history data of the same type of distribution transformers; Each of the same type of distribution transformers is treated as a data unit to obtain multiple data units, and multi-dimensional parameters corresponding to each data unit are generated. A preset number of initial cluster centers are randomly selected from the plurality of data units. The Euclidean distance between each data unit and the initial cluster center is calculated. The initial cluster center to which each data unit belongs is determined based on the Euclidean distance. Calculate the average value of the multi-dimensional parameters for each initial cluster center based on the data units contained in each initial cluster center; The average value of the multi-dimensional parameters is used as the new cluster center, and the step of calculating the Euclidean distance between each data unit and the initial cluster center is repeated until the new cluster center no longer changes, thereby obtaining multiple cluster centers.

2. The on-load tap changing method for a distribution transformer according to claim 1, characterized in that, The step of performing principal component dimensionality reduction on the operating data corresponding to the candidate distribution transformers to obtain dimensionality-reduced data includes: Obtain the multi-dimensional variables corresponding to the operational data, including digitized variables and non-digitized variables; The non-digital variables are converted into digital variables, and the digital variables are converted into a data matrix; The covariance matrix is ​​obtained by removing the mean from the data matrix, and the eigenvalues ​​are obtained by performing eigenvalue decomposition on the covariance matrix. Obtain a preset number of feature vector matrices corresponding to the feature values, and fuse the feature vector matrices to obtain the dimensionality-reduced data.

3. The on-load tap changing method for a distribution transformer according to claim 2, characterized in that, The step of inputting the dimensionality-reduced data into a preset voltage prediction model and outputting the voltage prediction value corresponding to the target time period includes: Select target dimensionality-reduced data corresponding to a preset time period from the dimensionality-reduced data, and generate time series data based on the target dimensionality-reduced data; The time series data is input into the preset voltage prediction model, and the preset voltage prediction model is used to predict the target time series data corresponding to the target time period; the target time period is the period after the preset time period. Obtain the regression average value of the target time series data, and output the regression average value as the voltage prediction value.

4. The on-load tap changing method for a distribution transformer according to claim 3, characterized in that, The step of predicting the target voltage level corresponding to the target time period based on the predicted voltage value and the actual voltage value includes: Obtain the operating voltage value corresponding to the operating data, and construct a comprehensive regression voltage calculation formula based on the operating voltage value and the voltage prediction value; Obtain the actual voltage value corresponding to the predicted voltage value, wherein the predicted voltage value and the actual voltage value correspond to the same time period; The comprehensive regression voltage calculation formula is adjusted using the actual voltage value to obtain the target voltage calculation formula; The final voltage prediction value is calculated using the target voltage calculation formula, and the target voltage level is determined based on the final voltage prediction value.

5. The on-load tap changing method for a distribution transformer according to claim 4, characterized in that, After the step of determining the target voltage level based on the final voltage prediction value, the method further includes: Determine whether the target voltage level is the same as the current voltage level; If they are different, when the tap changer has completed energy storage, there is no short circuit or overcurrent on the load side, the daily voltage adjustment frequency has not reached the limit, and the switch has not malfunctioned, the voltage level will be adjusted from the current voltage level to the target voltage level.

6. The on-load tap changing method for a distribution transformer according to claim 5, characterized in that, The step of adjusting the voltage level from the current voltage level to the target voltage level includes: If the on-load tap changer supports cross-range voltage adjustment, then the voltage range is directly adjusted to the target voltage range; If the on-load tap changer does not support cross-level voltage regulation, then the voltage level is adjusted to a level close to the target voltage level.

7. An on-load tap changing device for a distribution transformer, characterized in that, For performing the on-load tap changing method for a distribution transformer as described in any one of claims 1-6, the on-load tap changing device for the distribution transformer comprises: The data clustering module is used to cluster the historical load data corresponding to the target distribution transformer to obtain multiple cluster centers. The transformer selection module is used to select a corresponding target cluster center from multiple cluster centers according to the type of the target distribution transformer, and select a candidate distribution transformer with the closest Euclidean distance to the target distribution transformer through the target cluster center; The voltage prediction module is used to perform principal component dimensionality reduction on the operating data corresponding to the candidate distribution transformer to obtain dimensionality-reduced data, input the dimensionality-reduced data into a preset voltage prediction model, and output the voltage prediction value corresponding to the target time period. The voltage level prediction module is used to predict the target voltage level corresponding to the target time period based on the predicted voltage value and the actual voltage value.

8. An electronic device, characterized in that, It includes a processor and a memory, the memory storing machine-executable instructions that can be executed by the processor, the processor executing the machine-executable instructions to implement the on-load tap changing method for a distribution transformer according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the on-load tap changing method for distribution transformers as described in any one of claims 1-6.

Images