Non-intrusive anti-islanding detection method and device, electronic equipment and storage medium

Abstract

The invention discloses a non-intrusive anti-islanding detection method and device, electronic equipment and a storage medium. The method comprises the following steps: determining target time sequence voltage data of a target power distribution network system; determining a target operation state of the target power distribution network system through a prediction model according to the target time sequence voltage data; and determining whether the target power distribution network system has an island phenomenon according to the target operation state. According to the method, the target operation state is determined according to the target time sequence voltage data through the prediction model, whether the island phenomenon occurs in the target power distribution network is determined according to the target operation state, the process of preprocessing the target time sequence voltage data is avoided, the prediction process is simplified, and the detection efficiency is improved; and the problems of non-detection area appearance and power quality are solved, and the accuracy and reliability of island detection are improved.

Description

Technical Field

[0001] The present invention relates to the field of microgrid technology, and in particular to a non-invasive anti-islanding detection method, device, electronic equipment and storage medium. Background Art

[0002] With the increasing popularity of solar photovoltaic systems (PV), unplanned islanding has become a growing concern for power companies, especially in areas with high PV penetration. Unplanned islanding occurs when a portion of the grid is electrically isolated from the main grid due to a fault or maintenance activity and continues to be powered by local distributed energy resources (such as solar PV). This phenomenon can cause poor power quality, damage sensitive equipment, and pose a safety threat to power company employees performing maintenance work.

[0003] The existing unplanned island detection methods are mainly divided into passive detection, active detection and machine learning-based methods. Passive detection methods are simple and easy to implement, but they have non-detection areas, that is, in some cases, these parameter changes are not enough to trigger protection measures, resulting in detection failure. Active detection methods may reduce power quality and thus affect user experience. Machine learning-based methods usually involve two steps: feature extraction and classifier construction. Although this method can improve recognition accuracy, the model calculation cost is high, and due to the complexity of its integrated network, it is easy to cause overfitting. Summary of the invention

[0004] The present invention provides a non-intrusive anti-islanding detection method, device, electronic equipment and storage medium to solve the problem of inaccurate islanding detection.

[0005] According to one aspect of the present invention, a non-intrusive anti-islanding detection method is provided, comprising:

[0006] Determine target time-series voltage data of a target distribution network system; the target time-series voltage data is used to characterize voltage changes at a target common coupling point in a target distribution network system connected to a distributed power source;

[0007] The target operating state of the target distribution network system is determined through a prediction model according to the target time series voltage data; the target operating state is used to characterize whether the target distribution network system has an islanding phenomenon; the prediction model is used to characterize the relationship between the time series voltage data and the operating state of the target distribution network system; the prediction model is unsupervisedly trained by using reference time series voltage data, and after the first loss obtained by the unsupervised training satisfies the first preset error, the model that satisfies the first preset error is obtained by supervised training based on the original time series voltage data and the reference operating state; the unsupervised training is used to learn the internal structure and characteristics of the reference time series voltage data; the supervised training is used to learn the correlation between the original time series voltage data and the reference operating state based on the internal structure and characteristics of the reference time series voltage data; the islanding phenomenon is the phenomenon that an independent energized circuit is formed in the target distribution network system after the target distribution network system is powered off;

[0008] Whether an islanding phenomenon occurs in the target distribution network system is determined according to the target operating state.

[0009] According to another aspect of the present invention, a non-intrusive anti-islanding detection device is provided, comprising:

[0010] A target time-series voltage data determination module is used to determine target time-series voltage data of a target distribution network system; the target time-series voltage data is used to characterize voltage changes at a target common coupling point in a target distribution network system connected to a distributed power source;

[0011] A target operating state determination module is used to determine the target operating state of a target distribution network system through a prediction model according to the target time series voltage data; the target operating state is used to characterize whether an islanding phenomenon occurs in the target distribution network system; the prediction model is used to characterize the relationship between the time series voltage data and the operating state of the target distribution network system; the prediction model is trained unsupervisedly using reference time series voltage data, and after the first loss obtained by the unsupervised training satisfies a first preset error, the model satisfying the first preset error is obtained by supervised training based on the original time series voltage data and the reference operating state; the unsupervised training is used to learn the internal structure and characteristics of the reference time series voltage data; the supervised training is used to learn the correlation between the original time series voltage data and the reference operating state based on the internal structure and characteristics of the reference time series voltage data; the islanding phenomenon is the phenomenon that an independent energized circuit is formed in the target distribution network system after the target distribution network system is powered off;

[0012] The islanding phenomenon determination module is used to determine whether the target distribution network system has an islanding phenomenon according to the target operating state.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] at least one processor; and

[0015] a memory communicatively connected to 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 so that the at least one processor can execute the non-intrusive anti-islanding detection method described in any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the non-intrusive anti-islanding detection method described in any embodiment of the present invention when executed.

[0018] The technical solution of the embodiment of the present invention determines the target time-series voltage data of the target distribution network system; determines the target operating state of the target distribution network system through a prediction model based on the target time-series voltage data, and the prediction model can effectively capture the long-range dependency and global information in the target time-series voltage data, solves the problems of the appearance of non-detection areas and the quality of power information, thereby improving the accuracy of determining the target operating state; determines whether an island phenomenon occurs in the target distribution network system based on the target operating state, thereby improving the accuracy of island detection. The method determines the target operating state based on the target time-series voltage data through a prediction model, and determines whether an island phenomenon occurs in the target distribution network based on the target operating state, thereby avoiding the process of preprocessing the target time-series voltage data, simplifying the prediction process, and improving the efficiency of detection. At the same time, it also solves the problems of non-detection areas and power quality in existing detection methods, thereby improving the accuracy and reliability of island detection.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 A flowchart of a non-intrusive anti-islanding detection method provided by an embodiment of the present invention;

[0022] Figure 2 A schematic diagram of the structure of a target power distribution network system provided by an embodiment of the present invention;

[0023] Figure 3 A schematic diagram of the architecture of the general part of a TST model provided by an embodiment of the present invention;

[0024] Figure 4 A schematic structural diagram of a non-invasive anti-islanding detection device provided by an embodiment of the present invention;

[0025] Figure 5 A schematic diagram of the structure of an electronic device for implementing the non-invasive anti-islanding detection method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] Figure 1 The flowchart of a non-intrusive anti-islanding detection method provided by an embodiment of the present invention is applicable to the case where an islanding phenomenon occurs inside a distribution network when the power is off. The method can be executed by a non-intrusive anti-islanding detection device, which can be implemented in the form of hardware and / or software. The non-intrusive anti-islanding detection device can be configured in any electronic device with network communication function. Figure 1 As shown, the method includes:

[0029] S110, determining target time-series voltage data of a target distribution network system.

[0030] The target time-series voltage data is used to characterize the voltage change at the target common coupling point in the target distribution network system connected to the distributed power source.

[0031] Furthermore, the target time-series voltage data is time-series voltage data directly obtained from the target power distribution network system.

[0032] The target distribution network system may be a system generated by simulating an actual distribution network system; or it may be an actual distribution network system.

[0033] Furthermore, the simulation process of the target distribution network system is as follows: using the MATLAB / SIMULINK simulation platform, the IEEE 13-node target distribution system model is constructed according to the nodes in the actual distribution network system that are prone to islanding, such as Figure 2 As shown, these are nodes in the actual distribution network system that are prone to islanding, and 634 nodes are installed with PV power stations, where the PV power stations are distributed power sources.

[0034] Specifically, the target time-series voltage data of the target common coupling point at the current moment is obtained from the target distribution network system.

[0035] Furthermore, if the target distribution network system is a simulated distribution network system, the target time-series voltage data of the target common coupling point at the current moment is directly simulated and generated. If the target distribution network system is an actual distribution network system, the target time-series voltage data of the target common coupling point at the current moment is collected by the power monitoring equipment installed at the common coupling point.

[0036] S120. Determine a target operating state of a target distribution network system through a prediction model according to target time series voltage data.

[0037] The target operation state is used to characterize whether the target distribution network system has an islanding phenomenon. The target operation state includes at least: a normal operation state, an islanding state, and a fault state.

[0038] Among them, the prediction model is used to characterize the relationship between the time-series voltage data and the operating status of the target distribution network system; the prediction model is trained unsupervisedly using reference time-series voltage data. After the first loss obtained by the unsupervised training meets the first preset error, the model that meets the first preset error is trained in a supervised manner based on the original time-series voltage data and the reference operating status.

[0039] Among them, unsupervised training is used to learn the internal structure and characteristics of reference time series voltage data; supervised training is used to learn the correlation between original time series voltage data and reference operating status based on the internal structure and characteristics of reference time series voltage data.

[0040] Furthermore, the basic model framework of the preset model can adopt the time series Transformer neural network model (TST). The TST model is as follows Figure 3 As shown, it includes input layer, linear projection layer, position encoding layer, Transformer encoding layer and output layer.

[0041] Among them, the input layer is used to receive the original timing voltage data

[0042] Among them, the linear projection layer is used to map the original time series voltage data X to the d-dimensional vector space to obtain u t ,u t It can be expressed by the following formula:

[0043] u t =W in ·x t +b in ,

[0044] in, and is a trainable parameter of the linear projection layer; w represents the time series length of the original time series voltage data, m represents the number of voltage data in the original time series voltage data, It is the voltage data of a single time.

[0045] Among them, the position encoding layer retains the time series order of the original time series voltage data by adding position information to each time step. Position encoding enables the prediction model to use the position of each time step in the time series to capture the dependencies between time steps. The calculation formula of position encoding is as follows:

[0046]

[0047] Among them, PE (Positional Encoding) is the abbreviation of position encoding, which is used to represent the position information of the time step. The position encoding is generated using sine and cosine functions, and each position corresponds to a specific set of sine and cosine values; pos is the position, that is, the time step index; i is the feature dimension index of the input vector generated by the original time series voltage data; d is the vector dimension.

[0048] Among them, the Transformer encoding layer captures global dependencies through a multi-head self-attention mechanism and a feedforward neural network layer, further refining and enhancing feature representation. The multi-head self-attention mechanism calculates multiple attention heads in parallel, with each head calculating the attention in a different subspace to capture the dependencies in different subspaces of the input sequence. The calculation formula is as follows:

[0049]

[0050] Where Q, K, and V are query, key, and value respectively. k is the dimension of the key.

[0051] Furthermore, the Transformer encoding layer also includes: a feed-forward neural network layer (FFN), a residual connection, and a layer normalization. FFN can represent z at each time step. t For further processing:

[0052] FFN(x)=max(0,xW1+b1)W2+b2,

[0053] Among them, W1, W2, b1 and b2 are trainable parameters of FFN.

[0054] Furthermore, the TST model also includes a linear connection layer. The linear connection layer is used to represent the vector z of each time step. t Concatenate into a single vector Z.

[0055] In the above steps, the TST model is used as the basic model framework of the prediction model because the TST model has certain advantages in processing time series features and parallel computing, which can speed up the training speed and is not prone to overfitting problems.

[0056] Among them, the island phenomenon is the phenomenon that an independent energized circuit is formed in the target distribution network system after the target distribution network system is powered off.

[0057] Specifically, the target time-series voltage data is taken as input and input into the prediction model, and the target time-series voltage data is predicted by a preset model to obtain the target operating state of the target distribution network system.

[0058] The above steps use a preset model to effectively capture the long-range dependency and global information in the target time-series voltage data, solve the non-detection area and power quality problems existing in the existing detection methods, improve the accuracy of determining the target operating state, and thus improve the accuracy and reliability of island detection.

[0059] S130. Determine whether an islanding phenomenon occurs in the target distribution network system according to the target operating status.

[0060] Specifically, if the target operating state is a fault state or a normal operating state, the target distribution network system does not have an islanding phenomenon; if the target operating state is an islanding state, the target distribution network system has an islanding phenomenon, and an islanding phenomenon warning is issued.

[0061] Furthermore, if the target operating state is a fault state, a fault warning is performed.

[0062] Exemplarily, an experimental control group is constructed, and the control groups are LSTM (Long Short-Term Memory) model and XGB (eXtreme Gradient Boosting) model, respectively. The acquired target time series voltage data is divided into a clean data set and a noisy data set with 0dB SNR Gaussian white noise added, and are respectively input into the prediction model, LSTM model and XGB model for prediction to obtain the prediction results. The prediction results are: the accuracy of the prediction model on the clean data set is 95.8%, and the accuracy on the noisy data set is 98.3%; the accuracy of the LSTM model on the clean data set is 87%, and the accuracy on the noisy data set is slightly reduced. The accuracy of the XGB model on the clean data set is 99.5%, and the accuracy on the noisy data set is significantly reduced to 84%. Therefore, it can be seen from the prediction results that the prediction model can still maintain a high accuracy rate on the noisy data set.

[0063] From the above steps and the establishment of the control group, it can be seen that compared with other timing models, the prediction model has better robustness and generalization ability, and can maintain the efficiency and accuracy of island detection under different target timing voltage data conditions.

[0064] Optionally, the prediction model construction process includes steps A1-A3:

[0065] Step A1: Determine reference timing voltage data.

[0066] The reference time series voltage data is the data after the original time series voltage data is processed with missing values, noise reduction and standardization. The original time series voltage data is the time series voltage data of the target common coupling point at the historical moment. The target common coupling point is the coupling point of the area where the distributed power source is located.

[0067] Specifically, the original time-series voltage data of the acquired target common coupling point is processed for missing values, noise reduction and standardization to obtain reference time-series voltage data.

[0068] Step A2: using the reference time series voltage data as the input of the time series neural network model, performing unsupervised training on the time series neural network model to obtain a reference model.

[0069] Among them, the time series neural network model is the basic model framework of the prediction model, and the time series neural network model can be a TST model.

[0070] Specifically, the reference timing voltage data is used as the input of the timing neural network model, the timing neural network model is unsupervisedly trained based on the reference timing voltage data, and the weights of the timing neural network model are updated based on the first loss between the obtained reference prediction data and the reference timing voltage data, thereby obtaining a reference model.

[0071] Step A3: Perform supervised training on the reference model according to the original time series voltage data to obtain a prediction model.

[0072] Specifically, the reference timing voltage data is used as the input of the reference model, and the reference operating state corresponding to the reference timing voltage data is used as the output of the reference model. The reference model is trained, and the weight of the reference model is updated according to the second loss between the obtained predicted reference operating state and the reference operating state, thereby obtaining a prediction model.

[0073] Optionally, determining reference timing voltage data includes steps B1-B2:

[0074] Step B1: determining a target common coupling point according to a target distribution network system.

[0075] Specifically, the target common coupling point is determined according to the nodes included in the area connected to the distributed power source in the target distribution network system.

[0076] For example, Figure 2 As shown, the node 634 is connected to a distributed power source, and the node 634 can be the target common coupling point.

[0077] Step B2: determine the original time series voltage data of the target common coupling point, perform missing value, noise reduction and standardization processing on the original time series voltage data, and obtain reference time series voltage data.

[0078] Specifically, the original time series voltage data of the target common coupling point is obtained, the missing values ​​in the original time series voltage data are set to zero, and the high-frequency noise in the data is removed by using a low-pass filter or wavelet denoising method. Then, the outliers are detected and processed by the box plot method. Finally, the Z-score standardization method is used to convert the data into a standard normal distribution with a mean of 0 and a standard deviation of 1, and the standardized data is used as the reference time series voltage data.

[0079] Furthermore, Z-score standardization can be achieved by the following formula:

[0080]

[0081] Wherein, X is the original time series voltage data, μ is the mean of the original time series voltage data, and σ is the standard deviation of the original time series voltage data.

[0082] Optionally, the reference time series voltage data is used as the input of the time series neural network model, and the time series neural network model is unsupervisedly trained to obtain a reference model, including steps C1-C2:

[0083] Step C1, determining reference masking data according to reference timing voltage data.

[0084] Specifically, the reference timing voltage data is masked by a binary noise mask to obtain reference masked data.

[0085] The binary noise mask is a string of binary digits (ie, a digital sequence containing only 0 and 1), which is implemented by performing a bitwise operation on the reference timing voltage data to mask the timing voltage data corresponding to the specified bit.

[0086] In the above steps, the use of binary noise mask can simulate the noise and missing values ​​existing in the original time series voltage data, and can provide a basis for the reference model to learn the internal structure and characteristics of the time series voltage data according to the difference between the reference time series voltage data and the reference masked data.

[0087] Step C2: Use the reference masked data as the input of the timing neural network model, train the timing neural network model to obtain reference prediction data, calculate the first loss based on the reference prediction data and the reference timing voltage data, and correct the timing neural network model through the first loss until the first loss meets the first preset error. Then, use the timing neural network model corresponding to the first loss that meets the requirements as the reference model.

[0088] Specifically, the reference masked data is input into the time series neural network model, and the time series neural network model performs linear projection, position encoding and Transformer encoding layer processing on the reference masked data. The Transformer encoding layer extracts the data features of the reference masked data through the self-attention mechanism and the feedforward neural network layer. The reference prediction data X' is generated according to the data features of the obtained reference masked data, and the first loss is calculated using the mean square error (MSE) loss function based on the reference prediction data and the reference time series voltage data. The first loss can be expressed by the following formula:

[0089]

[0090] Among them, L MSE is the value corresponding to the first loss; x'(t,i) is the predicted value of the i-th reference prediction data at time t, and x(t,i) is the true value of the i-th reference timing voltage data at time t.

[0091] Furthermore, the Adam optimization algorithm is used to update the weights of the time series neural network model according to the first loss until the first loss meets the first preset error, and the time series neural network model corresponding to the first loss that meets the requirement is used as the reference model.

[0092] Furthermore, the reference masked data is input into the temporal neural network model in the form of (b,w,m), where b is the batch size, w represents the time series length, and m represents the number of variables.

[0093] The above steps and unsupervised learning can enable the time series neural network model to learn the internal structure and characteristics of the time series voltage data without labeled data, reduce the computational cost, and maintain high performance even with a limited number of samples.

[0094] Optionally, determining reference masking data according to reference timing voltage data includes steps D1-D2:

[0095] Step D1: Generate a binary noise mask according to reference timing voltage data.

[0096] Specifically, the desired noise ratio and the binary noise mask size are set according to the reference timing voltage data, and the binary noise mask is generated according to the desired noise ratio and the binary noise mask size.

[0097] Furthermore, the binary noise mask generation process is as follows: randomly generate a random number sequence equal to the size of the binary noise mask, compare each number in the random number sequence with the desired noise ratio, set the number less than the desired noise ratio to 1, and set the number greater than or equal to the desired noise ratio to 0, and finally obtain the binary noise mask.

[0098] The expected noise ratio is used to characterize the probability of 1 appearing in the binary noise mask. The larger the expected noise ratio, the more 1s there are in the binary noise mask; the smaller the expected noise ratio, the fewer 1s there are in the binary noise mask.

[0099] Step D2: mask the reference timing voltage data according to the binary noise mask to obtain reference masked data.

[0100] Specifically, a binary noise mask is used to randomly mask the reference timing voltage data to obtain reference masked data.

[0101] Furthermore, the masking process can be: performing a bitwise AND operation on the binary noise mask and the reference timing voltage data, that is, performing an AND operation with 1, then retaining the original timing voltage data of the corresponding bit of the reference timing voltage data; performing an AND operation with 0, then setting the original timing voltage data of the corresponding bit of the reference timing voltage data to 0.

[0102] The above steps can simulate the missing values ​​in the original time series voltage data.

[0103] Furthermore, the masking process may also be: combining the binary noise mask with the noise data to add noise to the reference timing voltage data, that is, 1 indicates adding noise to the reference timing voltage data, and 0 indicates not adding noise to the reference timing voltage data.

[0104] The above steps can simulate the noise of the original timing voltage data.

[0105] Optionally, supervised training is performed on the reference model according to the original time series voltage data to obtain a prediction model, including:

[0106] The original time series voltage data is used as the input of the reference model, and the reference operating state is used as the output. The reference model is trained to obtain the predicted reference operating state. The second loss is calculated based on the predicted reference operating state and the reference operating state. The reference model is corrected by the second loss until the second loss meets the second preset error. The reference model corresponding to the second loss that meets the requirement is used as the prediction model.

[0107] Specifically, the original time series voltage data is input into the reference model, and the reference model performs linear projection, position encoding and Transformer encoding layer processing on the reference mask data. Among them, the Transformer encoding layer extracts the data features of the reference mask data through the self-attention mechanism and the feedforward neural network layer, and the linear connection layer converts the representation vector z of each time step into t Connect into a single vector Z. Generate a predicted reference operating state based on the acquired data features and vector Z, and calculate the second loss using the cross entropy loss function based on the reference predicted data and the reference operating state. The second loss can be expressed by the following formula:

[0108]

[0109] Among them, L represents the value of the second loss, y i is the reference operating state corresponding to the i-th original time series voltage data, y i ' is the predicted reference operating state corresponding to the i-th original time series voltage data.

[0110] Furthermore, the Adam optimization algorithm is used to update the weights of the reference model according to the second loss until the second loss satisfies the second preset error, and the reference model corresponding to the first loss that meets the requirement is used as the preset model.

[0111] Furthermore, the original time series voltage data is input into the reference model in the form of (b,w,m), where b is the batch size, w represents the time series length, and m represents the number of variables.

[0112] Furthermore, before the original time-series voltage data is input into the reference model, a corresponding relationship between the original time-series voltage data and the reference state is established.

[0113] Furthermore, the corresponding relationship is established by manual labeling based on experience and automatic labeling based on rules, that is, the operating state corresponding to each time series voltage data in the original time series voltage data is labeled, such as normal operation, island state, fault state, etc.

[0114] The above steps are supervised training, and the prediction model obtained by training can accurately detect the island state of the target distribution network system. That is, the prediction model can directly determine the operating state of the target distribution network system based on the time series voltage data collected from the target distribution network system, without the need to pre-process the time series voltage data, which improves the accuracy of island detection while avoiding complex Fourier transform or wavelet transform steps, simplifying the model construction and training process.

[0115] The technical solution of this embodiment determines the target time-series voltage data of the target distribution network system; determines the target operating state of the target distribution network system through a prediction model based on the target time-series voltage data, and the prediction model can effectively capture the long-range dependency and global information in the target time-series voltage data, solves the problems of the appearance of non-detection areas and the quality of power information, thereby improving the accuracy of determining the target operating state; determines whether an island phenomenon occurs in the target distribution network system based on the target operating state, thereby improving the accuracy of island detection. This method determines the target operating state based on the target time-series voltage data through a prediction model, and determines whether an island phenomenon occurs in the target distribution network based on the target operating state, thereby avoiding the process of preprocessing the target time-series voltage data, simplifying the prediction process, and improving the efficiency of detection. At the same time, it also solves the problems of non-detection areas and power quality in existing detection methods, thereby improving the accuracy and reliability of island detection.

[0116] Figure 4 The schematic diagram of the structure of a non-intrusive anti-islanding detection device provided by an embodiment of the present invention. This embodiment can be applied to the case where an islanding phenomenon occurs inside a power distribution network when the power distribution network is powered off. The non-intrusive anti-islanding detection device can be implemented in the form of hardware and / or software. The non-intrusive anti-islanding detection device can be configured in any electronic device with network communication function. Figure 4 As shown, the device includes: a target time-series voltage data determination module 210, a target operation state determination module 220 and an islanding phenomenon determination module 230, wherein:

[0117] Target time-sequence voltage data determination module 210: used to determine target time-sequence voltage data of a target distribution network system; the target time-sequence voltage data is used to characterize the voltage change of a target common coupling point in a target distribution network system connected to a distributed power source;

[0118] Target operating state determination module 220: used to determine the target operating state of the target distribution network system through a prediction model according to the target time series voltage data; the target operating state is used to characterize whether the target distribution network system has an islanding phenomenon; the prediction model is used to characterize the relationship between the time series voltage data and the operating state of the target distribution network system; the prediction model is trained unsupervisedly using the reference time series voltage data, and after the first loss obtained by the unsupervised training satisfies the first preset error, the model that satisfies the first preset error is trained supervisedly according to the original time series voltage data and the reference operating state; the unsupervised training is used to learn the internal structure and characteristics of the reference time series voltage data; the supervised training is used to learn the correlation between the original time series voltage data and the reference operating state based on the internal structure and characteristics of the reference time series voltage data; the islanding phenomenon is the phenomenon that an independent energized circuit is formed in the target distribution network system after the target distribution network system is powered off;

[0119] The islanding phenomenon determination module 230 is used to determine whether an islanding phenomenon occurs in the target distribution network system according to the target operating state.

[0120] Optionally, the target operating state determination module 220 includes:

[0121] A reference time series voltage data determination unit is used to determine reference time series voltage data, where the reference time series voltage data is data after missing values, noise reduction and standardization processing are performed on the original time series voltage data, and the original time series voltage data is the time series voltage data of the target common coupling point at a historical moment, and the target common coupling point is a coupling point in the area where the distributed power source is located;

[0122] An unsupervised training unit is used to use the reference time series voltage data as the input of the time series neural network model, perform unsupervised training on the time series neural network model, and obtain a reference model;

[0123] The supervised training unit is used to perform supervised training on the reference model according to the original time series voltage data to obtain a prediction model.

[0124] Optionally, the reference timing voltage data determination unit includes:

[0125] Target common coupling point determination subunit: used to determine the target common coupling point according to the target distribution network system;

[0126] The reference time series voltage data determination subunit is used to determine the original time series voltage data of the target common coupling point, perform missing value, noise reduction and standardization processing on the original time series voltage data, and obtain the reference time series voltage data.

[0127] Optional, unsupervised training units, including:

[0128] Reference mask data determination subunit: used to determine reference mask data according to reference timing voltage data;

[0129] Unsupervised training subunit: used to take the reference masked data as the input of the time series neural network model, train the time series neural network model, obtain reference prediction data, calculate the first loss according to the reference prediction data and the reference time series voltage data, and correct the time series neural network model by the first loss until the first loss meets the first preset error, and then use the time series neural network model corresponding to the first loss that meets the requirements as the reference model.

[0130] Optionally, the subunit is determined with reference to the mask data, specifically for:

[0131] generating a binary noise mask based on reference timing voltage data;

[0132] The reference timing voltage data is masked according to the binary noise mask to obtain reference masked data.

[0133] Optional, supervised training unit, specifically for:

[0134] The original time series voltage data is used as the input of the reference model, and the reference operating state is used as the output. The reference model is trained to obtain the predicted reference operating state. The second loss is calculated based on the predicted reference operating state and the reference operating state. The reference model is corrected by the second loss until the second loss meets the second preset error. The reference model corresponding to the second loss that meets the requirement is used as the prediction model.

[0135] The non-invasive anti-islanding detection device provided in the embodiment of the present invention can execute the non-invasive anti-islanding detection method provided in any embodiment of the present invention mentioned above, and has the corresponding functions and beneficial effects of executing the non-invasive anti-islanding detection method. For detailed process, please refer to the relevant operations of the non-invasive anti-islanding detection method in the aforementioned embodiment.

[0136] Figure 5A schematic diagram of the structure of an electronic device for implementing the non-invasive anti-islanding detection method of an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0137] like Figure 5 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. 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.

[0138] A number of 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, speakers, etc.; a storage unit 18, such as a 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.

[0139] The processor 11 may be a variety of general and / or dedicated 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 dedicated 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 executes the various methods and processes described above, such as the non-intrusive anti-islanding detection method.

[0140] In some embodiments, the non-intrusive anti-islanding detection method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on 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 non-intrusive anti-islanding detection method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the non-intrusive anti-islanding detection method in any other appropriate manner (e.g., by means of firmware).

[0141] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0142] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0143] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0144] To provide interaction with a user, the systems and techniques described herein may 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 trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0145] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may 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.

[0146] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may 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 difficult management and weak business scalability in traditional physical hosts and VPS services.

[0147] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0148] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A non-invasive anti-islanding detection method, characterized in that: include: Determine target time-series voltage data of a target distribution network system; The target time-series voltage data is used to characterize the voltage change of the target common coupling point in the target distribution network system connected to the distributed power source; Determine the target operating state of the target distribution network system through a prediction model according to the target time series voltage data; The target operating state is used to characterize whether an islanding phenomenon occurs in the target distribution network system; The prediction model is used to characterize the relationship between the time series voltage data and the operating status of the target distribution network system; The prediction model is trained unsupervisedly using reference time series voltage data. After the first loss obtained by the unsupervised training satisfies the first preset error, supervised training is performed on the model that satisfies the first preset error based on the original time series voltage data and the reference operating state; the unsupervised training is used to learn the internal structure and characteristics of the reference time series voltage data; the supervised training is used to learn the correlation between the original time series voltage data and the reference operating state based on the internal structure and characteristics of the reference time series voltage data; the island phenomenon is the phenomenon that an independent energized circuit is formed in the target distribution network system after the target distribution network system is powered off; Whether an islanding phenomenon occurs in the target distribution network system is determined according to the target operating state.

2. The method according to claim 1, characterized in that: The process of constructing the prediction model includes: Determine reference time series voltage data, where the reference time series voltage data is data after missing value, noise reduction and standardization processing is performed on the original time series voltage data, and the original time series voltage data is the time series voltage data of the target common coupling point at a historical moment, and the target common coupling point is the coupling point of the area where the distributed power source is located; Using the reference time series voltage data as input of a time series neural network model, performing unsupervised training on the time series neural network model to obtain a reference model; The reference model is supervisedly trained according to the original time series voltage data to obtain the prediction model.

3. The method according to claim 2, characterized in that The step of determining reference timing voltage data comprises: Determine a target common coupling point according to a target distribution network system; The original time-series voltage data of the target common coupling point is determined, and the original time-series voltage data is processed for missing values, noise reduction, and standardization to obtain the reference time-series voltage data.

4. The method according to claim 2, characterized in that: The reference time series voltage data is used as the input of the time series neural network model, and the time series neural network model is trained unsupervised to obtain a reference model, including: Determine reference masking data according to the reference timing voltage data; The reference masked data is used as the input of the timing neural network model, the timing neural network model is trained to obtain reference prediction data, the first loss is calculated according to the reference prediction data and the reference timing voltage data, the timing neural network model is corrected by the first loss until the first loss meets the first preset error, and the timing neural network model corresponding to the first loss that meets the requirement is used as the reference model.

5. The method according to claim 4, characterized in that The step of determining reference masking data according to the reference timing voltage data comprises: generating a binary noise mask according to the reference timing voltage data; The reference timing voltage data is masked according to the binary noise mask to obtain reference masked data.

6. The method according to claim 2, characterized in that The step of performing supervised training on the reference model according to the original time series voltage data to obtain the prediction model comprises: The original time series voltage data is used as the input of the reference model, and the reference operating state is used as the output. The reference model is trained to obtain the predicted reference operating state. The second loss is calculated according to the predicted reference operating state and the reference operating state. The reference model is corrected by the second loss until the second loss meets the second preset error. The reference model corresponding to the second loss that meets the requirement is used as the prediction model.

7. A non-invasive anti-islanding detection device, characterized in that: include: A target time-series voltage data determination module is used to determine target time-series voltage data of a target distribution network system; The target time-series voltage data is used to characterize the voltage change of the target common coupling point in the target distribution network system connected to the distributed power source; A target operating state determination module, used to determine the target operating state of the target distribution network system through a prediction model according to the target time series voltage data; The target operating state is used to characterize whether an islanding phenomenon occurs in the target distribution network system; The prediction model is used to characterize the relationship between the time series voltage data and the operating status of the target distribution network system; The prediction model is trained unsupervisedly using reference time series voltage data. After the first loss obtained by the unsupervised training satisfies the first preset error, supervised training is performed on the model that satisfies the first preset error based on the original time series voltage data and the reference operating state; the unsupervised training is used to learn the internal structure and characteristics of the reference time series voltage data; the supervised training is used to learn the correlation between the original time series voltage data and the reference operating state based on the internal structure and characteristics of the reference time series voltage data; the island phenomenon is the phenomenon that an independent energized circuit is formed in the target distribution network system after the target distribution network system is powered off; The islanding phenomenon determination module is used to determine whether the target distribution network system has an islanding phenomenon according to the target operating state.

8. The device according to claim 7, characterized in that The target operating state determination module includes: A reference time series voltage data determination unit is used to determine reference time series voltage data, wherein the reference time series voltage data is data after missing value, noise reduction and standardization processing is performed on the original time series voltage data, wherein the original time series voltage data is the time series voltage data of the target common coupling point at a historical moment, and the target common coupling point is a coupling point in the area where the distributed power source is located; An unsupervised training unit, used to use the reference time series voltage data as an input of a time series neural network model, perform unsupervised training on the time series neural network model, and obtain a reference model; A supervised training unit is used to perform supervised training on the reference model according to the original time series voltage data to obtain the prediction model.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, 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 so that the at least one processor can perform the non-intrusive anti-islanding detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the non-intrusive anti-islanding detection method according to any one of claims 1 to 7 when executed.

Images