Method, device and equipment for identifying and positioning power flow section of power system and medium

Abstract

The invention relates to the technical field of power systems, and discloses a power system power flow section recognition and positioning method, device and equipment and a medium, and the method comprises the steps: collecting real-time power flow data in a power system; preprocessing the real-time power flow data to obtain power system node data; inputting the node data of the power system into an improved convolutional neural network module to extract spatial features of the power system; inputting the power system node data into a long short-term memory network module to capture power system time sequence characteristics; fusing the spatial characteristics of the power system and the time sequence characteristics of the power system, outputting the identification probability of the power flow section, and determining a target power flow section; and optimizing a power flow path of the target power flow section through the real-time power flow data of the target power flow section, and determining the position of the target power flow section. By applying the technical scheme of the invention, automatic identification from the power flow data to the power flow section can be realized, and the real-time performance and the accuracy of the system are obviously improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of power systems, and specifically to a method, device, equipment and computer-readable storage medium for identifying and locating power system flow sections. Background Art

[0002] Traditional power flow section identification methods mostly rely on power flow calculation and expert systems, which have the problems of slow calculation speed and low efficiency. Especially in large-scale power grids, the identification process needs to consider multiple parameters, which increases the calculation complexity and makes it impossible to identify key power flow sections in real time, seriously affecting the real-time dispatching capability of the power system.

[0003] Existing power flow section identification methods are usually based on static rules or linear models, which are difficult to accurately capture the nonlinear characteristics of the power system, resulting in low identification accuracy. Especially in the context of new energy access, the operating characteristics of the power system are more complex, and traditional methods are difficult to cope with the rapidly changing power flow distribution. The identified power flow section cannot fully reflect the dynamic changes of the system.

[0004] With the addition of new energy generation, distributed power sources and other factors, the operation status of the power system has become more dynamic and complex. Most existing identification methods rely on pre-set rules or analysis models based on historical data, which cannot cope with the rapid changes in grid operating conditions, lack adaptive capabilities, and are difficult to ensure the effectiveness and accuracy of identification in complex scenarios. Summary of the invention

[0005] In view of the above problems, an embodiment of the present invention provides a method, device, equipment and computer-readable storage medium for identifying and locating power system flow sections, which are used to solve the problem in the prior art that the identified flow sections cannot fully reflect the dynamic changes of the system for rapidly changing flow distribution.

[0006] According to one aspect of an embodiment of the present invention, a method for identifying and locating a power system power flow section is provided, the method comprising: collecting real-time power flow data in the power system; preprocessing the real-time power flow data to obtain power system node data;

[0007] Inputting the power system node data into an improved convolutional neural network module to extract the power system spatial features;

[0008] Inputting the power system node data into a long short-term memory network module to capture the power system time series characteristics;

[0009] The spatial characteristics of the power system and the temporal characteristics of the power system are integrated to output the recognition probability of the power flow section and determine the target power flow section;

[0010] The tidal path of the target tidal section is optimized and the position of the target tidal section is determined by using the real-time tidal data of the target tidal section.

[0011] Specifically, the present invention introduces a deep learning model that combines an improved convolutional neural network and a long short-term memory network to perform feature extraction and time series analysis on the power system flow data, and can quickly and accurately identify key flow sections in complex power systems. The present invention constructs an end-to-end model architecture to achieve automatic identification from flow data to sections, significantly improving the real-time performance and accuracy of the system. In addition, the fast positioning algorithm can accurately locate the position of the flow section and reduce recognition errors, thereby providing a more accurate reference for power dispatching. While improving the system safety and stability, the present invention improves the system dispatching efficiency, and is particularly suitable for complex and changeable new energy environments.

[0012] In an optional manner, preprocessing the real-time power flow data to obtain power system node data includes:

[0013] According to the node structure of the power system, the real-time power flow data is preprocessed to obtain the power system node data;

[0014] Wherein, the preprocessing includes: data standardization and denoising;

[0015] The real-time power flow data includes: voltage, active power, reactive power, power flow direction and power flow amount.

[0016] In an optional manner, the improved convolutional neural network module includes: a plurality of sequentially connected convolutional layers and a first fully connected layer; each of the convolutional layers is connected by a pooling layer; each of the convolutional layers includes a residual network and a self-attention mechanism module;

[0017] Each of the convolutional layers is used to calculate the features of the input data using the corresponding residual network and output the features of the preset dimensions; the features of the preset dimensions are calculated using the self-attention mechanism module and the enhanced features are output;

[0018] Among them, the input data of the first convolution layer is the power system node data, and the input data of the non-first convolution layer is the output of the pooling layer corresponding to the previous convolution layer;

[0019] The pooling layer is used to perform a maximum pooling operation on the output of the corresponding self-attention mechanism module and output a feature vector;

[0020] The first fully connected layer is used to take the output of each pooling layer as input and output the spatial characteristics of the power system.

[0021] Specifically, the convolution operation of each layer not only performs ordinary convolution operations, but also uses residual connections in the residual network (ResNet), which helps to alleviate the gradient vanishing problem in deep networks, especially in deep networks. The self-attention mechanism is added to each convolution layer, so that the model can model stronger feature correlations between different nodes. This can enhance the model's ability to capture key nodes and edge information, especially when faced with complex features of tidal sections, and can adaptively focus on the most relevant features.

[0022] In an optional manner, the power system node data is input into an improved convolutional neural network module to extract the power system spatial features, including:

[0023] The power system node data is input into the improved convolutional neural network module, and each convolution layer and each pooling layer network is used in turn to calculate the power system node data to obtain the feature vector of each layer; the feature vector of each layer is input into the fully connected layer to output the spatial characteristics of the power system.

[0024] Specifically, the improved convolutional neural network module of the present invention extracts spatial features through multi-layer convolution and residual connection, and enhances the training depth of the model. The present invention adds a self-attention mechanism to the convolution layer, so that the model can automatically pay attention to the key nodes and their relationships in the power system. After the improved convolutional neural network module of the present invention passes through several convolutional layers and pooling layers, the extracted spatial features are converted into vector representations through a fully connected layer, ready to be input into the subsequent timing analysis module for further identification of key flow sections, thereby improving the recognition accuracy of key flow sections.

[0025] In an optional manner, the power system node data is input into a long short-term memory network module to capture the power system time series characteristics, including:

[0026] Use the long short-term memory network module to perform time series calculation on the power system node data and output the time series characteristics of the power system;

[0027] The long short-term memory network module uses memory units and gating mechanisms to perform time series calculations on input data.

[0028] Specifically, the present invention captures the spatial characteristics of the power system while using the long short-term memory network to capture the temporal dynamic changes of the power system, effectively responds to the dynamic changes of the system, accurately locates the position of the flow section, and reduces recognition errors.

[0029] In an optional manner, the spatial characteristics of the power system and the temporal characteristics of the power system are integrated to output the recognition probability of the power flow section and determine the target power flow section, including:

[0030] The spatial features and temporal features of the power system are input into the second fully connected layer for feature fusion, and the recognition probability of each power flow section is output;

[0031] The flow section whose recognition probability exceeds a preset threshold is determined as the target flow section.

[0032] Specifically, the present invention simultaneously captures the spatial and temporal characteristics of the power system, effectively responds to the dynamic changes of the system, thereby accurately locating the position of the power flow section, reducing identification errors, and providing a more accurate reference basis for power dispatching.

[0033] In an optional manner, optimizing the flow path of the target flow section and determining the position of the target flow section by using the real-time flow data of the target flow section includes:

[0034] The objective function is set based on the estimated value of the target flow section and the label value of the actual flow section;

[0035] The objective function is iteratively optimized, and the tidal current path corresponding to the target tidal current section is adjusted until a preset optimization condition is reached to obtain the position of the target tidal current section.

[0036] Specifically, the fast positioning algorithm can accurately locate the position of the tidal section and reduce the identification error.

[0037] According to another aspect of an embodiment of the present invention, there is provided a device for identifying and locating a power system power flow section, comprising: a collection module, a processing module, an identification module and a positioning module;

[0038] The acquisition module is used to collect real-time power flow data in the power system; pre-process the real-time power flow data to obtain power system node data;

[0039] A processing module is used to input the power system node data into an improved convolutional neural network module to extract the power system spatial features; input the power system node data into a long short-term memory network module to capture the power system time series features;

[0040] An identification module, used to integrate the spatial characteristics of the power system and the temporal characteristics of the power system, output the identification probability of the power flow section, and determine the target power flow section;

[0041] The positioning module is used to optimize the tidal path of the target tidal section and determine the position of the target tidal section through the real-time tidal data of the target tidal section.

[0042] According to another aspect of an embodiment of the present invention, there is provided a device for identifying and locating a power system flow section, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; the memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform the operation of the power system flow section identification and locating method as described in any one of the above items.

[0043] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores at least one executable instruction, wherein the executable instruction enables an electric power system flow section identification and positioning device / apparatus to perform the operation of an electric power system flow section identification and positioning method as described in any one of the above.

[0044] The above description is only an overview of the technical solution of the embodiment of the present invention. In order to more clearly understand the technical means of the embodiment of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiment of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are only used to illustrate the embodiments and are not to be considered as limiting the present invention. In addition, the same reference symbols are used to represent the same components throughout the accompanying drawings. In the accompanying drawings:

[0046] Figure 1 A schematic diagram showing a flow chart of an embodiment of a method for identifying and locating a power system power flow section provided by the present invention;

[0047] Figure 2 A schematic diagram showing the structure of an embodiment of a power system power flow section identification and positioning device provided by the present invention;

[0048] Figure 3 A schematic structural diagram of an embodiment of a power system power flow section identification and positioning device provided by the present invention is shown. DETAILED DESCRIPTION

[0049] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0050] Figure 1 The flowchart of an embodiment of the method for identifying and locating a power system flow section of the present invention is shown, and the method is executed by a power system flow section identification and locating device. Figure 1 As shown, the method comprises the following steps:

[0051] Step 110: Collect real-time power flow data in the power system, pre-process the real-time power flow data, and obtain power system node data.

[0052] In an optional manner, preprocessing the real-time power flow data to obtain power system node data includes:

[0053] According to the node structure of the power system, the real-time power flow data is preprocessed to obtain the power system node data;

[0054] Wherein, the preprocessing includes: data standardization and denoising;

[0055] The real-time power flow data includes: voltage V, active power P, reactive power Q, power flow direction and power flow amount.

[0056] Specifically, the real-time power flow data collected from the power system includes but is not limited to: voltage, active power, reactive power, power flow direction and power flow amount.

[0057] Assuming that there are n nodes in the power system, the data set of real-time power flow data of each node in the power system can be expressed as Where: P i represents the active power of node i; Q i represents the reactive power of node i; V i represents the voltage of node i; F i Indicates the flow direction and size of node i.

[0058] Since the original data may contain noise or outliers, the data needs to be preprocessed before entering the deep learning model, including data standardization and denoising. The standardized data set is denoted as Represents the standardized node features, which are used for subsequent model training and recognition.

[0059] Step 120: Input the power system node data into the improved convolutional neural network module to extract the power system spatial features.

[0060] In an optional manner, the improved convolutional neural network module includes: a plurality of sequentially connected convolutional layers and a first fully connected layer; each of the convolutional layers is connected by a pooling layer; each of the convolutional layers includes a residual network and a self-attention mechanism module;

[0061] Each of the convolutional layers is used to calculate the features of the input data using the corresponding residual network and output the features of the preset dimensions; the features of the preset dimensions are calculated using the self-attention mechanism module and the enhanced features are output;

[0062] Among them, the input data of the first convolution layer is the power system node data, and the input data of the non-first convolution layer is the output of the pooling layer corresponding to the previous convolution layer;

[0063] The pooling layer is used to perform a maximum pooling operation on the output of the corresponding self-attention mechanism module and output a feature vector;

[0064] The first fully connected layer is used to take the output of each pooling layer as input and output the spatial characteristics of the power system.

[0065] Specifically, the improved convolutional neural network module also includes an input layer; the standardized data set D′ is input into the input layer, and the output of the convolutional layer connected by the residual network can be expressed as:

[0066] f Res =σ(W*D′+b)+D′;

[0067] Where W is the convolution kernel, b is the bias term, σ(·) is the activation function, * represents the convolution operation, and D′ is the input data. The residual connection plays a role in the layer-by-layer transmission of features in the model, which helps to alleviate the gradient vanishing problem in deep networks, especially in deep networks.

[0068] In the improved convolutional neural network, a self-attention mechanism is added after each convolution layer, so that the model can model stronger feature correlations between different nodes. This can enhance the model's ability to capture key nodes and edge information, especially when faced with complex features of tidal sections, and can adaptively focus on the most relevant features.

[0069] Specifically, the self-attention mechanism is calculated as follows:

[0070]

[0071] Where A is the computational output of the self-attention mechanism. Q, K, and V are the matrices of query, key, and value generated from the input features, respectively. k is the feature dimension. Through the self-attention mechanism, the model can capture the long-distance dependencies between nodes in the power system.

[0072] The parameters of the input feature generation query Q, key K and value matrix V mainly come from the output after each layer of convolution operation, that is, the output feature map of the convolution layer.

[0073] After adding the self-attention mechanism, the output feature map of the convolutional layer will be linearly transformed to generate query Q, key K and value V matrices. Each matrix is ​​generated from the convolution output features through an independent linear transformation. This process can generally be expressed as:

[0074] Q=W q ·X;

[0075] K=W k ·X;

[0076] V=W v ·X;

[0077] Where: X is the feature map output by the convolutional layer (i.e., the output of the convolutional layer). q , W k and W v are trainable weight matrices that generate query, key, and value matrices, respectively, which are usually simple linear transformations, i.e. a fully connected layer.

[0078] After adding the self-attention mechanism to the convolutional layer, the specific process of performing multi-layer convolution operations is to closely combine the output of the attention mechanism with the subsequent convolutional layer, thereby utilizing the correlation information between space and nodes to improve the feature extraction effect.

[0079] In an optional manner, the power system node data is input into an improved convolutional neural network module to extract the power system spatial features, including:

[0080] The power system node data is input into the improved convolutional neural network module, and each convolution layer and each pooling layer network is used in turn to calculate the power system node data to obtain the feature vector of each layer; the feature vector of each layer is input into the fully connected layer to output the spatial characteristics of the power system.

[0081] Specifically, the power system node data is input into the improved convolutional neural network module, and after passing through each convolution layer, the output feature map of the layer is obtained.

[0082] Assume that the output feature of the convolutional layer is X, and its dimensions are usually (N, C, H, W), where N is the batch size. C is the number of channels in the feature map. H and W are the spatial dimensions of the feature map, i.e., the height and width of the feature map. The output feature X of the convolutional layer is the input of the self-attention mechanism.

[0083] After each convolutional layer, a self-attention mechanism module is used to capture the spatial and inter-node correlations. The specific steps are as follows:

[0084] The output X from the convolutional layer is transformed linearly to generate the query Q, key K, and value V matrices:

[0085] Q=W q X, K = W k X, V = W v ·X;

[0086] Among them, W q , W k and W vare trainable linear transformation weights used to generate query Q, key K, and value V matrices respectively.

[0087] Calculate the dot product between the query Q and the key K to obtain the attention score matrix:

[0088]

[0089] Among them, QK T The dot product between the query and the key is calculated, reflecting the correlation between different nodes or features. k is the scaling factor, used to prevent the dot product value from being too large, usually d k = C. The softmax function is used to normalize these relevance scores so that they can represent the importance of each position to the rest of the positions. The final attention mechanism’s calculated output Z is the weighted sum of these weight pair values ​​V.

[0090] The computational output Z of the attention mechanism is an enhanced feature map, in which the dependencies between spaces and nodes have been modeled by the self-attention mechanism:

[0091] Z = Attention(Q,K,V).

[0092] After obtaining the calculated output Z of the attention mechanism, it is combined with the original output of the convolutional layer, usually by addition or concatenation. The specific method depends on the model design. The addition method is to add the attention output to the convolutional layer output, which means that the original features are enhanced by the attention mechanism, expressed as:

[0093] Y=X+Z.

[0094] Among them, X is the output of the convolutional layer, and Z is the feature after the attention mechanism.

[0095] The concatenation method is to concatenate the original convolutional layer output and the attention output in the channel dimension to increase the feature representation capability, which is expressed as:

[0096] Y=concat(X,Z).

[0097] The enhanced feature Y output by the self-attention mechanism module is used as the input of the next convolution layer, and the next convolution layer continues to extract higher-level features. This process can be repeated many times to form multiple layers of convolution operations. After each layer of convolution, the model will recalculate the attention mechanism to further enhance the feature correlation between different nodes.

[0098] After each convolutional layer, a maximum pooling operation is used to reduce the size of the feature map while retaining important spatial features. Assume that the features after pooling are:

[0099] f pool =MaxPool(f Res );

[0100] After L layers of convolution and pooling operations, the feature vector is finally obtained:

[0101] f CNN-att =Attention(f pool ).

[0102] After several convolutional layers and pooling layers, the extracted spatial features are converted into vector representations through a fully connected layer to obtain the spatial features of the power system.

[0103] Step 130: Input the power system node data into the long short-term memory network module to capture the power system time series characteristics.

[0104] In an optional manner, the power system node data is input into a long short-term memory network module to capture the power system time series characteristics, including:

[0105] Use the long short-term memory network module to perform time series calculation on the power system node data and output the time series characteristics of the power system;

[0106] The long short-term memory network module uses memory units and gating mechanisms to perform time series calculations on input data.

[0107] Specifically, the expression of the long short-term memory network module is:

[0108] f LSTM =LSTM(f CNN );

[0109] Among them, f LSTM Represents the temporal characteristics of the output of the long short-term memory network module.

[0110] The long short-term memory network module models the time series changes of the power system by introducing memory units and gate mechanisms. The input data includes the active power, reactive power, voltage and power flow characteristics of each node. The forget gate of the long short-term memory network module determines whether to retain or forget the past memory state, the input gate introduces new time series information, and the output gate generates the final output characteristics in combination with the current memory state. This structure can effectively capture the long-term dependence and short-term dynamic changes of power flow and power in the power system, gradually update the memory state, and thus accurately model the time series characteristics of the system. The final output time series characteristics are used for subsequent power flow section identification and decision-making.

[0111] Step 140: integrating the spatial characteristics of the power system and the temporal characteristics of the power system, outputting the recognition probability of the power flow section, and determining the target power flow section.

[0112] In an optional manner, the spatial characteristics of the power system and the temporal characteristics of the power system are integrated to output the recognition probability of the power flow section and determine the target power flow section, including:

[0113] The spatial features and temporal features of the power system are input into the second fully connected layer for feature fusion, and the recognition probability of each power flow section is output;

[0114] The flow section whose recognition probability exceeds a preset threshold is determined as the target flow section.

[0115] Specifically, the extracted spatial features and temporal features are fused through the fully connected layer, the recognition probability of each power flow section is output, and the threshold T is set to determine whether it is a key power flow section:

[0116]

[0117] Among them, σ(·) represents the Sigmoid function, represents the recognition result of the key power flow section, W and b are the weight and bias of the fully connected layer.

[0118] Exemplarily, after the spatial features are fused with the temporal features, the fused feature vector is processed by a fully connected layer, the role of which is to map the high-dimensional feature vector to a probability space. Specifically, through the linear transformation of the fully connected layer, combined with an activation function such as the Sigmoid function, an identification probability between 0 and 1 is output, and this probability value indicates the possibility that a certain flow section is a critical flow section. By setting a threshold, such as T = 0.5, the model will determine whether a node or flow section is critical based on the identification probability. If the identification probability exceeds the threshold, the section is considered to be a critical flow section, indicating that it plays an important role in stability and security in the system. The identification result mainly reflects whether there is a critical flow section.

[0119] Step 150: Optimize the tidal path of the target tidal section and determine the position of the target tidal section using the real-time tidal data of the target tidal section.

[0120] In an optional manner, optimizing the flow path of the target flow section and determining the position of the target flow section by using the real-time flow data of the target flow section includes:

[0121] The objective function is set based on the estimated value of the target flow section and the label value of the actual flow section;

[0122] The objective function is iteratively optimized, and the tidal current path corresponding to the target tidal current section is adjusted until a preset optimization condition is reached to obtain the position of the target tidal current section.

[0123] After identifying the key tidal section, the present invention further designs a fast positioning algorithm to locate the specific location of the tidal section. Through the real-time data feedback of the tidal section, the tidal path is optimized by the gradient descent method, and the location of the tidal section is finally determined. Assuming that the position vector of the tidal section is P, by minimizing the objective function:

[0124]

[0125] Among them, y i is the actual section label value, is an estimate, and the precise location of the section is obtained through optimization. Using fast positioning algorithms and real-time data feedback, the optimization objective function is set (such as minimizing the difference between the identified section and the true label), and the tidal path is adjusted through iterative optimization to infer the specific section location. Finally, the accuracy of the position is verified through decision criteria to achieve accurate positioning of key tidal sections.

[0126] By introducing a deep learning model, the present invention realizes the rapid and efficient identification and positioning of key power flow sections in the power system, significantly improving the real-time performance and accuracy of the system. Compared with traditional methods, the architecture combining improved CNN and LSTM can simultaneously capture the spatial and temporal characteristics of the power system and effectively respond to the dynamic changes of the system. In addition, the fast positioning algorithm can accurately locate the position of the power flow section and reduce the recognition error, thereby providing a more accurate reference for power dispatching. Finally, the present invention improves the dispatching efficiency of the system while ensuring the safety and stability of the system, and is particularly suitable for complex and changeable new energy environments.

[0127] Figure 2 FIG. 1 is a schematic diagram showing a structure of an embodiment of a power system flow section identification and positioning device according to the present invention; Figure 2 As shown, the device 200 includes: a collection module 210, a processing module 220, an identification module 230 and a positioning module 240;

[0128] The acquisition module 210 is used to acquire real-time power flow data in the power system; pre-process the real-time power flow data to obtain power system node data;

[0129] A processing module 220, configured to input the power system node data into an improved convolutional neural network module to extract power system spatial features;

[0130] Inputting the power system node data into a long short-term memory network module to capture the power system time series characteristics;

[0131] The identification module 230 is used to integrate the spatial characteristics of the power system and the temporal characteristics of the power system, output the identification probability of the power flow section, and determine the target power flow section;

[0132] The positioning module 240 is used to optimize the tidal path of the target tidal section and determine the position of the target tidal section through the real-time tidal data of the target tidal section.

[0133] In an optional manner, preprocessing the real-time power flow data to obtain power system node data includes:

[0134] According to the node structure of the power system, the real-time power flow data is preprocessed to obtain the power system node data;

[0135] Wherein, the preprocessing includes: data standardization and denoising;

[0136] The real-time power flow data includes: voltage, active power, reactive power, power flow direction and power flow amount.

[0137] In an optional manner, the improved convolutional neural network module includes: a plurality of sequentially connected convolutional layers and a first fully connected layer; each of the convolutional layers is connected by a pooling layer; each of the convolutional layers includes a residual network and a self-attention mechanism module;

[0138] Each of the convolutional layers is used to calculate the features of the input data using the corresponding residual network and output the features of the preset dimensions; the features of the preset dimensions are calculated using the self-attention mechanism module and the enhanced features are output;

[0139] Among them, the input data of the first convolution layer is the power system node data, and the input data of the non-first convolution layer is the output of the pooling layer corresponding to the previous convolution layer;

[0140] The pooling layer is used to perform a maximum pooling operation on the output of the corresponding self-attention mechanism module and output a feature vector;

[0141] The first fully connected layer is used to take the output of each pooling layer as input and output the spatial characteristics of the power system.

[0142] In an optional manner, the power system node data is input into an improved convolutional neural network module to extract the power system spatial features, including:

[0143] The power system node data is input into the improved convolutional neural network module, and each convolution layer and each pooling layer network is used in turn to calculate the power system node data to obtain the feature vector of each layer; the feature vector of each layer is input into the fully connected layer to output the spatial characteristics of the power system.

[0144] In an optional manner, the power system node data is input into a long short-term memory network module to capture the power system time series characteristics, including:

[0145] Use the long short-term memory network module to perform time series calculation on the power system node data and output the time series characteristics of the power system;

[0146] The long short-term memory network module uses memory units and gating mechanisms to perform time series calculations on input data.

[0147] In an optional manner, the spatial characteristics of the power system and the temporal characteristics of the power system are integrated to output the recognition probability of the power flow section and determine the target power flow section, including:

[0148] The spatial features and temporal features of the power system are input into the second fully connected layer for feature fusion, and the recognition probability of each power flow section is output;

[0149] The flow section whose recognition probability exceeds a preset threshold is determined as the target flow section.

[0150] In an optional manner, optimizing the flow path of the target flow section and determining the position of the target flow section by using the real-time flow data of the target flow section includes:

[0151] The objective function is set based on the estimated value of the target flow section and the label value of the actual flow section;

[0152] The objective function is iteratively optimized, and the tidal current path corresponding to the target tidal current section is adjusted until a preset optimization condition is reached to obtain the position of the target tidal current section.

[0153] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0154] Figure 3 The schematic diagram of the structure of an embodiment of the power system power flow section identification and positioning device of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the device.

[0155] like Figure 3 As shown, the power system power flow section identification and positioning device may include: a processor (processor) 302 , a communication interface (Communications Interface) 304 , a memory (memory) 306 , and a communication bus 308 .

[0156] The processor 302, the communication interface 304, and the memory 306 communicate with each other via the communication bus 308. The communication interface 304 is used to communicate with other devices such as a client or other server network elements. The processor 302 is used to execute the program 310, which can specifically execute the relevant steps in the above-mentioned embodiment of the method for identifying and locating the power system power flow section.

[0157] Specifically, the program 310 may include program code including computer executable instructions.

[0158] The processor 302 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiment of the present invention. The one or more processors included in the power system flow section identification and positioning device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0159] The memory 306 is used to store the program 310. The memory 306 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0160] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system or other device. In addition, the embodiments of the present invention are not directed to any particular programming language.

[0161] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. Similarly, in order to simplify the present invention and help understand one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, the various features of the embodiments of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. Wherein, the claims that follow the specific embodiment are hereby expressly incorporated into the specific embodiment, wherein each claim itself is a separate embodiment of the present invention.

[0162] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and further may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive.

[0163] It should be noted that the above embodiments illustrate the present invention rather than limit it, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising a number of different elements and by means of a suitably programmed computer. In a unit claim enumerating a number of devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.

Claims

1. A method for identifying and locating a power system power flow section, characterized in that: The method comprises: Collecting real-time power flow data in the power system; preprocessing the real-time power flow data to obtain power system node data; Inputting the power system node data into an improved convolutional neural network module to extract the power system spatial features; Inputting the power system node data into a long short-term memory network module to capture the power system time series characteristics; The spatial characteristics of the power system and the temporal characteristics of the power system are integrated to output the recognition probability of the power flow section and determine the target power flow section; The tidal path of the target tidal section is optimized and the position of the target tidal section is determined by using the real-time tidal data of the target tidal section.

2. The method for identifying and locating power system flow sections according to claim 1, characterized in that: Preprocessing the real-time power flow data to obtain power system node data includes: According to the node structure of the power system, the real-time power flow data is preprocessed to obtain the power system node data; Wherein, the preprocessing includes: data standardization and denoising; The real-time power flow data includes: voltage, active power, reactive power, power flow direction and power flow amount.

3. The method for identifying and locating power system flow sections according to claim 1, characterized in that: The improved convolutional neural network module includes: a plurality of sequentially connected convolutional layers and a first fully connected layer; each of the convolutional layers is connected by a pooling layer; each of the convolutional layers includes a residual network and a self-attention mechanism module; Each of the convolutional layers is used to calculate the features of the input data using the corresponding residual network and output the features of the preset dimensions; the features of the preset dimensions are calculated using the self-attention mechanism module and the enhanced features are output; Among them, the input data of the first convolution layer is the power system node data, and the input data of the non-first convolution layer is the output of the pooling layer corresponding to the previous convolution layer; The pooling layer is used to perform a maximum pooling operation on the output of the corresponding self-attention mechanism module and output a feature vector; The first fully connected layer is used to take the output of each pooling layer as input and output the spatial characteristics of the power system.

4. The method for identifying and locating power system flow sections according to claim 3, characterized in that: The power system node data is input into the improved convolutional neural network module to extract the power system spatial features, including: The power system node data is input into the improved convolutional neural network module, and each convolution layer and each pooling layer network is used in turn to calculate the power system node data to obtain the feature vector of each layer; the feature vector of each layer is input into the fully connected layer to output the spatial characteristics of the power system.

5. The method for identifying and locating power system flow sections according to claim 1, characterized in that: The power system node data is input into the long short-term memory network module to capture the power system time series characteristics, including: Use the long short-term memory network module to perform time series calculation on the power system node data and output the time series characteristics of the power system; The long short-term memory network module uses memory units and gating mechanisms to perform time series calculations on input data.

6. The method for identifying and locating power system power flow sections according to claim 1, characterized in that: The spatial characteristics of the power system and the temporal characteristics of the power system are integrated to output the recognition probability of the power flow section and determine the target power flow section, including: The spatial features and temporal features of the power system are input into the second fully connected layer for feature fusion, and the recognition probability of each power flow section is output; The flow section whose recognition probability exceeds a preset threshold is determined as the target flow section.

7. The method for identifying and locating power system flow sections according to claim 1, characterized in that: Optimizing the flow path of the target flow section and determining the position of the target flow section by using the real-time flow data of the target flow section includes: The objective function is set based on the estimated value of the target flow section and the label value of the actual flow section; The objective function is iteratively optimized, and the tidal current path corresponding to the target tidal current section is adjusted until a preset optimization condition is reached to obtain the position of the target tidal current section.

8. A power system flow section identification and positioning device, characterized in that: The device comprises: a collection module, a processing module, an identification module and a positioning module; The acquisition module is used to collect real-time power flow data in the power system; pre-process the real-time power flow data to obtain power system node data; A processing module is used to input the power system node data into an improved convolutional neural network module to extract the power system spatial features; input the power system node data into a long short-term memory network module to capture the power system time series features; An identification module, used to integrate the spatial characteristics of the power system and the temporal characteristics of the power system, output the identification probability of the power flow section, and determine the target power flow section; The positioning module is used to optimize the tidal path of the target tidal section and determine the position of the target tidal section through the real-time tidal data of the target tidal section.

9. A power system flow section identification and positioning device, characterized in that: include: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform the operation of the power system power flow section identification and positioning method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The storage medium stores at least one executable instruction, and when the executable instruction is executed on the power system flow section identification and positioning device / apparatus, the power system flow section identification and positioning device / apparatus performs the operation of the power system flow section identification and positioning method as described in any one of claims 1-7.

Images