CN115800274B - 5G distribution network feeder automation self-adaptation method, device and storage medium - Google Patents
5G distribution network feeder automation self-adaptation method, device and storage medium Download PDFInfo
- Publication number
- CN115800274B CN115800274B CN202310077578.6A CN202310077578A CN115800274B CN 115800274 B CN115800274 B CN 115800274B CN 202310077578 A CN202310077578 A CN 202310077578A CN 115800274 B CN115800274 B CN 115800274B
- Authority
- CN
- China
- Prior art keywords
- time delay
- feeder automation
- dependency
- distribution network
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The invention discloses a 5G distribution network feeder automation self-adaptation method, a device and a storage medium, wherein the method comprises the following steps: acquiring real-time operation data and historical feeder automation data of a 5G power distribution network; the historical feeder automation data comprise the historical data of fault processing by different feeder automation modes under different communication delays; inputting real-time operation data into a preset time delay dependency calculation model to obtain the current time delay dependency; calculating a time delay dependency interval of each feeder automation mode according to the maximum fault processing time delay and the average feeder automation time delay in different feeder automation modes; and comparing the current time delay dependence with a time delay dependence interval to adjust a feeder automation mode of the 5G power distribution network. The invention can switch the feeder automation mode of the fastest fault processing corresponding to different network time delays, and realize the self-adaptive switching of feeder automation caused by 5G communication sporadic time delay, thereby ensuring the operation safety of the power distribution network.
Description
Technical Field
The invention relates to the technical field of feeder automation, in particular to a 5G power distribution network feeder automation self-adaptation method, a device and a storage medium.
Background
Feeder automation is an important power distribution network fault self-healing control technology, and the power supply reliability is improved by detecting faults, isolating faults and finally recovering power supply. The intelligent distributed feeder automation technology is used as a current mainstream feeder automation mode, and the measurement and control information is exchanged in a peer-to-peer manner mainly by adopting the same communication mode between power distribution terminals, so that the control decision is realized to complete the rapid fault isolation, accurate positioning and self-healing recovery of the power distribution network. At present, the intelligent distributed feeder automation technology mainly depends on an optical fiber communication mode to realize information interaction, however, the construction cost of the optical fiber channel of the urban power distribution network is high, and the implementation difficulty is high. With the rapid development and wide application of novel communication technologies represented by 5G communication technology, the problem can be effectively solved, and the 5G communication technology has the characteristics of flexible networking mode, long transmission distance, high safety performance and the like, and is suitable for intelligent distributed feeder automation technology.
However, the wireless communication technology has the problem of low reliability, is easily influenced by external environment, so that the communication delay is increased, and the requirement of high real-time communication interaction between power distribution terminals is difficult to meet. Therefore, how to provide a feeder automation self-adaptive method for a 5G power distribution network for uncertain fluctuation time delay of 5G communication becomes a problem to be solved in the present day.
Disclosure of Invention
The invention aims to solve the technical problem of providing a 5G power distribution network feeder automation self-adaption method, a device and a storage medium, which are based on real-time operation data of a 5G power distribution network and historical data of fault processing by different feeder automation modes under different communication time delays, and output a feeder automation mode which is applicable to the fastest distribution network fault response of the real-time network time delay through LSTM deep learning of the historical data, so that the feeder automation mode which corresponds to the fastest fault processing can be switched out by different network time delays, the self-adaption switching of the feeder automation caused by the 5G communication sporadic time delay is realized, and the operation safety of the power distribution network is ensured.
In order to achieve the above object, an embodiment of the present invention provides a feeder automation adaptive method for a 5G power distribution network, including:
acquiring real-time operation data and historical feeder automation data of a 5G power distribution network; the historical feeder automation data comprise historical data of fault processing by different feeder automation modes under different communication delays;
inputting the real-time operation data into a preset time delay dependency calculation model to obtain the current time delay dependency output by the time delay dependency calculation model; the time delay dependency calculation model is obtained by training according to the historical feeder automation data and comprises an improved long-short-term memory neural network model and a residual error module;
calculating a time delay dependency interval of each feeder automation mode according to the maximum fault processing time delay and the average feeder automation time delay in different feeder automation modes;
and comparing the current time delay dependence with the time delay dependence interval, and adjusting a feeder automation mode of the 5G power distribution network according to the comparison result.
As an improvement of the scheme, the training method of the time delay dependency calculation model specifically comprises the following steps:
the history feeder automation data is used as input data and is input into an improved long-short-time memory neural network model for training, and initial time delay dependency of the output of the improved long-short-time memory neural network model is obtained;
sequentially inputting the initial time delay dependency degree to an improved data pooling layer and a residual error module to obtain the time delay dependency degree output by the residual error module;
and optimizing the improved long-short-time memory neural network model and the residual error module according to a preset loss function to obtain a trained time delay dependency calculation model.
As an improvement of the scheme, the improved long-short-time memory neural network model comprises a plurality of layers of long-short-time memory neural networks, and each layer of long-short-time memory neural network is provided withNnA training unit; wherein, the liquid crystal display device comprises a liquid crystal display device,,nrepresent the firstnLayer length time memory neural networks.
As an improvement of the scheme, the residual error module comprises a cavity convolution, an activation function and a batch normalization function; the calculation formula of the receptive field in the cavity convolution is as follows:
in the method, in the process of the invention,receptive field representing the current layer,/->Represents the receptive field of the next layer, m represents the size of the equivalent convolution kernel, L represents the first layer to +.>The product of the layer steps.
As an improvement of the above solution, the optimizing the improved long-short-time memory neural network model and the residual error module according to a preset loss function to obtain a trained time delay dependency calculation model specifically includes:
back-propagating the improved long-short-term memory neural network model and the residual error module according to a preset loss function;
and carrying out iterative updating on the improved long-short-term memory neural network model and parameters in the residual error module until preset iterative conditions are met, so as to obtain a trained time delay dependency calculation model.
As an improvement of the above solution, the comparing the current time delay dependency and the time delay dependency interval, and adjusting the feeder automation mode of the 5G power distribution network according to the comparison result specifically includes:
comparing the current time delay dependency with a time delay dependency interval to obtain a time delay dependency interval in which the current time delay dependency falls;
and adjusting a feeder automation mode corresponding to the time delay dependency interval in which the current time delay dependency falls into a current feeder automation mode of the 5G power distribution network.
As an improvement of the above scheme, the calculation formula of the delay dependency is:
in the method, in the process of the invention,yin order to be a time delay dependency,for maximum processing delay of fault->Average delay for feeder automation.
The embodiment of the invention also provides a feeder automation self-adapting device of the 5G power distribution network, which comprises the following components:
the acquisition module is used for acquiring real-time operation data and historical feeder automation data of the 5G power distribution network; the historical feeder automation data comprise historical data of fault processing by different feeder automation modes under different communication delays;
the first calculation module is used for inputting the real-time operation data into a preset time delay dependency calculation model to obtain the current time delay dependency output by the time delay dependency calculation model; the time delay dependency calculation model is obtained by training according to the historical feeder automation data and comprises an improved long-short-term memory neural network model and a residual error module;
the second calculation module is used for calculating the time delay dependency interval of each feeder automation mode according to the maximum fault processing time delay and the average feeder automation time delay in different feeder automation modes;
and the adjusting module is used for comparing the current time delay dependence with the time delay dependence interval and adjusting a feeder automation mode of the 5G power distribution network according to the comparison result.
The embodiment of the invention also provides a 5G power distribution network feeder automation self-adapting device, which comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, wherein the 5G power distribution network feeder automation self-adapting method is realized when the processor executes the computer program.
The embodiment of the invention also provides a computer readable storage medium, which comprises a stored computer program, wherein the computer program is used for controlling equipment where the computer readable storage medium is located to execute the 5G distribution network feeder automation self-adaption method.
Compared with the prior art, the 5G distribution network feeder automation self-adaptation method, device and storage medium provided by the embodiment of the invention have the beneficial effects that: acquiring real-time operation data and historical feeder automation data of a 5G power distribution network; the historical feeder automation data comprise historical data of fault processing by different feeder automation modes under different communication delays; inputting the real-time operation data into a preset time delay dependency calculation model to obtain the current time delay dependency output by the time delay dependency calculation model; the time delay dependency calculation model is obtained by training according to the historical feeder automation data and comprises an improved long-short-term memory neural network model and a residual error module; calculating a time delay dependency interval of each feeder automation mode according to the maximum fault processing time delay and the average feeder automation time delay in different feeder automation modes; and comparing the current time delay dependence with the time delay dependence interval, and adjusting a feeder automation mode of the 5G power distribution network according to the comparison result. The embodiment of the invention outputs the feeder automation mode of the fastest distribution network fault response applicable to the real-time network time delay based on the real-time operation data of the 5G distribution network and the historical data of the fault processing by different feeder automation modes under different communication time delays through the LSTM deep learning historical data, so that the feeder automation mode of the fastest fault processing can be switched corresponding to different network time delays, the self-adaptive switching of the feeder automation caused by the 5G communication sporadic time delay is realized, and the operation safety of the distribution network is ensured.
Drawings
FIG. 1 is a schematic flow chart of a preferred embodiment of a feeder automation adaptation method for a 5G power distribution network provided by the present invention;
FIG. 2 is a schematic diagram of a preferred embodiment of a feeder automation adaptation device for a 5G power distribution network according to the present invention;
fig. 3 is a schematic structural diagram of another preferred embodiment of a feeder automation adaptive device for a 5G power distribution network provided by the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, fig. 1 is a schematic flow chart of a preferred embodiment of a feeder automation adaptive method for a 5G power distribution network according to the present invention. The feeder automation self-adapting method of the 5G power distribution network comprises the following steps:
s1, acquiring real-time operation data and historical feeder automation data of a 5G power distribution network; the historical feeder automation data comprise historical data of fault processing by different feeder automation modes under different communication delays;
s2, inputting the real-time operation data into a preset time delay dependency calculation model to obtain the current time delay dependency output by the time delay dependency calculation model; the time delay dependency calculation model is obtained by training according to the historical feeder automation data and comprises an improved long-short-term memory neural network model and a residual error module;
s3, calculating a time delay dependency interval of each feeder automation mode according to the maximum fault processing time delay and the average feeder automation time delay in different feeder automation modes;
s4, comparing the current time delay dependency with the time delay dependency interval, and adjusting a feeder automation mode of the 5G power distribution network according to a comparison result.
Specifically, the embodiment of the invention acquires real-time operation data and historical feeder automation data of the 5G power distribution network. The historical feeder automation data comprise the historical data of fault processing by different feeder automation modes under different communication delays. And inputting the real-time operation data into a preset time delay dependency calculation model to obtain the current time delay dependency output by the time delay dependency calculation model. The time delay dependency calculation model is obtained by training according to historical feeder automation data and comprises an improved long-short-time memory neural network model LSTM and a residual error module. And calculating the time delay dependency interval of each feeder automation mode according to the maximum fault processing time delay and the average feeder automation time delay in different feeder automation modes. And comparing the current time delay dependence with a time delay dependence interval, and adjusting a feeder automation mode of the 5G power distribution network according to the comparison result.
It should be noted that, in the embodiment of the present invention, considering the weak adaptability of the single-layer LSTM network, the disadvantages of over-fitting and gradient disappearance may occur, and the data features of different depths are extracted from the data by using the multi-layer stacked LSTM, the processed data is subjected to fusion analysis, and the stacked LSTM performs repeated training on the sample features to weaken the weak points of over-fitting and gradient disappearance. Because the increase of the network layer number can cause the serious overfitting problem of the model during training to cause the classification effect to be not added, a residual block is added to improve the neural network, an output result is transmitted to an improved data pooling layer as input, and the improved data pooling layer and the residual module are utilized to optimize the extraction of the characteristic information, so that the training performance is further improved.
The embodiment of the invention outputs the feeder automation mode of the fastest distribution network fault response applicable to the real-time network time delay based on the real-time operation data of the 5G distribution network and the historical data of the fault processing by different feeder automation modes under different communication time delays through the LSTM deep learning historical data, so that the feeder automation mode of the fastest fault processing can be switched corresponding to different network time delays, the self-adaptive switching of the feeder automation caused by the 5G communication sporadic time delay is realized, and the operation safety of the distribution network is ensured.
In another preferred embodiment, the training method of the delay dependency calculation model specifically includes:
the history feeder automation data is used as input data and is input into an improved long-short-time memory neural network model for training, and initial time delay dependency of the output of the improved long-short-time memory neural network model is obtained;
sequentially inputting the initial time delay dependency degree to an improved data pooling layer and a residual error module to obtain the time delay dependency degree output by the residual error module;
and optimizing the improved long-short-time memory neural network model and the residual error module according to a preset loss function to obtain a trained time delay dependency calculation model.
Specifically, based on the fault processing conditions of different feeder automation modes with different communication delays under the historical conditions, in order to select the feeder mode with the fastest fault processing speed under the real-time communication delay, the historical data of the fault processing is used as training samples by using the different feeder automation modes with different communication delays, and a long-short-term memory neural network model (LSTM) is constructed to train the samples. In order to prevent the conditions of over-fitting, gradient disappearance and the like in the training process, the embodiment of the invention adopts the multilayer stacked LSTM network to solve the problem of weak adaptability of the single-layer LSTM network, and the LSTM stacked layer by layer repeatedly trains sample characteristics to weaken the weak points of over-fitting and gradient disappearance, and the processed data are subjected to fusion analysis. And taking the historical feeder automation data as input data, inputting the input data into the improved long-short-time memory neural network model for training, and obtaining the initial time delay dependence of the output of the improved long-short-time memory neural network model. And sequentially inputting the initial time delay dependence into the improved data pooling layer and the residual error module to obtain the time delay dependence output by the residual error module. And optimizing the improved long-short-time memory neural network model and the residual error module according to a preset loss function to obtain a trained time delay dependency calculation model.
The typical LSTM structure is divided into three layers, including an input layer, an output layer and a hidden layer, wherein the input layer is used for inputting a feature sequence set to be trained, and the input and the output have no one-to-one correspondence.
Input node: this node is marked asThe hidden node is represented as a common neuron, and receives as output the output of the hidden node at the last time point and the current input.
Wherein, the liquid crystal display device comprises a liquid crystal display device,representing the tanh activation function, +.>Input representing the current power distribution network delay +.>Representing the delay dependency output of the power distribution network in the last stage, < ->And->Respectively representing the power distribution network time delay dependency weight of input nodes for introducing historical time delay input and upper-stage output of the power distribution network, and +.>And the time delay input deviation term of the power distribution network represents the input node.
An input door: to be used forThe input gates are named, the input multiplier gates are a salient feature of the LSTM model, and as with the input nodes, the gate inputs are the output and current input of the hidden node of the previous time node.
Wherein, the liquid crystal display device comprises a liquid crystal display device,representing sigmoid function activation,/->Representing the current power distribution network delay input->Represents the delay dependency output of the power distribution network in the last stage, < ->And->Distribution network time delay dependency weight representing input gate distribution network historical time delay input and upper stage output respectively,/>Representing the time delay input deviation term of the power distribution network of the input gate.
Forgetting to gate: by this method, the network can learn to refresh the content of the internal state.
Wherein, the liquid crystal display device comprises a liquid crystal display device,and->Distribution network time delay dependency weight representing distribution network time delay input and upper-stage output of forgetting gate refreshing respectively, < ->Representing a power distribution network delay refreshing deviation term.
Output door: the input of the gate is the output of the hidden node at the last moment and the current input.
Wherein, the liquid crystal display device comprises a liquid crystal display device,and->Distribution network delay dependency weights representing the delay input and the upper phase output of the distribution network of the output gate respectively,/->And outputting a deviation term representing the time delay dependency of the power distribution network.
Internal state: the core of each memory cell is a node with linear activation, the cell is indexed by the linear activation node c, the internal state of the cell is indexed, and the internal state is named:
Wherein, the liquid crystal display device comprises a liquid crystal display device,input node input at time t is indicated, < >>Represents the input gate input at time t,/>Represents the internal state quantity at time t-1, < >>Indicating the internal state refresh content at time t.
Will representMultiplying the output gate of (2) by the internal state->To generate a distribution network delay dependency outputted by a memory unit>And based on the value of network delay dependency +.>And analyzing and selecting a feeder line automatic switching mode with the fastest fault response.
Wherein, the liquid crystal display device comprises a liquid crystal display device,representing a term-wise multiplication.
The problem of feature information loss is improved by adding a pooling layer and a residual block because the problem of serious overfitting generated by the model during training can be solved by increasing the network layer number, so that the classification effect is not increased. And adding a residual block to improve the residual neural network, transmitting an output result as input to an improved data pooling layer, and optimizing the extraction of the characteristic information by utilizing the improved data pooling layer and the residual block to further improve the training performance and ensure that the classification effect is better.
As a preferable scheme, the improved long-short time memory neural network model comprises a plurality of layers of long-short time memory neural networks, and each layer of long-short time memory neural network is provided withNnA training unit; wherein, the liquid crystal display device comprises a liquid crystal display device,,nrepresent the firstnLayer length time memory neural networks.
In particular, embodiments of the present invention contemplate adaptation of a single layer LSTM networkIn order to prevent the conditions of over fitting, gradient disappearance and the like in the training process, the embodiment of the invention adopts a multilayer stacked LSTM network to solve the problem of weak adaptability of a single-layer LSTM network, and the LSTM stacked layer by layer repeatedly trains sample characteristics to weaken the weaknesses of over fitting and gradient disappearance, and the processed data are subjected to fusion analysis. Wherein, the number of stacked layers of the LSTM is preferably 2-4, and each long-short-time memory neural network is provided withNnA training unit; wherein, the liquid crystal display device comprises a liquid crystal display device,,nrepresent the firstnLayer length time memory neural networks.
In yet another preferred embodiment, the residual module includes a hole convolution, an activation function, and a batch normalization function; the calculation formula of the receptive field in the cavity convolution is as follows:
in the method, in the process of the invention,receptive field representing the current layer,/->Represents the receptive field of the next layer, m represents the size of the equivalent convolution kernel, L represents the first layer to +.>The product of the layer steps.
Specifically, in the embodiment of the invention, the classification effect is not improved due to the fact that the model generates a serious overfitting problem during training due to the increase of the LSTM network layer number, so that a residual block is added to improve the neural network, an output result is used as input to be transmitted to an improved data pooling layer, and the extraction of characteristic information is optimized by utilizing the improved data pooling layer and the residual module, so that the training performance is further improved, and the classification effect is better. The residual module comprises a cavity convolution, an activation function and a batch normalization function. The cavity convolution expands the convolution kernel receptive field by adding zero padding in the common convolution, so that the convolution receptive field is larger on the premise of not changing the feature resolution, and the scope of perception information is increased. The calculation formula of the receptive field in the cavity convolution is as follows:
in the method, in the process of the invention,receptive field representing the current layer,/->Represents the receptive field of the next layer, m represents the size of the equivalent convolution kernel, L represents the first layer to +.>The product of the layer steps.
In another preferred embodiment, the optimizing the improved long-short-time memory neural network model and the residual module according to a preset loss function to obtain a trained time delay dependency calculation model specifically includes:
back-propagating the improved long-short-term memory neural network model and the residual error module according to a preset loss function;
and carrying out iterative updating on the improved long-short-term memory neural network model and parameters in the residual error module until preset iterative conditions are met, so as to obtain a trained time delay dependency calculation model.
In another preferred embodiment, the comparing the current time delay dependency and the time delay dependency interval, and adjusting a feeder automation mode of the 5G power distribution network according to the comparison result specifically includes:
comparing the current time delay dependency with a time delay dependency interval to obtain a time delay dependency interval in which the current time delay dependency falls;
and adjusting a feeder automation mode corresponding to the time delay dependency interval in which the current time delay dependency falls into a current feeder automation mode of the 5G power distribution network.
Specifically, the embodiment of the invention compares the current time delay dependency and the time delay dependency interval to obtain the time delay dependency interval in which the current time delay dependency falls. And adjusting the feeder automation mode corresponding to the time delay dependency interval in which the current time delay dependency falls into the current feeder automation mode of the 5G power distribution network.
In another preferred embodiment, the calculation formula of the delay dependency is:
in the method, in the process of the invention,yin order to be a time delay dependency,for maximum processing delay of fault->Average delay for feeder automation.
Specifically, three types of feeder automation methods based on the current common use can be classified according to the degree of dependence on communication, namely an interactive feeder automation technology, an interrogation feeder automation technology and a voltage-current feeder automation technology. And dividing a time delay dependency interval according to the dependency degree of different feeder automation modes on the communication time delay so as to output classification by the LSTM model. The calculation formula of the time delay dependency is as follows:
in the method, in the process of the invention,yin order to be a time delay dependency,for maximum processing delay of fault->Average delay for feeder automation.
The time delay dependency interval of the three feeder automation modes can be divided as follows according to the time delay dependency:
time delay dependency interval [ 0-0.3 ] of voltage-current feeder automation mode;
time delay dependency interval [ 0.3-0.6 ] of an interrogation type feeder automation mode;
the time delay dependency interval [ 0.6-1 ] of the interactive feeder automation mode.
It should be noted that, the embodiment of the invention can also classify a layer of Softmax used at the end of the time delay dependency calculation model, obtain the feeder automation mode of the fastest fault response under different network time delays according to the classification result, and switch the appropriate feeder mode according to the result so as to realize the rapid processing of the power distribution network faults under different real-time communication time delays.
Softmax function calculation:
wherein K is the total number of outputs, the denominator is normalization, the sum of indexes on all output nodes is formed, the sum of outputs is ensured to be 1, and the softmax function is also called a normalization index function and is used for displaying the multi-classification result in a probability form. Wherein, the liquid crystal display device comprises a liquid crystal display device,is the output value of the kth node, K summed by denominators is the number of output nodes, ++>Is summed from the first output node. The output symbol is for the sake of conspicuity and easy discrimination.
When (when)When the system adjusts and selects a voltage-current feeder automation technology to switch the feeder;
when (when)When the system adjusts and selects an inquiry type feeder automation technology to switch the feeder;
when (when)And when the system adjusts and selects the interactive feeder automation technology to switch the feeder.
According to the embodiment of the invention, the data are learned and trained through the multilayer stacked LSTM long short-time memory neural network model, the extraction of the characteristic information is optimized through adding the improved data pool and the cavity convolution residual block, finally, the data normalization classification processing is carried out through the Softmax function, the real-time power distribution network communication time delay is acquired and input, the feeder automation mode of the current network time delay for processing faults most rapidly is obtained according to the training and learning result, and the corresponding feeder automation strategy is deployed. The feeder switching mode aiming at the real-time communication time delay can eliminate the influence of the uncertainty of the communication time delay on the feeder automation technology.
Correspondingly, the invention also provides a 5G distribution network feeder automation self-adapting device, which can realize all the flows of the 5G distribution network feeder automation self-adapting method in the embodiment.
Referring to fig. 2, fig. 2 is a schematic structural diagram of a preferred embodiment of a feeder automation adaptive device for a 5G power distribution network according to the present invention. The 5G distribution network feeder automation self-adapting device comprises:
an acquisition module 201, configured to acquire real-time operation data and historical feeder automation data of a 5G power distribution network; the historical feeder automation data comprise historical data of fault processing by different feeder automation modes under different communication delays;
a first calculation module 202, configured to input the real-time operation data into a preset time delay dependency calculation model, so as to obtain a current time delay dependency output by the time delay dependency calculation model; the time delay dependency calculation model is obtained by training according to the historical feeder automation data and comprises an improved long-short-term memory neural network model and a residual error module;
a second calculation module 203, configured to calculate a time delay dependency interval of each feeder automation mode according to the maximum fault processing time delay and the average feeder automation time delay in different feeder automation modes;
and the adjusting module 204 is configured to compare the current time delay dependency and the time delay dependency interval, and adjust a feeder automation mode of the 5G power distribution network according to a comparison result.
Preferably, the training method of the time delay dependency calculation model specifically includes:
the history feeder automation data is used as input data and is input into an improved long-short-time memory neural network model for training, and initial time delay dependency of the output of the improved long-short-time memory neural network model is obtained;
sequentially inputting the initial time delay dependency degree to an improved data pooling layer and a residual error module to obtain the time delay dependency degree output by the residual error module;
and optimizing the improved long-short-time memory neural network model and the residual error module according to a preset loss function to obtain a trained time delay dependency calculation model.
Preferably, the improved long-short time memory neural network model comprises multiple layers of long-short time memory neural networks, and each layer of long-short time memory neural network is provided withNnA training unit; wherein, the liquid crystal display device comprises a liquid crystal display device,,nrepresent the firstnLayer length time memory neural networks.
Preferably, the residual module comprises a hole convolution, an activation function and a batch normalization function; the calculation formula of the receptive field in the cavity convolution is as follows:
in the method, in the process of the invention,receptive field representing the current layer,/->Represents the receptive field of the next layer, m represents the size of the equivalent convolution kernel, L represents the first layer to +.>The product of the layer steps.
Preferably, the optimizing the improved long-short-time memory neural network model and the residual error module according to a preset loss function to obtain a trained time delay dependency calculation model specifically includes:
back-propagating the improved long-short-term memory neural network model and the residual error module according to a preset loss function;
and carrying out iterative updating on the improved long-short-term memory neural network model and parameters in the residual error module until preset iterative conditions are met, so as to obtain a trained time delay dependency calculation model.
Preferably, the adjustment module 204 is specifically configured to:
comparing the current time delay dependency with a time delay dependency interval to obtain a time delay dependency interval in which the current time delay dependency falls;
and adjusting a feeder automation mode corresponding to the time delay dependency interval in which the current time delay dependency falls into a current feeder automation mode of the 5G power distribution network.
Preferably, the calculation formula of the delay dependency is:
in the method, in the process of the invention,yin order to be a time delay dependency,for maximum processing delay of fault->Average delay for feeder automation.
In specific implementation, the working principle, control flow and technical effects of the 5G power distribution network feeder automation self-adapting device provided by the embodiment of the present invention are the same as those of the 5G power distribution network feeder automation self-adapting method in the above embodiment, and are not described herein.
Referring to fig. 3, fig. 3 is a schematic structural diagram of another preferred embodiment of a feeder automation adaptive device for a 5G power distribution network according to the present invention. The 5G distribution grid feeder automation adaptation device comprises a processor 301, a memory 302 and a computer program stored in the memory 302 and configured to be executed by the processor 301, wherein the processor 301 implements the 5G distribution grid feeder automation adaptation method according to any of the embodiments above when executing the computer program.
Preferably, the computer program may be divided into one or more modules/units (e.g. computer program 1, computer program 2, … …) stored in the memory 302 and executed by the processor 301 to complete the present invention. The one or more modules/units may be a series of computer program instruction segments capable of performing specific functions for describing the execution of the computer program in the 5G distribution grid feeder automation adaptation device.
The processor 301 may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application SpecificIntegrated Circuit, ASIC), field programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., the general purpose processor may be a microprocessor, or the processor 301 may be any conventional processor, the processor 301 being the control center of the 5G distribution grid feeder automation adaptation, the various interfaces and lines being utilized to connect the various parts of the 5G distribution grid feeder automation adaptation.
The memory 302 mainly includes a program storage area, which may store an operating system, application programs required for at least one function, and the like, and a data storage area, which may store related data and the like. In addition, the memory 302 may be a high-speed random access memory, a nonvolatile memory such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash Card (Flash Card), etc., or the memory 302 may be other volatile solid-state memory devices.
It should be noted that the foregoing 5G power distribution network feeder automation adaptive device may include, but is not limited to, a processor, a memory, and those skilled in the art will understand that the schematic structural diagram of fig. 3 is merely an example of the foregoing 5G power distribution network feeder automation adaptive device, and does not limit the foregoing 5G power distribution network feeder automation adaptive device, and may include more or fewer components than illustrated, or may combine some components, or different components.
The embodiment of the invention also provides a computer readable storage medium, which comprises a stored computer program, wherein when the computer program runs, equipment where the computer readable storage medium is located is controlled to execute the 5G distribution network feeder automation self-adaption method according to any embodiment.
The embodiment of the invention provides a 5G power distribution network feeder automation self-adaption method, a device and a storage medium, wherein real-time operation data and historical feeder automation data of a 5G power distribution network are obtained; the historical feeder automation data comprise historical data of fault processing by different feeder automation modes under different communication delays; inputting the real-time operation data into a preset time delay dependency calculation model to obtain the current time delay dependency output by the time delay dependency calculation model; the time delay dependency calculation model is obtained by training according to the historical feeder automation data and comprises an improved long-short-term memory neural network model and a residual error module; calculating a time delay dependency interval of each feeder automation mode according to the maximum fault processing time delay and the average feeder automation time delay in different feeder automation modes; and comparing the current time delay dependence with the time delay dependence interval, and adjusting a feeder automation mode of the 5G power distribution network according to the comparison result. The embodiment of the invention outputs the feeder automation mode of the fastest distribution network fault response applicable to the real-time network time delay based on the real-time operation data of the 5G distribution network and the historical data of the fault processing by different feeder automation modes under different communication time delays through the LSTM deep learning historical data, so that the feeder automation mode of the fastest fault processing can be switched corresponding to different network time delays, the self-adaptive switching of the feeder automation caused by the 5G communication sporadic time delay is realized, and the operation safety of the distribution network is ensured.
It should be noted that the system embodiments described above are merely illustrative, and that the units described as separate units may or may not be physically separate, and that units shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the system embodiment of the present invention, the connection relationship between the modules represents that there is a communication connection between them, and may be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
While the foregoing is directed to the preferred embodiments of the present invention, it will be appreciated by those skilled in the art that changes and modifications may be made without departing from the principles of the invention, such changes and modifications are also intended to be within the scope of the invention.
Claims (9)
1. The feeder automation self-adaptation method for the 5G power distribution network is characterized by comprising the following steps of:
acquiring real-time operation data and historical feeder automation data of a 5G power distribution network; the historical feeder automation data comprise historical data of fault processing by different feeder automation modes under different communication delays;
inputting the real-time operation data into a preset time delay dependency calculation model to obtain the current time delay dependency output by the time delay dependency calculation model; the time delay dependency calculation model is obtained by training according to the historical feeder automation data and comprises an improved long-short-term memory neural network model and a residual error module;
calculating a time delay dependency interval of each feeder automation mode according to the maximum fault processing time delay and the average feeder automation time delay in different feeder automation modes;
comparing the current time delay dependence with a time delay dependence interval, and adjusting a feeder automation mode of the 5G power distribution network according to a comparison result;
the training method of the time delay dependency calculation model specifically comprises the following steps:
the history feeder automation data is used as input data and is input into an improved long-short-time memory neural network model for training, and initial time delay dependency of the output of the improved long-short-time memory neural network model is obtained;
sequentially inputting the initial time delay dependency degree to an improved data pooling layer and a residual error module to obtain the time delay dependency degree output by the residual error module;
and optimizing the improved long-short-time memory neural network model and the residual error module according to a preset loss function to obtain a trained time delay dependency calculation model.
2. The method as claimed in claim 1The feeder automation self-adapting method of the 5G power distribution network is characterized in that the improved long-short-time memory neural network model comprises a plurality of layers of long-short-time memory neural networks, and each layer of long-short-time memory neural network is provided withNnA training unit; wherein, the liquid crystal display device comprises a liquid crystal display device,,nrepresent the firstnLayer length time memory neural networks.
3. The 5G distribution grid feeder automation adaptation method of claim 2, wherein the residual module comprises a hole convolution, an activation function, and a batch normalization function; the calculation formula of the receptive field in the cavity convolution is as follows:
4. A 5G distribution grid feeder automation adaptation method according to claim 3, wherein the optimizing the improved long-short-term memory neural network model and the residual module according to a preset loss function to obtain a trained time delay dependency calculation model specifically comprises:
back-propagating the improved long-short-term memory neural network model and the residual error module according to a preset loss function;
and carrying out iterative updating on the improved long-short-term memory neural network model and parameters in the residual error module until preset iterative conditions are met, so as to obtain a trained time delay dependency calculation model.
5. The feeder automation self-adaptation method of a 5G power distribution network according to claim 4, wherein the comparing the current time delay dependency and the time delay dependency interval, and adjusting the feeder automation mode of the 5G power distribution network according to the comparison result, specifically comprises:
comparing the current time delay dependency with a time delay dependency interval to obtain a time delay dependency interval in which the current time delay dependency falls;
and adjusting a feeder automation mode corresponding to the time delay dependency interval in which the current time delay dependency falls into a current feeder automation mode of the 5G power distribution network.
7. The utility model provides a 5G distribution network feeder automation self-adaptation device which characterized in that includes:
the acquisition module is used for acquiring real-time operation data and historical feeder automation data of the 5G power distribution network; the historical feeder automation data comprise historical data of fault processing by different feeder automation modes under different communication delays;
the first calculation module is used for inputting the real-time operation data into a preset time delay dependency calculation model to obtain the current time delay dependency output by the time delay dependency calculation model; the time delay dependency calculation model is obtained by training according to the historical feeder automation data and comprises an improved long-short-term memory neural network model and a residual error module;
the second calculation module is used for calculating the time delay dependency interval of each feeder automation mode according to the maximum fault processing time delay and the average feeder automation time delay in different feeder automation modes;
the adjusting module is used for comparing the current time delay dependence with the time delay dependence interval and adjusting a feeder automation mode of the 5G power distribution network according to the comparison result;
the training method of the time delay dependency calculation model specifically comprises the following steps:
the history feeder automation data is used as input data and is input into an improved long-short-time memory neural network model for training, and initial time delay dependency of the output of the improved long-short-time memory neural network model is obtained;
sequentially inputting the initial time delay dependency degree to an improved data pooling layer and a residual error module to obtain the time delay dependency degree output by the residual error module;
and optimizing the improved long-short-time memory neural network model and the residual error module according to a preset loss function to obtain a trained time delay dependency calculation model.
8. A 5G distribution grid feeder automation adaptation device comprising a processor and a memory, the memory having a computer program stored therein and being configured to be executed by the processor, the processor implementing the 5G distribution grid feeder automation adaptation method of any one of claims 1 to 6 when the computer program is executed by the processor.
9. A computer readable storage medium, wherein the computer readable storage medium stores a computer program, and when the computer program is executed by a device where the computer readable storage medium is located, the feeder automation adaptive method of the 5G power distribution network according to any one of claims 1 to 6 is implemented.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310077578.6A CN115800274B (en) | 2023-02-08 | 2023-02-08 | 5G distribution network feeder automation self-adaptation method, device and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310077578.6A CN115800274B (en) | 2023-02-08 | 2023-02-08 | 5G distribution network feeder automation self-adaptation method, device and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115800274A CN115800274A (en) | 2023-03-14 |
CN115800274B true CN115800274B (en) | 2023-06-30 |
Family
ID=85430368
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310077578.6A Active CN115800274B (en) | 2023-02-08 | 2023-02-08 | 5G distribution network feeder automation self-adaptation method, device and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115800274B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117791597A (en) * | 2024-02-23 | 2024-03-29 | 广东电网有限责任公司广州供电局 | Power distribution network fault self-healing method and system based on machine learning |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113036731B (en) * | 2021-04-25 | 2022-02-08 | 武汉大学 | Point-to-point communication power distribution network feeder fault isolation method and self-healing system |
CN114566964B (en) * | 2022-04-29 | 2022-07-29 | 国网天津市电力公司电力科学研究院 | Power distribution network feeder automation control method, device, equipment and storage medium |
CN114726105A (en) * | 2022-06-02 | 2022-07-08 | 国网天津市电力公司电力科学研究院 | Power distribution network feeder automatic switching method, device and medium based on communication time delay |
-
2023
- 2023-02-08 CN CN202310077578.6A patent/CN115800274B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN115800274A (en) | 2023-03-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Rangapuram et al. | Deep state space models for time series forecasting | |
CN110162018B (en) | Incremental equipment fault diagnosis method based on knowledge distillation and hidden layer sharing | |
CN107944410B (en) | Cross-domain facial feature analysis method based on convolutional neural network | |
CN111815033A (en) | Offshore wind power prediction method based on RCNN and meteorological time sequence characteristics | |
CN110276441B (en) | Trapezoidal overlapped kernel pulse estimation method based on deep learning | |
CN113723007B (en) | Equipment residual life prediction method based on DRSN and sparrow search optimization | |
CN108879732B (en) | Transient stability evaluation method and device for power system | |
CN115800274B (en) | 5G distribution network feeder automation self-adaptation method, device and storage medium | |
CN110837888A (en) | Traffic missing data completion method based on bidirectional cyclic neural network | |
CN113486578A (en) | Method for predicting residual life of equipment in industrial process | |
CN115081316A (en) | DC/DC converter fault diagnosis method and system based on improved sparrow search algorithm | |
CN111461463A (en) | Short-term load prediction method, system and equipment based on TCN-BP | |
CN115204302A (en) | Unmanned aerial vehicle small sample fault diagnosis system and method | |
CN111612125A (en) | Novel HTM time pool method and system for online learning | |
CN112766603A (en) | Traffic flow prediction method, system, computer device and storage medium | |
CN112232577A (en) | Power load probability prediction system and method for multi-core intelligent meter | |
CN113469427A (en) | Convolution LSTM-based photovoltaic power station day-ahead power generation amount prediction method and system | |
CN113591728A (en) | Electric energy quality disturbance classification method based on integrated deep learning | |
CN112651519A (en) | Secondary equipment fault positioning method and system based on deep learning theory | |
WO2022039675A1 (en) | Method and apparatus for forecasting weather, electronic device and storage medium thereof | |
CN115018193A (en) | Time series wind energy data prediction method based on LSTM-GA model | |
CN111783688B (en) | Remote sensing image scene classification method based on convolutional neural network | |
Alam | Recurrent neural networks in electricity load forecasting | |
CN109447843B (en) | Photovoltaic output prediction method and device | |
CN116432023A (en) | Novel power system fault classification method based on sample transfer learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |