CN112396535A - Management method, device, equipment and storage medium of smart power grid - Google Patents
Management method, device, equipment and storage medium of smart power grid Download PDFInfo
- Publication number
- CN112396535A CN112396535A CN202011285277.5A CN202011285277A CN112396535A CN 112396535 A CN112396535 A CN 112396535A CN 202011285277 A CN202011285277 A CN 202011285277A CN 112396535 A CN112396535 A CN 112396535A
- Authority
- CN
- China
- Prior art keywords
- frequency data
- neural network
- frequency
- smart grid
- low
- 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.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
Abstract
The invention discloses a management method and a management device of a smart grid, electronic equipment and a storage medium, belonging to the technical field of smart grids, wherein the method comprises the following steps: acquiring low-frequency data of a specific meter, wherein the low-frequency data is initial data acquired to the smart grid according to a set sampling frequency; reconstructing the collected low-frequency data by utilizing a neural network based on a preset super-resolution perception model to obtain reconstructed target high-frequency data; and performing state estimation on the smart grid by using the target high-frequency data so as to manage the smart grid. According to the scheme, low-frequency data in the smart grid are collected, the neural network is used for reconstructing the low-frequency data to obtain target high-frequency data based on a preset super-resolution perception model, state estimation is further carried out on the smart grid based on the target high-frequency data, the smart grid is managed according to a state estimation result, more hidden problems in the smart grid are discovered in time, and therefore safety and operation efficiency of the system are improved, and reliability and stability are achieved.
Description
Technical Field
The present invention relates to the field of smart grids, and in particular, to a method and an apparatus for managing a smart grid, an electronic device, and a storage medium.
Background
Smart grids are an important infrastructure that continues to provide a safe and economical supply of electricity for modern society. State estimation of the smart grid plays a crucial role in system monitoring, and helps system operators to sense the operation state of the system so as to make accurate control decisions. However, more frequent state estimation results help to find more hidden problems in the smart grid, thereby improving the safety and efficiency of the system. Therefore, high frequency data needs to be acquired to perform more frequent state estimation on the smart grid based on the high frequency data.
The existing hardware equipment is limited, and the collection and transmission of high-frequency data are still difficult work; in addition, due to the influence of communication faults or network attacks and the like, data acquired in a traditional mode is easy to lose or be tampered, so that high-frequency data is difficult to acquire, the estimation frequency of the smart grid is influenced, and the reliability is reduced.
Therefore, smart grids require new approaches to support high frequency awareness of system operating conditions based on existing metering infrastructure.
Disclosure of Invention
The application provides a management method and device of a smart grid, electronic equipment and a storage medium, which can solve the technical problem that high-frequency data are difficult to obtain and the estimation frequency of the smart grid is influenced.
The invention provides a management method of a smart grid in a first aspect, which comprises the following steps:
acquiring low-frequency data of a specific meter, wherein the low-frequency data is initial data acquired to a smart grid according to a set sampling frequency;
reconstructing the collected low-frequency data by utilizing a neural network based on a preset super-resolution perception model to obtain reconstructed target high-frequency data;
and performing state estimation on the smart grid by using the target high-frequency data so as to manage the smart grid.
Optionally, the step of reconstructing the acquired low-frequency data by using a neural network based on a preset super-resolution sensing model to obtain reconstructed high-frequency data includes:
generating high-frequency data according to the low-frequency data and the loss information measured in advance by using a neural network based on a preset super-resolution perception model;
the preset super-resolution perception model has the following calculation formula:
La＝↓aH+e
wherein L isaFor the collected low-frequency data, H represents the original high-frequency data, e represents the noise vector corresponding to the lost information, ↓aRepresents a down-sampling function, a represents a down-sampling factor;
and updating the neural network by using the high-frequency data to acquire target high-frequency data.
Optionally, the step of generating high-frequency data according to the low-frequency data and the pre-acquired loss information by using a neural network based on a preset super-resolution sensing model includes:
extracting low-frequency features in pre-measured low-frequency data according to a preset time sequence rule by utilizing three one-dimensional convolution layers of a neural network;
performing information completion on the high-frequency features corresponding to the low-frequency features by using a residual error structure of the neural network to obtain high-frequency feature vectors after the information completion;
and reconstructing the high-frequency characteristic vector according to a preset subsequence by using the three one-dimensional convolution layers of the neural network to obtain high-frequency data.
Optionally, the step of performing information completion on the high-frequency feature corresponding to the low-frequency feature by using the residual structure of the neural network to obtain a high-frequency feature vector after the information completion includes:
generating a high-frequency feature corresponding to the low-frequency feature according to a preset corresponding relation between low-frequency data and high-frequency data by using a residual error structure of the neural network;
and performing information completion on the high-frequency features according to the lost information to obtain the high-frequency feature vectors after the information completion.
Optionally, the step of updating the neural network by using the high-frequency data to obtain target high-frequency data includes:
and determining an error loss function value by using the high-frequency data, wherein the calculation formula of the loss function is as follows:
where MSE represents the error loss function value, β is the down-sampling factor,
representing predicted high-frequency data, H, output by a neural network modeliβThe method comprises the steps of representing sampling high-frequency data obtained by sampling original high-frequency data by taking beta as a down-sampling factor, wherein N represents the number of estimated high-frequency data, and i represents a natural number;
and determining whether updating of the neural network is finished or not according to the error loss function value so as to obtain target high-frequency data.
Optionally, the step of determining whether updating the neural network is completed according to the error loss function value to obtain target high-frequency data includes:
comparing the error loss function value with a preset error loss function threshold;
if the error loss function value is larger than the preset error loss function threshold, determining to update the neural network until the error loss function value is smaller than or equal to the preset error loss function threshold;
and if the error loss function value is smaller than or equal to the preset error loss function threshold, stopping updating the neural network, and calculating target high-frequency data according to the updated neural network.
Optionally, the step of stopping updating the neural network, and calculating target high-frequency data according to the updated neural network includes:
after the neural network is stopped to be updated, determining an average absolute error by using the high-frequency data corresponding to the updated neural network, wherein a calculation formula of the average absolute error is as follows:
wherein MAPE represents the mean absolute error,
representing estimated high frequency data, HiβRepresenting sampled high frequency data, N representing the number of high frequency data, i representing a natural number;
and determining a signal-to-noise ratio by using the updated high-frequency data corresponding to the neural network, wherein the signal-to-noise ratio calculation formula is as follows:
where, SNR represents the signal-to-noise ratio,
representing estimated high frequency data, HiβRepresenting sampled high-frequency data, N representing the number of estimated high-frequency data, and i representing a natural number;
and determining that the neural network is updated according to the average absolute error and the signal-to-noise ratio so as to obtain target high-frequency data corresponding to the updated neural network.
A second aspect of the present invention provides a management apparatus for a smart grid, the apparatus including:
the acquisition module is used for acquiring low-frequency data of a specific meter, wherein the low-frequency data is initial data acquired to the smart grid according to a set sampling frequency;
the reconstruction module is used for reconstructing the collected low-frequency data by utilizing a neural network based on a preset super-resolution perception model to obtain reconstructed target high-frequency data;
and the estimation module is used for carrying out state estimation on the smart grid by utilizing the target high-frequency data so as to manage the smart grid.
A third aspect of the present invention provides an electronic device comprising: the management method of the smart grid comprises a memory, a processor and a communication bus, wherein the communication bus is respectively connected with the memory and the processor in a communication mode, the memory is coupled with the processor, a computer program is stored in the memory, and when the processor executes the computer program, all steps in the management method of the smart grid in the first aspect are realized.
A fourth aspect of the present invention provides a storage medium, which is a computer-readable storage medium having a computer program stored thereon, where the computer program, when executed by a processor, implements the steps of the management method for a smart grid according to the first aspect.
The management method of the smart power grid provided by the invention comprises the following steps: acquiring low-frequency data of a specific meter, wherein the low-frequency data is initial data acquired to the smart grid according to a set sampling frequency; reconstructing the collected low-frequency data by utilizing a neural network based on a preset super-resolution perception model to obtain reconstructed target high-frequency data; and performing state estimation on the smart grid by using the target high-frequency data so as to manage the smart grid. According to the scheme, low-frequency data in the intelligent network are collected, the neural network is utilized to reconstruct the low-frequency data to obtain target high-frequency data based on a preset super-resolution perception model, the integrity of the data is provided, the direct collection and transmission of the high-frequency data are avoided, the phenomenon that the collected high-frequency data are lost and tampered is avoided, more accurate and frequent state estimation is further carried out on the intelligent power grid based on the target high-frequency data, the intelligent power grid is managed according to a state estimation structure, more hidden problems in the intelligent power grid are discovered in time, the safety and the efficiency of the system are improved, and the reliability is achieved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a management system architecture diagram of a smart grid according to an embodiment of the present invention;
fig. 2 is a flowchart illustrating steps of a management method for a smart grid according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a neural network according to an embodiment of the present invention;
fig. 4a is a block diagram of a management apparatus of a smart grid according to an embodiment of the present invention;
FIG. 4b is a block diagram of a refinement module of the reconstruction module provided in the embodiment of FIG. 4 a;
FIG. 4c is a block diagram of a refinement module of the generation module provided in the embodiment of FIG. 4 b;
fig. 5 is an architecture diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, features and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The technical problem that high-frequency data are difficult to obtain in the prior art, and the estimation frequency of the smart grid is affected is solved.
In order to solve the technical problem, the invention provides a management method and device of a smart grid, an electronic device and a storage medium.
Referring to fig. 1, a management system architecture diagram of a smart grid according to an embodiment of the present invention is shown, where the management system mainly includes: the intelligent management system comprises a networking device, a measuring device, a data processing device and an error eliminating device, wherein the networking device is used for enabling the intelligent management system to be networked with the external Internet so as to realize the networking state of the intelligent power grid; the measuring device is used for collecting low-frequency data according to a preset sampling frequency, and can receive a collection instruction issued by a worker or a manager through a networking function before collection and carry out measurement based on the collection instruction. The data processing device is used for receiving the low-frequency data collected by the measuring device and performing data processing, and specifically, the data processing device may include but is not limited to: the device comprises a reconstruction module and an estimation module, wherein the reconstruction module is used for receiving low-frequency data collected by a measuring device and reconstructing the received low-frequency data to obtain target high-frequency data; the estimation module carries out state estimation on the smart power grid based on the high-frequency data obtained through reconstruction; error elimination means are used to detect anomalous data and eliminate apparent errors in the measurements. Through the intelligent power grid management system, the low-frequency data can be reconstructed by utilizing the neural network to obtain the target high-frequency data, the direct acquisition and transmission of the high-frequency data are avoided, the phenomena of loss and falsification of the acquired high-frequency data are avoided, the state estimation is further carried out on the intelligent power grid based on the target high-frequency data, the intelligent power grid is managed according to the state estimation structure, more hidden problems in the intelligent power grid are discovered in time, the safety and the efficiency of the system are improved, and the reliability is realized.
Referring to fig. 2, a flowchart of steps of a management method for a smart grid according to an embodiment of the present invention is shown, where the embodiment of the present invention provides a management method for a smart grid, the method includes the following steps:
step S201: and acquiring low-frequency data of the specific meter, wherein the low-frequency data is initial data acquired to the smart grid according to a set sampling frequency.
The acquisition frequency refers to the frequency of acquired low-frequency data, the low-frequency data are incomplete data, namely the received original measurement data are incomplete low-frequency data, the specific meter samples only the power grid according to the set sampling frequency to obtain initial low-frequency data, and the initial low-frequency data are the data characteristics of an initialized zero vector; the low frequency data is displayed or transmitted by a specific meter after the acquisition is finished. The low-frequency data may affect the management strategy of the smart grid management system, for example, the management can be performed only in a low-frequency decision mode, and thus, the efficiency and the security of the smart grid are affected.
Step S202: and reconstructing the collected low-frequency data by utilizing a neural network based on a preset super-resolution perception model to obtain reconstructed target high-frequency data.
In order to realize high-frequency decision of the smart grid, more frequent state estimation needs to be realized by using high-frequency data so as to find more hidden problems, and therefore, the safety and reliability of the system are improved.
Since the input of the state estimation is a vector of values measured by a plurality of specific meters at a specific time point, the specific meters can be measurement devices or acquisition devices for measuring or acquiring the intelligent network, and the like, and are not limited herein; due to the influence of data transmission or communication failure or network attack, the data measured by high specific meters can be lost or tampered, and the data lost by each specific meter can be recovered, so that the low-frequency data can be restored into high-frequency data. The method is based on a super-resolution sensing model, and the high-frequency data can be acquired by using the existing low-frequency data.
Firstly, a super-resolution perception model is established, and the calculation formula of the super-resolution perception model is as follows:
La＝↓aH+e
wherein L isaLow-frequency data collected for a specific meter, H represents original high-frequency data, e represents noise vector corresponding to lost information, ↓aRepresents a down-sampling function, a represents a down-sampling factor; for example, assuming that the original high frequency data H is a measurement vector with a sampling frequency of 60Hz, if a is equal to 10, then LaIs a measurement vector with a sampling frequency of 6 Hz.
Further, the super-resolution sensing model can be regarded as a maximum a posteriori estimation model, in the maximum a posteriori estimation, the high frequency data aiming at prediction or estimation is made to be maximum, and the calculation formula of the maximum a posteriori estimation model is as follows:
wherein the content of the first and second substances,
representing estimated high frequency data calculated by the neural network,
representing the posterior probability at which the estimated high frequency data is obtained, P (L)a,↓βH) Representing likelihood functions, ↓βH represents that original high-frequency data is sampled by taking beta as a down-sampling factor, and P (↓βH) Representation ↓βThe prior probability of H. The prior probability is mainly related to the error of the original sampling data caused by a specific meter, the error can be understood as lost data or noise, and the prior probability follows a Gaussian distribution because the measured noise follows a normal distribution.
In one embodiment of step S202, the method includes:
s2021: generating high-frequency data according to the low-frequency data and the loss information measured in advance by using a neural network based on a preset super-resolution sensing model;
s2022: and updating the neural network by using the high-frequency data to acquire target high-frequency data.
In particular, a set of raw data is given as input to a super-resolution perceptual model, such as the received raw low-frequency data collected, to obtain two sets of down-sampled data, respectively low-frequency data sets LaDownsampling of high frequency data sets HβAnd corresponding down-sampling factors a and beta, wherein the super-resolution perception model takes low-frequency data as input, generates a group of estimated high-frequency data and enables the estimated high-frequency data to be close to the real down-sampled original high-frequency data as much as possible.
Further, in a detailed implementation of the foregoing embodiment, the step of generating the high-frequency data according to the low-frequency data and the pre-acquired loss information by using a neural network based on a preset super-resolution sensing model includes:
extracting low-frequency features in pre-measured low-frequency data according to a preset time sequence rule by utilizing three one-dimensional convolution layers of a neural network; performing information completion on the high-frequency features corresponding to the low-frequency features by using a residual error structure of the neural network to obtain high-frequency feature vectors after the information completion; and reconstructing the high-frequency characteristic vector according to a preset subsequence by using three one-dimensional convolution layers of the neural network to obtain high-frequency data.
Referring to fig. 3, a schematic diagram of a neural network structure provided in this embodiment is shown; specifically, the structure 300 of the neural network at least includes three one-dimensional convolutional layers 301, a residual structure 302 and another part of three one-dimensional convolutional layers 303, where the residual structure includes a large global residual connection and 22 local residual blocks, the local residual block is a super-resolution sensing module, and k represents the number of the super-resolution sensing modules, and then k is equal to 22; extracting the characteristics of the low-frequency data by using the three one-dimensional convolutional layers, and performing information completion on the high-frequency characteristics corresponding to the low-frequency characteristics by using a residual structure to obtain high-frequency characteristic vectors after the information completion; and reconstructing the high-frequency characteristic vector according to a preset subsequence by using three one-dimensional convolution layers of the neural network to obtain high-frequency data.
It should be noted that, this embodiment defines a time sequence, senses the system state by processing the measurement vector z corresponding to the received low frequency data, and represents the kth row value of the measurement vector z and the first measurement value of the kth specific meter in a discrete sequence, where the formula of the first measurement value is as follows:
wherein z isk(t) denotes the first measured value of the kth special meter at the time t, fk(t) represents a first measurement function of the kth specific meter at time t;
further, the measurement z may be caused by low sampling frequency, limited communication channel capacity, communication error, or even network attack of the conventional special meterk(t) partial data loss, i.e. loss of information, and therefore the present embodiment also defines a comparison with the first measurement value zk(t) obtaining a second measured value from the new time series of different time instants t, wherein the calculation formula of the second measured value is as follows:
wherein the content of the first and second substances,
representing a second measurement, g, of the kth particular meter at time tk(t) represents a second measurement function of the kth specific meter at time t;
in addition, since the high frequency data is complete data, the embodiment further defines a high frequency time sequence for representing the high frequency feature vector corresponding to the high frequency data, specifically, using Zk(t) represents a polymerization time series, and the polymerization time series is represented by
The method comprises the following specific steps:
wherein Z isk(t) represents the high frequency characteristic vector value of the kth specific meter at time t.
It should be noted that the aggregation time series may be understood as a correspondence relationship between low-frequency data and high-frequency data, or a correspondence relationship between low-frequency features and high-frequency features.
Further, the step of performing information completion on the high-frequency features corresponding to the low-frequency features by using a residual error structure of the neural network to obtain high-frequency feature vectors after the information completion comprises: generating a high-frequency feature corresponding to the low-frequency feature according to a preset corresponding relation between the low-frequency data and the high-frequency data by utilizing a residual error structure of the neural network; and performing information completion on the high-frequency characteristics according to the lost information to obtain high-frequency characteristic vectors after the information completion.
Specifically, before data reconstruction is performed, a high-frequency feature vector needs to be acquired, and the high-frequency feature vector can generate a high frequency by using a low-frequency feature according to the preset corresponding relationship between the low-frequency data and the high-frequency data; and then, carrying out information completion on the high-frequency characteristic by using the lost information, namely the noise of the meter, so as to obtain a high-frequency characteristic vector after the information completion. It should be noted that, in terms of the neural network structure, the global residual connection is used to force the neural network to learn the lost information, rather than forming the noise signal (lost information) itself; performing information completion by using the local residual block to obtain a closer high-frequency characteristic; it will be appreciated that, after completion of the information, high resolution features are obtained that contain more details of the system mode, resulting in better performance. The local residual block provides the possibility to train a deeper network.
In another embodiment of this step S202, after obtaining the high frequency data, step S2022 includes:
s20221: and determining an error loss function value by using the high-frequency data, wherein the calculation formula of the loss function is as follows:
where MSE represents the error loss function value, β is the down-sampling factor,
representing predicted high-frequency data, H, output by a neural network modeliβThe method comprises the steps of sampling original high-frequency data by taking beta as a down-sampling factor to obtain sampled high-frequency data, wherein N represents the number of estimated high-frequency data, and i represents a natural number.
And calculating an error loss function value by utilizing the estimated high-frequency data and the sampled high-frequency data, wherein the error loss function value is a value obtained by calculating a mean square error function, and whether the high-frequency data obtained by calculation of the neural network accords with the expected sampled high-frequency data can be analyzed through the error loss function value.
And determining whether to update the neural network by using the calculated loss function value,
s20222: and determining whether updating of the neural network is finished or not according to the error loss function value so as to obtain target high-frequency data.
Specifically, step S20222 includes:
step S202221: comparing the error loss function value with a preset error loss function threshold; it should be noted that, according to the comparison result between the error loss function value and the preset error loss function threshold, the next step is determined, wherein if the error loss function value is greater than the preset error loss function threshold, step S202222 is executed, and then step S202223 is executed.
Step S202222: if the error loss function value is larger than the preset error loss function threshold, determining to update the neural network, and stopping updating the neural network until the error loss function value is smaller than or equal to the preset error loss function threshold;
wherein the process of determining to update the neural network is as follows:
determining a weight coefficient and a bias coefficient of the neural network by utilizing a preset first moment estimation and a preset second moment estimation;
the first moment estimate is calculated as follows:
υdW:＝β1υdW+(1-β1)dW
the second moment estimate is calculated as follows:
υdb:＝β1υdb+(1-β1)db
the calculation formula of the weight coefficient and the bias coefficient is as follows:
SdW:＝β2SdW+(1-β2)dW2
Sdb:＝β2Sdb+(1-β2)db2
wherein, beta1And beta2Indicating an exponential decay rate of momentum, preferably, beta1Value of 0.9, beta2The value of the amount of the active carbon is 0.999,
and
is represented by beta1And beta2The power, upsilon, of the current time step tdWAnd upsilondbExpressed as the gradient with deviation, SdWAnd SdbAn exponentially decaying average of the squares of the gradients representing the presence of a deviation,
and
w represents a weight coefficient of the neural network and b represents a bias coefficient of the neural network for the estimated value after deviation correction;
and
indicating the learning rate.
Specifically, the neural network parameter optimization algorithm provided in this embodiment adjusts the parameter values of the deep neural network by using the adaptive moment estimation algorithm for the nonlinear low-frequency data, so as to obtain the data characteristics. Obtaining a gradient and an exponential decay average value of the square of the gradient by calculating a first moment estimation and a second moment estimation in the self-adaptive moment estimation; and further calculating a weight coefficient and a bias coefficient of the neural network for correcting the first moment estimation and the second moment estimation to offset data deviation caused by the initialized zero vector, namely updating parameters of the neural network. In addition, the algorithm updates the descending direction according to the exponential decay average value of the gradient, divides the square exponential decay average value of the gradient by the learning rate to obtain faster convergence speed and less oscillation, so that the high-frequency data calculated or estimated by the neural network is as close to the value of the original high-frequency data as possible, namely the difference between the high-frequency data obtained by the updated neural network and the original high-frequency data is smaller, and the high-frequency data obtained by the updated neural network is used as the target high-frequency data to obtain more accurate high-frequency data so as to provide data integrity.
Further, after the weight coefficient and the bias coefficient of the neural network are obtained through calculation, the neural network is updated by using the weight coefficient and the bias coefficient. Specifically, by setting an error loss function threshold in advance, the error loss function threshold is used to determine whether to perform updating of the neural network. Comparing the obtained error loss function value with a preset error loss function threshold, determining to update the neural network when the error loss function value is greater than or equal to the error loss function threshold, calculating high-frequency data by using the updated neural network to obtain updated high-frequency data, wherein if the error loss function value is calculated by using the updated high-frequency data and compared with the preset error loss function value, when the error loss function value is still greater than the preset error loss function threshold, continuing to update the neural network, and circulating until the error loss function value corresponding to the high-frequency data obtained by calculating by using the updated neural network is less than the preset error loss function threshold, namely until the error loss function value corresponding to the updated high-frequency data is less than the preset error loss function threshold, the update is stopped.
Step S202223: and if the error loss function value is smaller than or equal to the preset error loss function threshold, stopping updating the neural network, and calculating target high-frequency data according to the updated neural network.
Wherein, the step of stopping updating the neural network and calculating the target high-frequency data according to the updated neural network comprises the following steps:
after the neural network is stopped to be updated, determining an average absolute error by using the high-frequency data corresponding to the updated neural network, wherein a calculation formula of the average absolute error is as follows:
wherein MAPE represents the mean absolute error,
representing estimated high frequency data, HiβRepresenting sampled high frequency data, N representing the number of high frequency data, i representing a natural number;
and determining the signal-to-noise ratio by using the updated high-frequency data corresponding to the neural network, wherein the signal-to-noise ratio calculation formula is as follows:
where, SNR represents the signal-to-noise ratio,
representing estimated high frequency data, HiβIndicating miningAnd sampling high-frequency data, wherein N represents the number of estimated high-frequency data, and i represents a natural number.
Further, the neural network is determined to be updated according to the average absolute error and the signal-to-noise ratio, so that target high-frequency data corresponding to the updated neural network are obtained.
The average absolute error and the signal-to-noise ratio are calculated by utilizing the estimated high-frequency data and the sampled high-frequency data, and the updated neural network is evaluated by utilizing the obtained average absolute error and the obtained signal-to-noise ratio respectively so as to determine whether the updating of the neural network is complete or not. Specifically, since the related high-frequency data is related to the super-resolution model, the average absolute error and the signal-to-noise ratio calculated by the related high-frequency data can be used for evaluating the performance of the super-resolution model so as to determine whether to update the neural network, and the reliability is achieved. For example, the evaluation process may be: comparing the average absolute error with a preset average absolute error threshold, wherein if the average absolute error meets the preset average absolute error threshold condition, the average absolute error is less than or equal to the preset average absolute error threshold; and comparing the signal-to-noise ratio with a preset signal-to-noise ratio threshold, wherein the signal-to-noise ratio is smaller than or equal to the preset signal-to-noise ratio threshold. And (4) determining whether the neural network is updated or not by setting the evaluation conditions and combining the evaluation results of the average absolute error and the signal-to-noise ratio. It should be noted that the performance of the super-resolution model (neural network) may be evaluated by using the average absolute error MAPE alone, and the performance of the super-resolution model (neural network) may also be evaluated by using the signal-to-noise ratio alone, which is not further described in this embodiment. Furthermore, after determining whether to update the neural network, the target high-frequency data is acquired, so that the direct acquisition and transmission of the high-frequency data are avoided, the phenomena of loss and falsification of the acquired high-frequency data are avoided, the integrity of the data is provided, and the reliability is realized.
Step S203: and performing state estimation on the smart grid by using the target high-frequency data so as to manage the smart grid.
The state estimation is an indispensable step of the management step of the smart grid, and the state estimation is usually performed by performing data processing based on the acquired or received collected low-frequency data, and performing state estimation after the data processing. According to the method, the system state vector is calculated, and the estimated value of the system state variable is made to be as close to the actual value as possible by utilizing a maximum likelihood estimation method.
Specifically, the state estimation process can be regarded as generalized load flow calculation, and is represented by the following model:
z＝h(x)+e
where z represents the vector of the measured low frequency data, assuming z ∈ [ m,1 ]; x represents a vector of a state variable of the smart grid, and x belongs to [ n,1 ]; h (x) represents a functional relationship between the measured low-frequency data and the state variable of the smart grid; e represents the noise vector, e ∈ [ m,1 ]; wherein m is more than n.
In the state estimation process, in order to make the estimated value of the system state variable closer to the actual value, a maximum likelihood estimation mode is adopted, and a maximum likelihood estimation model or a calculation formula is as follows:
wherein the content of the first and second substances,
indicating at target state estimation
The maximum probability, p (z), is the probability distribution density function of the low frequency data vector z.
Furthermore, the measured noise vector or the corresponding value obeys normal distribution, and the state estimation method adopts a weighted least squares method, aiming at obtaining the extreme value of the target state function, and the specific calculation formula is as follows:
Jw(x)＝[z-h(x)]TW[z-h(x)]
wherein, Jw(x) Representing the objective state function, z represents the vector of the measured low frequency data, assuming z ∈ [ m,1]](ii) a x represents a vector of state variables of the smart grid, x belongs to [ n,1]](ii) a h (x) represents a functional relationship between the measured low-frequency data and the state variable of the smart grid; w ═ R-1And R is a diagonal matrix of measurement error variances.
Further, after obtaining an extreme value of the target state function, the extreme value is used as a state estimation value, and the state estimation value is used as a basis for managing the smart grid. If the corresponding relation table of the state estimation and the strategy is preset, the preset corresponding relation table of the state estimation and the strategy is searched for the value of the state estimation to obtain a corresponding target strategy, and the smart grid is managed according to the target strategy.
The invention provides a management method of a smart power grid, which comprises the following steps: acquiring low-frequency data of a specific meter, wherein the low-frequency data is initial data acquired to the smart grid according to a set sampling frequency; reconstructing the collected low-frequency data by utilizing a neural network based on a preset super-resolution perception model to obtain reconstructed target high-frequency data; and performing state estimation on the smart grid by using the target high-frequency data so as to manage the smart grid. According to the scheme, low-frequency data in the intelligent network are collected, the neural network is utilized to reconstruct the low-frequency data to obtain target high-frequency data based on a preset super-resolution perception model, the integrity of the data is provided, the direct collection and transmission of the high-frequency data are avoided, the phenomenon that the collected high-frequency data are lost and tampered is avoided, more accurate and frequent state estimation is further carried out on the intelligent power grid based on the target high-frequency data, the intelligent power grid is managed according to a state estimation structure, more hidden problems in the intelligent power grid are discovered in time, the safety and the efficiency of the system are improved, and the reliability is achieved.
Referring to fig. 4a, fig. 4a is a block diagram of a management apparatus of a smart grid according to an embodiment of the present invention, the management apparatus of the smart grid corresponding to an execution subject processor of a management method of the smart grid, the apparatus 400 including: an acquisition module 401, a reconstruction module 402 and an estimation module 403;
the acquisition module 401 is configured to acquire low-frequency data of a specific meter, where the low-frequency data is initial data acquired to a smart grid according to a set sampling frequency;
a reconstruction module 402, configured to reconstruct the acquired low-frequency data by using a neural network based on a preset super-resolution sensing model, so as to obtain reconstructed target high-frequency data;
and an estimating module 403, configured to perform state estimation on the smart grid by using the target high-frequency data to manage the smart grid.
The invention provides a management device of a smart grid, which comprises: an acquisition module 401, a reconstruction module 402 and an estimation module 403. The acquisition module 401 is configured to acquire low-frequency data of a specific meter, where the low-frequency data is initial data acquired to a smart grid according to a set sampling frequency; the reconstruction module 402 is configured to reconstruct the acquired low-frequency data by using a neural network based on a preset super-resolution sensing model to obtain reconstructed target high-frequency data; the estimation module 403 is configured to perform state estimation on the smart grid by using the target high-frequency data to manage the smart grid. According to the scheme, low-frequency data in the intelligent network are collected, the neural network is utilized to reconstruct the low-frequency data to obtain target high-frequency data based on a preset super-resolution perception model, the integrity of the data is provided, the direct collection and transmission of the high-frequency data are avoided, the phenomenon that the collected high-frequency data are lost and tampered is avoided, more accurate and frequent state estimation is further carried out on the intelligent power grid based on the target high-frequency data, the intelligent power grid is managed according to a state estimation structure, more hidden problems in the intelligent power grid are discovered in time, the safety and the efficiency of the system are improved, and the reliability is achieved.
Further, referring to fig. 4b, which is a block diagram of a refinement module of a part of modules provided in the embodiment of fig. 4a, the reconstruction module 402 may be further subdivided into: the generating module 4021 and the updating module 4022, but are not limited thereto.
The generating module 4021 is configured to generate high-frequency data according to the low-frequency data and the loss information measured in advance by using a neural network based on a preset super-resolution sensing model.
The updating module 4022 is configured to update the neural network with the high-frequency data to obtain target high-frequency data.
Further, referring to fig. 4c, for a block diagram of another refinement module of part of the modules provided in the embodiment of fig. 4b, the generating module 4021 may include: an extraction module 40211, a completion module 40212, and a reconstruction sub-module 40213, but is not limited thereto.
The extracting module 40211 is configured to extract low-frequency features in the pre-measured low-frequency data according to a preset time sequence rule by using three one-dimensional convolution layers of the neural network.
The completion module 40212 is configured to perform information completion on the high-frequency features corresponding to the low-frequency features by using a residual structure of the neural network, so as to obtain high-frequency feature vectors after the information completion.
The reconstruction submodule 40213 is configured to reconstruct the high-frequency feature vector according to a preset subsequence by using three one-dimensional convolution layers of the neural network, so as to obtain high-frequency data.
It should be noted that, the management device for a smart grid provided in this embodiment is an item corresponding to the management method for a smart grid, and technical features of modules included in the management device for a smart grid provided in this embodiment are similar to or similar to the steps of the method described above, and for description of technical features of the management device, reference may be made to the description of the management method for a smart grid in the foregoing embodiment, which is not further described in this embodiment.
The present invention provides an electronic device, please refer to fig. 5, which is an architecture diagram of the electronic device according to an embodiment of the present invention, and the electronic device includes: the smart grid management method comprises a memory 501, a processor 502 and a communication bus 503, wherein the communication bus 503 is respectively connected with the memory 501 and the processor 502 in a communication mode, the memory 501 is coupled with the processor 502, a computer program is stored on the memory 501, and when the processor 502 executes the computer program, each step of the smart grid management method is realized.
For example, the computer program of the management method of the smart grid mainly includes: acquiring low-frequency data of a specific meter, wherein the low-frequency data is initial data acquired to the smart grid according to a set sampling frequency; reconstructing the collected low-frequency data by utilizing a neural network based on a preset super-resolution perception model to obtain reconstructed target high-frequency data; and performing state estimation on the smart grid by using the target high-frequency data so as to manage the smart grid. In addition, the computer program may also be divided into one or more modules, which are stored in the memory and executed by the processor to accomplish the present invention. One or more of the modules may be a series of computer program instruction segments capable of performing certain functions, the instruction segments being used to describe the execution of a computer program in a computing device. For example, the computer program may be divided into an acquisition module 401, a reconstruction module 402 and an estimation module 403 as shown in fig. 4 a.
The Processor 402 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The invention further provides a storage medium, which is a computer-readable storage medium, and a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps in the management method of the smart grid are implemented.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, a division of modules is merely a division of logical functions, and an actual implementation may have another division, for example, a plurality of modules or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or modules, and may be in an electrical, mechanical or other form.
Modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical modules, may be located in one place, or may be distributed on a plurality of network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
In addition, functional modules in the embodiments of the present invention may be integrated into one processing module, or each of the modules may exist alone physically, or two or more modules are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode.
The integrated module, if implemented in the form of a software functional module and sold or used as a separate product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
It should be noted that, for the sake of simplicity, the above-mentioned method embodiments are described as a series of acts or combinations, but those skilled in the art should understand that the present invention is not limited by the described order of acts, as some steps may be performed in other orders or simultaneously according to the present invention. Further, those skilled in the art will appreciate that the embodiments described in the specification are presently preferred and that no acts or modules are necessarily required of the invention.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
In the above description, for a person skilled in the art, there are variations on the specific implementation and application scope according to the ideas of the embodiments of the present invention, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (10)
1. A management method of a smart grid, the method comprising:
acquiring low-frequency data of a specific meter, wherein the low-frequency data is initial data acquired to a smart grid according to a set sampling frequency;
reconstructing the collected low-frequency data by utilizing a neural network based on a preset super-resolution perception model to obtain reconstructed target high-frequency data;
and performing state estimation on the smart grid by using the target high-frequency data so as to manage the smart grid.
2. The management method of the smart grid according to claim 1, wherein the step of reconstructing the collected low-frequency data by using a neural network based on a preset super-resolution sensing model to obtain reconstructed high-frequency data comprises:
generating high-frequency data according to the low-frequency data and the loss information measured in advance by using a neural network based on a preset super-resolution perception model;
the preset super-resolution perception model has the following calculation formula:
La＝↓aH+e
wherein L isaFor the collected low-frequency data, H represents the original high-frequency data, e represents the noise vector corresponding to the lost information, ↓aRepresents a down-sampling function, a represents a down-sampling factor;
and updating the neural network by using the high-frequency data to acquire target high-frequency data.
3. The smart grid management method according to claim 2, wherein the step of generating high-frequency data from the low-frequency data and the pre-acquired loss information by using a neural network based on a preset super-resolution sensing model comprises:
extracting low-frequency features in pre-measured low-frequency data according to a preset time sequence rule by utilizing three one-dimensional convolution layers of a neural network;
performing information completion on the high-frequency features corresponding to the low-frequency features by using a residual error structure of the neural network to obtain high-frequency feature vectors after the information completion;
and reconstructing the high-frequency characteristic vector according to a preset subsequence by using the three one-dimensional convolution layers of the neural network to obtain high-frequency data.
4. The management method of the smart grid according to claim 3, wherein the step of performing information completion on the high-frequency features corresponding to the low-frequency features by using a residual structure of the neural network to obtain information-completed high-frequency feature vectors includes:
generating a high-frequency feature corresponding to the low-frequency feature according to a preset corresponding relation between low-frequency data and high-frequency data by using a residual error structure of the neural network;
and performing information completion on the high-frequency features according to the lost information to obtain the high-frequency feature vectors after the information completion.
5. The method for managing the smart grid according to any one of claims 2 to 4, wherein the step of updating the neural network with the high frequency data to obtain the target high frequency data comprises:
and determining an error loss function value by using the high-frequency data, wherein the calculation formula of the loss function is as follows:
where MSE represents the error loss function value, β is the down-sampling factor,
representing predicted high-frequency data, H, output by a neural network modeliβThe method comprises the steps of representing sampling high-frequency data obtained by sampling original high-frequency data by taking beta as a down-sampling factor, wherein N represents the number of estimated high-frequency data, and i represents a natural number;
and determining whether updating of the neural network is finished or not according to the error loss function value so as to obtain target high-frequency data.
6. The method for managing a smart grid according to claim 5, wherein the step of determining whether updating the neural network to obtain the target high frequency data is completed according to the error loss function value comprises:
comparing the error loss function value with a preset error loss function threshold;
if the error loss function value is larger than the preset error loss function threshold, determining to update the neural network until the error loss function value is smaller than or equal to the preset error loss function threshold;
and if the error loss function value is smaller than or equal to the preset error loss function threshold, stopping updating the neural network, and calculating target high-frequency data according to the updated neural network.
7. The method for managing the smart grid according to claim 6, wherein the step of stopping updating the neural network and calculating the target high-frequency data according to the updated neural network comprises:
after the neural network is stopped to be updated, determining an average absolute error by using the high-frequency data corresponding to the updated neural network, wherein a calculation formula of the average absolute error is as follows:
wherein MAPE represents the mean absolute error,
representing estimated high frequency data, HiβRepresenting sampled high frequency data, N representing the number of high frequency data, i representing a natural number;
and determining a signal-to-noise ratio by using the updated high-frequency data corresponding to the neural network, wherein the signal-to-noise ratio calculation formula is as follows:
where, SNR represents the signal-to-noise ratio,
representing estimated high frequency data, HiβRepresenting sampled high-frequency data, N representing the number of estimated high-frequency data, and i representing a natural number;
and determining that the neural network is updated according to the average absolute error and the signal-to-noise ratio so as to obtain target high-frequency data corresponding to the updated neural network.
8. A management device of a smart grid, comprising:
the acquisition module is used for acquiring low-frequency data of a specific meter, wherein the low-frequency data is initial data acquired to the smart grid according to a set sampling frequency;
the reconstruction module is used for reconstructing the collected low-frequency data by utilizing a neural network based on a preset super-resolution perception model to obtain reconstructed target high-frequency data;
and the estimation module is used for carrying out state estimation on the smart grid by utilizing the target high-frequency data so as to manage the smart grid.
9. An electronic device, comprising: the smart grid management system comprises a memory, a processor and a communication bus, wherein the communication bus is respectively connected with the memory and the processor in a communication mode, the memory is coupled with the processor, and the memory stores a computer program which is executed by the processor to realize each step in the smart grid management method according to any one of claims 1 to 7.
10. A storage medium, which is a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the computer program implements the steps in the management method of the smart grid according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011285277.5A CN112396535A (en) | 2020-11-17 | 2020-11-17 | Management method, device, equipment and storage medium of smart power grid |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011285277.5A CN112396535A (en) | 2020-11-17 | 2020-11-17 | Management method, device, equipment and storage medium of smart power grid |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112396535A true CN112396535A (en) | 2021-02-23 |
Family
ID=74600571
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011285277.5A Withdrawn CN112396535A (en) | 2020-11-17 | 2020-11-17 | Management method, device, equipment and storage medium of smart power grid |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112396535A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113596020A (en) * | 2021-07-28 | 2021-11-02 | 深圳供电局有限公司 | Smart grid false data injection attack vulnerability detection method |
| CN114419339A (en) * | 2022-03-30 | 2022-04-29 | 南方电网数字电网研究院有限公司 | Method and device for training data reconstruction model based on electric power portrait |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705699A (en) * | 2019-10-18 | 2020-01-17 | 厦门美图之家科技有限公司 | Super-resolution reconstruction method and device, electronic equipment and readable storage medium |
| CN111798491A (en) * | 2020-07-13 | 2020-10-20 | 哈尔滨工业大学 | Maneuvering target tracking method based on Elman neural network |
-
2020
- 2020-11-17 CN CN202011285277.5A patent/CN112396535A/en not_active Withdrawn
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705699A (en) * | 2019-10-18 | 2020-01-17 | 厦门美图之家科技有限公司 | Super-resolution reconstruction method and device, electronic equipment and readable storage medium |
| CN111798491A (en) * | 2020-07-13 | 2020-10-20 | 哈尔滨工业大学 | Maneuvering target tracking method based on Elman neural network |
Non-Patent Citations (1)
| Title |
|---|
| GAOQI LIANG 等: "Super Resolution Perception for Improving Data Completeness in Smart Grid State Estimation", 《ENGINEERING》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113596020A (en) * | 2021-07-28 | 2021-11-02 | 深圳供电局有限公司 | Smart grid false data injection attack vulnerability detection method |
| CN114419339A (en) * | 2022-03-30 | 2022-04-29 | 南方电网数字电网研究院有限公司 | Method and device for training data reconstruction model based on electric power portrait |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106505557B (en) | Remote measurement error identification method and device | |
| CN114297036B (en) | Data processing method, device, electronic equipment and readable storage medium | |
| CN116703183B (en) | Novel power distribution network system power load assessment method | |
| US5627760A (en) | Method and apparatus for real time recursive parameter energy management system | |
| CN112396535A (en) | Management method, device, equipment and storage medium of smart power grid | |
| Cheng et al. | Enhanced state estimation and bad data identification in active power distribution networks using photovoltaic power forecasting | |
| CN111368259B (en) | Distribution network voltage data missing filling method and related device | |
| CN114978956B (en) | Method and device for detecting abnormal mutation points of performance of intelligent city network equipment | |
| CN113409166A (en) | XGboost model-based method and device for detecting abnormal electricity consumption behavior of user | |
| Park et al. | Missing-insensitive short-term load forecasting leveraging autoencoder and LSTM | |
| CN114290960A (en) | Method and device for acquiring battery health degree of power battery and vehicle | |
| CN117290679A (en) | Running state detection method and device of current transformer and electronic equipment | |
| CN117407770A (en) | High-voltage switch cabinet fault mode classification and prediction method based on neural network | |
| CN115494349B (en) | Active power distribution network single-phase earth fault section positioning method | |
| CN111350638A (en) | Method and device for calculating power generation loss of wind generating set | |
| CN116298670A (en) | Intelligent fault positioning method and system suitable for multi-branch distribution line | |
| CN115796338A (en) | Photovoltaic power generation power prediction model construction and photovoltaic power generation power prediction method | |
| US20220046039A1 (en) | Method, device, and computer program product for abnormality detection | |
| CN117591983B (en) | Multi-index anomaly detection method and device, electronic equipment and storage medium | |
| Chu et al. | A relaxed support vector data description algorithm based fault detection in distribution systems | |
| CN113395031B (en) | Complex system fault diagnosis method and system based on multistage model | |
| CN117411190B (en) | Power distribution network topology identification method, equipment, system and medium based on multi-source information | |
| CN116029621B (en) | Method and device for evaluating operation of electric energy meter in transformer area, electronic equipment and storage medium | |
| CN117009902A (en) | Data detection method, device, equipment and storage medium | |
| CN117874429A (en) | Interconnected power system boundary data normal correction and recursion optimizing method |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210223 |
|
| WW01 | Invention patent application withdrawn after publication |