CN112036682A - Early warning method and device for frequent power failure - Google Patents
Early warning method and device for frequent power failure Download PDFInfo
- Publication number
- CN112036682A CN112036682A CN202010664816.XA CN202010664816A CN112036682A CN 112036682 A CN112036682 A CN 112036682A CN 202010664816 A CN202010664816 A CN 202010664816A CN 112036682 A CN112036682 A CN 112036682A
- Authority
- CN
- China
- Prior art keywords
- power failure
- user
- historical
- prediction model
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 37
- 239000011159 matrix material Substances 0.000 claims abstract description 70
- 238000012549 training Methods 0.000 claims abstract description 60
- 230000007613 environmental effect Effects 0.000 claims abstract description 18
- 239000013598 vector Substances 0.000 claims description 40
- 238000012360 testing method Methods 0.000 claims description 38
- 230000006870 function Effects 0.000 claims description 12
- 238000003062 neural network model Methods 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 7
- 239000000126 substance Substances 0.000 claims description 6
- 238000012544 monitoring process Methods 0.000 claims description 5
- 238000012423 maintenance Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- 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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Game Theory and Decision Science (AREA)
- Health & Medical Sciences (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method and a device for early warning of frequent power failure, wherein the method comprises the following steps: acquiring a power failure user with frequent power failure and performing frequent power failure marking; obtaining historical user data and historical environmental condition data of the power outage user during historical power outage; extracting characteristic data and constructing a historical characteristic matrix; inputting the historical characteristic matrix into a prediction model for training and learning until convergence to obtain a converged prediction model; constructing a real-time feature matrix; and inputting the real-time characteristic matrix into the converged prediction model for prediction to obtain a prediction result, and performing power failure early warning reminding on the frequent power failure user based on the prediction result. In the embodiment of the invention, the power failure risk which is possibly existed in the next time when the power failure is frequently used is predicted, and the power failure risk is accurately pushed to the user for early warning, so that the power failure is prevented in advance, the user loss is reduced, and the user experience is improved.
Description
Technical Field
The invention relates to the technical field of intelligent power grids, in particular to a method and a device for early warning of frequent power failure.
Background
In the prior art, the reason for frequent power failure of a user cannot be accurately known, so that the user cannot be accurately warned of the condition that power failure is likely to occur again in the future according to related data conditions, or corresponding lines and the like are overhauled in advance, the power failure frequency of the user is reduced, and the power utilization experience of the user is improved.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, and provides a method and a device for early warning of frequent power failure, which are used for predicting the power failure risk possibly existing next time when frequent power failure is used, and accurately pushing the power failure risk to a user for early warning, so that the power failure is prevented in advance, the user loss is reduced, and the user experience is improved.
In order to solve the technical problem, an embodiment of the present invention provides a method for warning frequent power outage, where the method includes:
obtaining a power failure user with frequent power failure, determining the reason of the power failure caused by the power failure user, and marking the power failure user with frequent power failure based on the reason of the power failure;
obtaining historical user data and historical environmental condition data of the power outage user during historical power outage;
extracting historical data features based on the frequent power failure mark, historical user data and historical environment condition data, and constructing a historical feature matrix based on the extracted historical data features;
inputting the historical characteristic matrix into a prediction model for training and learning until convergence, and obtaining a converged prediction model;
acquiring current user data set current environment data of the power failure user in real time, and constructing a real-time characteristic matrix based on the frequent power failure mark and the current user data set current environment data;
and inputting the real-time characteristic matrix into the converged prediction model for prediction to obtain a prediction result, and performing power failure early warning reminding on the frequent power failure user based on the prediction result.
Optionally, the obtaining of the power outage user with frequent power outage includes:
acquiring all users with power failure in a preset time period in a certain jurisdiction;
judging whether the power failure times of the user in a preset time period reach a power failure time threshold value or not;
and if so, confirming that the user is the power failure user with frequent power failure.
Optionally, the reason causing the power failure includes any one of a power supply equipment reason, an overload problem reason, a weather reason or an arrearage reason;
the determining the reason of the power failure caused by the power failure user comprises the following steps:
and determining the reason of the power failure caused by the power failure user based on the position information of the power failure user, the power fee payment time, the prompting record and the weather condition information of the power failure.
Optionally, the historical user data includes power load at the moment of power failure, equipment state information of the power supply equipment, line load information, and load information of the power supply equipment; the historical environmental condition data comprises weather condition information during power failure;
the historical data feature extraction based on the frequent power failure mark, the historical user data and the historical environmental condition data comprises the following steps:
and performing assignment representation on the frequent power failure mark, the historical user data and the historical environment condition data of each power failure user, and extracting corresponding assignments as historical data characteristics.
Optionally, the constructing a historical feature matrix based on the extracted historical data features includes:
and constructing a one-dimensional historical feature matrix of each power failure user based on the extracted historical data features.
Optionally, the prediction model is a deep neural network model;
the inputting the historical characteristic matrix into the prediction model for training and learning until convergence to obtain the converged prediction model comprises the following steps:
regularizing the parameter vector of each layer of the network in the prediction model to obtain a regularization item;
updating the loss function in the prediction model by using the regularization term to obtain an updated prediction model;
dividing the historical feature matrix into a training set and a testing set, inputting the training set into the updated prediction model for feature learning training, and obtaining a trained prediction model;
inputting the test set into the trained prediction model for testing to obtain a test result;
judging whether the test result reaches a preset target or not, if so, obtaining a convergent prediction model
If not, resetting the coefficients of all layer parameter vectors of the trained prediction model by adopting a back propagation algorithm, and after resetting, continuing training by using the training set data until convergence is reached or the training frequency reaches a threshold value.
Optionally, before the regularizing the parameter vector of each layer network in the prediction model, the regularizing includes:
compressing the output nodes of each layer network in the prediction model in proportion;
the ratio is two to one.
Optionally, the ratio of the training set to the test set is 9: 1.
Optionally, the reset formula for resetting the coefficients of all layer parameter vectors of the trained prediction model by using the back propagation algorithm is as follows:
wherein the content of the first and second substances,
for updated coefficients of the i-th and j-th parameter vectors in the trained prediction model, WijThe coefficients of the ith parameter vector and the jth parameter vector in the trained prediction model are represented, lr is the learning rate, weight _ decay is the weight attenuation coefficient, momentum is the impulse, and L is the updated loss function.
In addition, the embodiment of the invention also provides a warning device for frequent power failure, which comprises:
a marking module: the system comprises a power failure user for obtaining frequent power failure, determining the reason of the power failure caused by the power failure user, and marking the frequent power failure for the power failure user based on the reason of the power failure;
a historical data acquisition module: the power failure monitoring system is used for obtaining historical user data and historical environment condition data of the power failure user during historical power failure;
a feature matrix construction module: the historical data feature extraction module is used for extracting historical data features based on the frequent power failure mark, the historical user data and the historical environment condition data, and constructing a historical feature matrix based on the extracted historical data features;
training a learning module: the prediction model is used for inputting the historical characteristic matrix into the prediction model for training and learning until convergence, and the convergence prediction model is obtained;
a real-time matrix construction module: the real-time characteristic matrix is used for acquiring the current environment data of the current user data set of the power failure user in real time and constructing a real-time characteristic matrix based on the frequent power failure mark and the current environment data of the current user data set;
the prediction early warning module: and the real-time feature matrix is input into the converged prediction model for prediction to obtain a prediction result, and the power failure early warning reminding is carried out on the frequent power failure user based on the prediction result.
In the embodiment of the invention, whether the user is a frequent power failure user or not can be separated, when the user is the frequent power failure user, the users are effectively monitored, the power failure risk possibly existing next time of the users is predicted by a corresponding method, and the corresponding power failure risk is accurately pushed to the users in real time, so that the users can prevent power failure in advance, the user loss is reduced, and the user experience is improved; the power supply enterprise can effectively monitor various states of each power supply device, and the power utilization states and weather states of the users frequently having power failure, so that the users can timely know the corresponding device conditions, timely carry out device maintenance, reduce power failure from the source, provide a good power utilization experience for the users, and reduce the loss of the users caused by power failure.
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 flow chart illustrating a method for warning of frequent blackouts in an embodiment of the present invention;
fig. 2 is a schematic structural component diagram of the warning device for frequent power failure in the embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious 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.
Examples
Referring to fig. 1, fig. 1 is a flow chart illustrating an early warning method for frequent power outage according to an embodiment of the present invention.
As shown in fig. 1, a method for warning of frequent power failure includes:
s11: obtaining a power failure user with frequent power failure, determining the reason of the power failure caused by the power failure user, and marking the power failure user with frequent power failure based on the reason of the power failure;
in the embodiment of the present invention, the obtaining of the power outage user with frequent power outage includes: acquiring all users with power failure in a preset time period in a certain jurisdiction; judging whether the power failure times of the user in a preset time period reach a power failure time threshold value or not; and if so, confirming that the user is the power failure user with frequent power failure.
Further, the reason causing the power failure includes any one reason of a power supply equipment reason, an overload problem reason, a weather reason or an arrearage reason; the determining the reason of the power failure caused by the power failure user comprises the following steps: and determining the reason of the power failure caused by the power failure user based on the position information of the power failure user, the power fee payment time, the prompting record and the weather condition information of the power failure.
Specifically, all users who have had power failure within a certain jurisdiction within a certain time (historically) are obtained, and through the preset power failure time threshold, when the power failure time reaches the preset power failure time threshold within a specified time length, the user is considered as a frequent power failure user, and the frequent power failure user can be obtained.
Acquiring data information such as position information of the power failure user, electric charge payment time, whether collection urging record exists or not, weather condition information during power failure and the like, and performing matching judgment according to the data information so as to distinguish the power failure of the power failure user, which is equipment problem (equipment failure, overload caused by sudden change of weather, equipment trip caused by thunder, and the like), or power failure of self-reason (arrearage); judging whether the same power supply equipment (such as a transformer) is shared or not according to the position information of the power failure users, and considering frequent power failure caused by equipment failure firstly if the users of the same power supply equipment are all power failure users; if the users are not sharing the same power supply equipment, firstly, the electricity fee payment time of the users and whether the record of collection hastening exists are compared to determine whether the users are defaulting power failure users; when determining that the power failure user is not an owing power failure user, considering that the frequent power failure user is a frequent power failure caused by personal equipment failure or line failure, and needing to obtain the power load, weather condition information and the like of the power failure user during power failure; the causes of blackouts by blackout users who are not arrearage blackouts are analyzed to determine whether they are due to equipment failure, overload, or weather.
When the power failure users are determined to be frequent power failures caused by equipment faults, overload of electric loads, weather conditions or arrearages, the power failure users are marked with frequent power failures caused by the frequent power failures.
S12: obtaining historical user data and historical environmental condition data of the power outage user during historical power outage;
in the specific implementation process of the invention, the historical user data comprises the power load at the moment of power failure, equipment state information of the power supply equipment, line load information and load information of the power supply equipment; the historical environmental condition data includes weather condition information at the time of a power outage.
Specifically, the power utilization load, the equipment load and the weather conditions (including temperature, whether thunderstorm weather and the like) of the users with power failure (non-defaulting users) before historical power failure and at the moment of power failure are obtained, whether the equipment load exceeds the load borne by the equipment, whether the power utilization load of the users exceeds the overload load of the line where the users are located and whether the weather conditions are extreme weather (extreme conditions such as extremely high or extremely low temperature, thunderstorm weather and the like) are judged, and therefore the users can be determined to be frequent power failure caused by equipment failure, power utilization overload, weather conditions and the like.
S13: extracting historical data features based on the frequent power failure mark, historical user data and historical environment condition data, and constructing a historical feature matrix based on the extracted historical data features;
in a specific implementation process of the present invention, the extracting historical data features based on the frequent outage flag, the historical user data, and the historical environmental condition data includes: and performing assignment representation on the frequent power failure mark, the historical user data and the historical environment condition data of each power failure user, and extracting corresponding assignments as historical data characteristics.
Further, the constructing a historical feature matrix based on the extracted historical data features includes: and constructing a one-dimensional historical feature matrix of each power failure user based on the extracted historical data features.
Specifically, data features of each user with frequent power failure are extracted, and a one-dimensional feature matrix is constructed by the extracted features.
Extracting data such as power load, power supply equipment (equipment such as a transformer) equipment state information, line load, load of the power supply equipment, weather condition, power failure cause and the like of a frequent power failure user at the moment of frequent power failure; wherein, the electric load is represented by 0 in the matrix under the normal load, and is represented by 1 under the abnormal condition (exceeding the normal electric load); the power supply equipment state information is well represented by 0, generally represented by 1, and the difference is represented by 2; the line load is normally represented by 0 and the overload by 1; the load of the power supply equipment is normally represented by 0, and the overload is represented by 1; the weather condition is represented by 0, the temperature is over high or over low by 1, the thunderstorm day is represented by 3, and the rainy day (without thunder) is represented by 4; the cause of the power failure is represented by a power supply failure of 0, the line failure of 1, the weather cause of 3, and the cause of the user is represented by 4.
The historical data features are extracted through the rules, a one-dimensional historical feature matrix is constructed, the historical feature matrix is normalized to be within the range of 0-1, the redundancy rate of the data can be reduced, and calculation is facilitated.
S14: inputting the historical characteristic matrix into a prediction model for training and learning until convergence, and obtaining a converged prediction model;
in the specific implementation process of the invention, the prediction model is a deep neural network model; the inputting the historical characteristic matrix into the prediction model for training and learning until convergence to obtain the converged prediction model comprises the following steps: regularizing the parameter vector of each layer of the network in the prediction model to obtain a regularization item; updating the loss function in the prediction model by using the regularization term to obtain an updated prediction model; dividing the historical feature matrix into a training set and a testing set, inputting the training set into the updated prediction model for feature learning training, and obtaining a trained prediction model; inputting the test set into the trained prediction model for testing to obtain a test result; judging whether the test result reaches a preset target or not, if so, obtaining a convergent prediction model; if not, resetting the coefficients of all layer parameter vectors of the trained prediction model by adopting a back propagation algorithm, and after resetting, continuing training by using the training set data until convergence is reached or the training frequency reaches a threshold value.
Further, before the regularizing the parameter vectors of each layer network in the prediction model, the regularizing includes: compressing the output nodes of each layer network in the prediction model in proportion; the ratio is two to one.
Further, the ratio of the training set to the test set is 9: 1.
Further, the reset formula for resetting the coefficients of all layer parameter vectors of the trained prediction model by using the back propagation algorithm is as follows:
wherein the content of the first and second substances,
for trainingUpdated coefficients of the ith and jth parameter vectors in the subsequent prediction model, WijThe coefficients of the ith parameter vector and the jth parameter vector in the trained prediction model are represented, lr is the learning rate, weight _ decay is the weight attenuation coefficient, momentum is the impulse, and L is the updated loss function.
Specifically, the prediction model may be a deep neural network model, and after determining the neural network model as the prediction model, first, the output nodes of each layer network in the prediction model need to be compressed in proportion, which is generally one-half; firstly, regularization processing is carried out on parameter vectors of each layer of network in the prediction model, after regularization is carried out on vector parameters of all layers of the prediction model, regularization items are obtained, the obtained regularization items are used for updating the loss function, after updating is completed, a historical feature matrix constructed by using historical data is subjected to division of a training set and a test set, and the general division ratio is 9: 1; after a training set and a test set are divided, inputting a feature matrix in the training set into a prediction model for feature learning training, after one training is completed, testing the prediction model by using test data in the test set to obtain a test result, judging whether the test result is in a preset range, if so, the prediction model is a convergence model, if not, coefficients of all layer parameter vectors of the trained prediction model need to be reset by using a back propagation algorithm, and in the resetting process, a formula for resetting is provided:
wherein the content of the first and second substances,
for updated coefficients of the i-th and j-th parameter vectors in the trained prediction model, WijThe coefficients of the ith parameter vector and the jth parameter vector in the trained prediction model are represented, lr is the learning rate, weight _ decay is the weight attenuation coefficient, momentum is the impulse, and L is the updated loss function.
After updating with the back propagation algorithm, training continues with training data of the training set until convergence is reached or a training time threshold is reached.
S15: acquiring current user data set current environment data of the power failure user in real time, and constructing a real-time characteristic matrix based on the frequent power failure mark and the current user data set current environment data;
in the specific implementation process of the invention, the current environmental data of the current user data set of the power failure user is acquired, namely the current environmental data comprises the current power utilization load, the equipment state information of the power supply equipment, the line load information and the load information of the power supply equipment; the current environmental condition data includes current weather condition information.
Then, the real-time feature extraction is performed and the real-time feature matrix is constructed according to the manner in the step S13.
S16: and inputting the real-time characteristic matrix into the converged prediction model for prediction to obtain a prediction result, and performing power failure early warning reminding on the frequent power failure user based on the prediction result.
In the specific embodiment of the invention, the real-time characteristic matrix is input into a convergent prediction model for prediction to obtain a prediction result, and power failure early warning reminding is carried out on the frequent power failure user according to the prediction result; when the power failure user is in arrearage power failure and does not reach a preset payment period, monitoring whether the user pays the fee or not through a payment system in real time, reminding the user under the condition that the user does not pay the fee, wherein the reminding is carried out according to grades, the reminding when a payment notice is given is ordinary reminding, and the reminding is pushed through short messages or mails or WeChat and the like; the power failure early warning prompt is to send the power failure early warning prompt which is not paid to a user by means of a telephone and the like when power failure countdown is about to occur; when the power failure user is not due power failure and the prediction result is that power failure is possible to occur, the prediction result is pushed to the user in a power failure early warning mode.
In the embodiment of the invention, whether the user is a frequent power failure user or not can be separated, when the user is the frequent power failure user, the users are effectively monitored, the power failure risk possibly existing next time of the users is predicted by a corresponding method, and the corresponding power failure risk is accurately pushed to the users in real time, so that the users can prevent power failure in advance, the user loss is reduced, and the user experience is improved; the power supply enterprise can effectively monitor various states of each power supply device, and the power utilization states and weather states of the users frequently having power failure, so that the users can timely know the corresponding device conditions, timely carry out device maintenance, reduce power failure from the source, provide a good power utilization experience for the users, and reduce the loss of the users caused by power failure.
Examples
Referring to fig. 2, fig. 2 is a schematic structural component diagram of an early warning device for frequent power failure in an embodiment of the present invention.
As shown in fig. 2, an early warning device for frequent power failure includes:
the marking module 21: the system comprises a power failure user for obtaining frequent power failure, determining the reason of the power failure caused by the power failure user, and marking the frequent power failure for the power failure user based on the reason of the power failure;
in the embodiment of the present invention, the obtaining of the power outage user with frequent power outage includes: acquiring all users with power failure in a preset time period in a certain jurisdiction; judging whether the power failure times of the user in a preset time period reach a power failure time threshold value or not; and if so, confirming that the user is the power failure user with frequent power failure.
Further, the reason causing the power failure includes any one reason of a power supply equipment reason, an overload problem reason, a weather reason or an arrearage reason; the determining the reason of the power failure caused by the power failure user comprises the following steps: and determining the reason of the power failure caused by the power failure user based on the position information of the power failure user, the power fee payment time, the prompting record and the weather condition information of the power failure.
Specifically, all users who have had power failure within a certain jurisdiction within a certain time (historically) are obtained, and through the preset power failure time threshold, when the power failure time reaches the preset power failure time threshold within a specified time length, the user is considered as a frequent power failure user, and the frequent power failure user can be obtained.
Acquiring data information such as position information of the power failure user, electric charge payment time, whether collection urging record exists or not, weather condition information during power failure and the like, and performing matching judgment according to the data information so as to distinguish the power failure of the power failure user, which is equipment problem (equipment failure, overload caused by sudden change of weather, equipment trip caused by thunder, and the like), or power failure of self-reason (arrearage); judging whether the same power supply equipment (such as a transformer) is shared or not according to the position information of the power failure users, and considering frequent power failure caused by equipment failure firstly if the users of the same power supply equipment are all power failure users; if the users are not sharing the same power supply equipment, firstly, the electricity fee payment time of the users and whether the record of collection hastening exists are compared to determine whether the users are defaulting power failure users; when determining that the power failure user is not an owing power failure user, considering that the frequent power failure user is a frequent power failure caused by personal equipment failure or line failure, and needing to obtain the power load, weather condition information and the like of the power failure user during power failure; the causes of blackouts by blackout users who are not arrearage blackouts are analyzed to determine whether they are due to equipment failure, overload, or weather.
When the power failure users are determined to be frequent power failures caused by equipment faults, overload of electric loads, weather conditions or arrearages, the power failure users are marked with frequent power failures caused by the frequent power failures.
The history data acquisition module 22: the power failure monitoring system is used for obtaining historical user data and historical environment condition data of the power failure user during historical power failure;
in the specific implementation process of the invention, the historical user data comprises the power load at the moment of power failure, equipment state information of the power supply equipment, line load information and load information of the power supply equipment; the historical environmental condition data includes weather condition information at the time of a power outage.
Specifically, the power utilization load, the equipment load and the weather conditions (including temperature, whether thunderstorm weather and the like) of the users with power failure (non-defaulting users) before historical power failure and at the moment of power failure are obtained, whether the equipment load exceeds the load borne by the equipment, whether the power utilization load of the users exceeds the overload load of the line where the users are located and whether the weather conditions are extreme weather (extreme conditions such as extremely high or extremely low temperature, thunderstorm weather and the like) are judged, and therefore the users can be determined to be frequent power failure caused by equipment failure, power utilization overload, weather conditions and the like.
Feature matrix construction module 23: the historical data feature extraction module is used for extracting historical data features based on the frequent power failure mark, the historical user data and the historical environment condition data, and constructing a historical feature matrix based on the extracted historical data features;
in a specific implementation process of the present invention, the extracting historical data features based on the frequent outage flag, the historical user data, and the historical environmental condition data includes: and performing assignment representation on the frequent power failure mark, the historical user data and the historical environment condition data of each power failure user, and extracting corresponding assignments as historical data characteristics.
Further, the constructing a historical feature matrix based on the extracted historical data features includes: and constructing a one-dimensional historical feature matrix of each power failure user based on the extracted historical data features.
Specifically, data features of each user with frequent power failure are extracted, and a one-dimensional feature matrix is constructed by the extracted features.
Extracting data such as power load, power supply equipment (equipment such as a transformer) equipment state information, line load, load of the power supply equipment, weather condition, power failure cause and the like of a frequent power failure user at the moment of frequent power failure; wherein, the electric load is represented by 0 in the matrix under the normal load, and is represented by 1 under the abnormal condition (exceeding the normal electric load); the power supply equipment state information is well represented by 0, generally represented by 1, and the difference is represented by 2; the line load is normally represented by 0 and the overload by 1; the load of the power supply equipment is normally represented by 0, and the overload is represented by 1; the weather condition is represented by 0, the temperature is over high or over low by 1, the thunderstorm day is represented by 3, and the rainy day (without thunder) is represented by 4; the cause of the power failure is represented by a power supply failure of 0, the line failure of 1, the weather cause of 3, and the cause of the user is represented by 4.
The historical data features are extracted through the rules, a one-dimensional historical feature matrix is constructed, the historical feature matrix is normalized to be within the range of 0-1, the redundancy rate of the data can be reduced, and calculation is facilitated.
Training the learning module 24: the prediction model is used for inputting the historical characteristic matrix into the prediction model for training and learning until convergence, and the convergence prediction model is obtained;
in the specific implementation process of the invention, the prediction model is a deep neural network model; the inputting the historical characteristic matrix into the prediction model for training and learning until convergence to obtain the converged prediction model comprises the following steps: regularizing the parameter vector of each layer of the network in the prediction model to obtain a regularization item; updating the loss function in the prediction model by using the regularization term to obtain an updated prediction model; dividing the historical feature matrix into a training set and a testing set, inputting the training set into the updated prediction model for feature learning training, and obtaining a trained prediction model; inputting the test set into the trained prediction model for testing to obtain a test result; judging whether the test result reaches a preset target or not, if so, obtaining a convergent prediction model; if not, resetting the coefficients of all layer parameter vectors of the trained prediction model by adopting a back propagation algorithm, and after resetting, continuing training by using the training set data until convergence is reached or the training frequency reaches a threshold value.
Further, before the regularizing the parameter vectors of each layer network in the prediction model, the regularizing includes: compressing the output nodes of each layer network in the prediction model in proportion; the ratio is two to one.
Further, the ratio of the training set to the test set is 9: 1.
Further, the reset formula for resetting the coefficients of all layer parameter vectors of the trained prediction model by using the back propagation algorithm is as follows:
wherein the content of the first and second substances,
for updated coefficients of the i-th and j-th parameter vectors in the trained prediction model, WijThe coefficients of the ith parameter vector and the jth parameter vector in the trained prediction model are represented, lr is the learning rate, weight _ decay is the weight attenuation coefficient, momentum is the impulse, and L is the updated loss function.
Specifically, the prediction model may be a deep neural network model, and after determining the neural network model as the prediction model, first, the output nodes of each layer network in the prediction model need to be compressed in proportion, which is generally one-half; firstly, regularization processing is carried out on parameter vectors of each layer of network in the prediction model, after regularization is carried out on vector parameters of all layers of the prediction model, regularization items are obtained, the obtained regularization items are used for updating the loss function, after updating is completed, a historical feature matrix constructed by using historical data is subjected to division of a training set and a test set, and the general division ratio is 9: 1; after a training set and a test set are divided, inputting a feature matrix in the training set into a prediction model for feature learning training, after one training is completed, testing the prediction model by using test data in the test set to obtain a test result, judging whether the test result is in a preset range, if so, the prediction model is a convergence model, if not, coefficients of all layer parameter vectors of the trained prediction model need to be reset by using a back propagation algorithm, and in the resetting process, a formula for resetting is provided:
wherein the content of the first and second substances,
for updated coefficients of the i-th and j-th parameter vectors in the trained prediction model, WijThe coefficients of the ith parameter vector and the jth parameter vector in the trained prediction model are represented, lr is the learning rate, weight _ decay is the weight attenuation coefficient, momentum is the impulse, and L is the updated loss function.
After updating with the back propagation algorithm, training continues with training data of the training set until convergence is reached or a training time threshold is reached.
Real-time matrix construction module 25: the real-time characteristic matrix is used for acquiring the current environment data of the current user data set of the power failure user in real time and constructing a real-time characteristic matrix based on the frequent power failure mark and the current environment data of the current user data set;
in the specific implementation process of the invention, the current environmental data of the current user data set of the power failure user is acquired, namely the current environmental data comprises the current power utilization load, the equipment state information of the power supply equipment, the line load information and the load information of the power supply equipment; the current environmental condition data includes current weather condition information.
Then, the real-time feature extraction is performed according to the above-mentioned manner in the feature matrix construction module 23, and a real-time feature matrix is constructed.
The prediction and early warning module 26: and the real-time feature matrix is input into the converged prediction model for prediction to obtain a prediction result, and the power failure early warning reminding is carried out on the frequent power failure user based on the prediction result.
In the specific embodiment of the invention, the real-time characteristic matrix is input into a convergent prediction model for prediction to obtain a prediction result, and power failure early warning reminding is carried out on the frequent power failure user according to the prediction result; when the power failure user is in arrearage power failure and does not reach a preset payment period, monitoring whether the user pays the fee or not through a payment system in real time, reminding the user under the condition that the user does not pay the fee, wherein the reminding is carried out according to grades, the reminding when a payment notice is given is ordinary reminding, and the reminding is pushed through short messages or mails or WeChat and the like; the power failure early warning prompt is to send the power failure early warning prompt which is not paid to a user by means of a telephone and the like when power failure countdown is about to occur; when the power failure user is not due power failure and the prediction result is that power failure is possible to occur, the prediction result is pushed to the user in a power failure early warning mode.
In the embodiment of the invention, whether the user is a frequent power failure user or not can be separated, when the user is the frequent power failure user, the users are effectively monitored, the power failure risk possibly existing next time of the users is predicted by a corresponding method, and the corresponding power failure risk is accurately pushed to the users in real time, so that the users can prevent power failure in advance, the user loss is reduced, and the user experience is improved; the power supply enterprise can effectively monitor various states of each power supply device, and the power utilization states and weather states of the users frequently having power failure, so that the users can timely know the corresponding device conditions, timely carry out device maintenance, reduce power failure from the source, provide a good power utilization experience for the users, and reduce the loss of the users caused by power failure.
Those skilled in the art will appreciate that all or part of the steps in the methods of the above embodiments may be implemented by associated hardware instructed by a program, which may be stored in a computer-readable storage medium, and the storage medium may include: read Only Memory (ROM), Random Access Memory (RAM), magnetic or optical disks, and the like.
In addition, the method and the device for early warning of frequent blackout according to the embodiment of the present invention are described in detail, a specific example is adopted herein to explain the principle and the implementation manner of the present invention, and the description of the embodiment is only used to help understanding the method and the core idea of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (10)
1. A method for early warning of frequent power failure is characterized by comprising the following steps:
obtaining a power failure user with frequent power failure, determining the reason of the power failure caused by the power failure user, and marking the power failure user with frequent power failure based on the reason of the power failure;
obtaining historical user data and historical environmental condition data of the power outage user during historical power outage;
extracting historical data features based on the frequent power failure mark, historical user data and historical environment condition data, and constructing a historical feature matrix based on the extracted historical data features;
inputting the historical characteristic matrix into a prediction model for training and learning until convergence, and obtaining a converged prediction model;
acquiring current user data set current environment data of the power failure user in real time, and constructing a real-time characteristic matrix based on the frequent power failure mark and the current user data set current environment data;
and inputting the real-time characteristic matrix into the converged prediction model for prediction to obtain a prediction result, and performing power failure early warning reminding on the frequent power failure user based on the prediction result.
2. The warning method as claimed in claim 1, wherein the obtaining of the blackout user with frequent blackouts comprises:
acquiring all users with power failure in a preset time period in a certain jurisdiction;
judging whether the power failure times of the user in a preset time period reach a power failure time threshold value or not;
and if so, confirming that the user is the power failure user with frequent power failure.
3. The early warning method as claimed in claim 1, wherein the reason causing the power failure comprises any one reason of a power supply equipment reason, an overload problem reason, a weather reason or an arrearage reason;
the determining the reason of the power failure caused by the power failure user comprises the following steps:
and determining the reason of the power failure caused by the power failure user based on the position information of the power failure user, the power fee payment time, the prompting record and the weather condition information of the power failure.
4. The early warning method according to claim 1, wherein the historical user data comprises power loads at the moment of power failure, equipment state information of power supply equipment, line load information and load information of the power supply equipment; the historical environmental condition data comprises weather condition information during power failure;
the historical data feature extraction based on the frequent power failure mark, the historical user data and the historical environmental condition data comprises the following steps:
and performing assignment representation on the frequent power failure mark, the historical user data and the historical environment condition data of each power failure user, and extracting corresponding assignments as historical data characteristics.
5. The early warning method of claim 1, wherein the constructing a historical feature matrix based on the extracted historical data features comprises:
and constructing a one-dimensional historical feature matrix of each power failure user based on the extracted historical data features.
6. The warning method of claim 1, wherein the predictive model is a deep neural network model;
the inputting the historical characteristic matrix into the prediction model for training and learning until convergence to obtain the converged prediction model comprises the following steps:
regularizing the parameter vector of each layer of the network in the prediction model to obtain a regularization item;
updating the loss function in the prediction model by using the regularization term to obtain an updated prediction model;
dividing the historical feature matrix into a training set and a testing set, inputting the training set into the updated prediction model for feature learning training, and obtaining a trained prediction model;
inputting the test set into the trained prediction model for testing to obtain a test result;
judging whether the test result reaches a preset target or not, if so, obtaining a convergent prediction model;
if not, resetting the coefficients of all layer parameter vectors of the trained prediction model by adopting a back propagation algorithm, and after resetting, continuing training by using the training set data until convergence is reached or the training frequency reaches a threshold value.
7. The warning method of claim 6, wherein before the regularizing the parameter vectors of each layer of the network in the prediction model, the method comprises:
compressing the output nodes of each layer network in the prediction model in proportion;
the ratio is two to one.
8. The warning method of claim 6 wherein the training set and the test set are in a ratio of 9: 1.
9. The warning method according to claim 6, wherein the formula for resetting the coefficients of all layer parameter vectors of the trained predictive model by using the back propagation algorithm is as follows:
wherein the content of the first and second substances,
for updated coefficients of the i-th and j-th parameter vectors in the trained prediction model, WijThe coefficients of the ith parameter vector and the jth parameter vector in the trained prediction model are represented, lr is the learning rate, weight _ decay is the weight attenuation coefficient, momentum is the impulse, and L is the updated loss function.
10. An early warning device for frequent power failure, the device comprising:
a marking module: the system comprises a power failure user for obtaining frequent power failure, determining the reason of the power failure caused by the power failure user, and marking the frequent power failure for the power failure user based on the reason of the power failure;
a historical data acquisition module: the power failure monitoring system is used for obtaining historical user data and historical environment condition data of the power failure user during historical power failure;
a feature matrix construction module: the historical data feature extraction module is used for extracting historical data features based on the frequent power failure mark, the historical user data and the historical environment condition data, and constructing a historical feature matrix based on the extracted historical data features;
training a learning module: the prediction model is used for inputting the historical characteristic matrix into the prediction model for training and learning until convergence, and the convergence prediction model is obtained;
a real-time matrix construction module: the real-time characteristic matrix is used for acquiring the current environment data of the current user data set of the power failure user in real time and constructing a real-time characteristic matrix based on the frequent power failure mark and the current environment data of the current user data set;
the prediction early warning module: and the real-time feature matrix is input into the converged prediction model for prediction to obtain a prediction result, and the power failure early warning reminding is carried out on the frequent power failure user based on the prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010664816.XA CN112036682A (en) | 2020-07-10 | 2020-07-10 | Early warning method and device for frequent power failure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010664816.XA CN112036682A (en) | 2020-07-10 | 2020-07-10 | Early warning method and device for frequent power failure |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112036682A true CN112036682A (en) | 2020-12-04 |
Family
ID=73579037
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010664816.XA Pending CN112036682A (en) | 2020-07-10 | 2020-07-10 | Early warning method and device for frequent power failure |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112036682A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113112064A (en) * | 2021-04-09 | 2021-07-13 | 浙江和达科技股份有限公司 | Method and system for early warning water supply pipe network penetration, electronic equipment and storage medium |
| CN115276190A (en) * | 2022-09-22 | 2022-11-01 | 荣耀终端有限公司 | Charging reminding method, electronic device and storage medium |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102722759A (en) * | 2012-05-17 | 2012-10-10 | 河海大学 | Method for predicting power supply reliability of power grid based on BP neural network |
| US20160142894A1 (en) * | 2014-11-13 | 2016-05-19 | Mobiltron, Inc. | Systems and methods for real time detection and reporting of personal emergencies |
| CN106549375A (en) * | 2016-12-06 | 2017-03-29 | 国网浙江慈溪市供电公司 | Power distribution network frequently power failure management-control method and device |
| CN108734381A (en) * | 2018-04-11 | 2018-11-02 | 国网山东省电力公司 | A kind of outage information collaborative management method, apparatus and system |
| CN108805259A (en) * | 2018-05-23 | 2018-11-13 | 北京达佳互联信息技术有限公司 | neural network model training method, device, storage medium and terminal device |
| CN108830406A (en) * | 2018-05-29 | 2018-11-16 | 贵州黔驰信息股份有限公司 | A kind of main distribution based on data mining has a power failure configuration method in advance |
| CN109272125A (en) * | 2018-10-30 | 2019-01-25 | 深圳供电局有限公司 | A kind of frequent power failure discriminance analysis processing method and system |
| CN109492814A (en) * | 2018-11-15 | 2019-03-19 | 中国科学院深圳先进技术研究院 | A kind of Forecast of Urban Traffic Flow prediction technique, system and electronic equipment |
| CN109858663A (en) * | 2018-11-19 | 2019-06-07 | 中国农业大学 | Distribution network failure power failure INFLUENCING FACTORS analysis based on BP neural network |
| CN110210686A (en) * | 2019-06-13 | 2019-09-06 | 郑州轻工业学院 | A kind of electricity charge risk model construction method of electric power big data |
| CN110490359A (en) * | 2019-07-04 | 2019-11-22 | 广州供电局有限公司 | Consider extreme meteorological dynamic power distribution network scope of power outage prediction technique and system |
| CN110929226A (en) * | 2019-11-26 | 2020-03-27 | 广州供电局有限公司 | Power distribution network power failure prediction method, device and system |
-
2020
- 2020-07-10 CN CN202010664816.XA patent/CN112036682A/en active Pending
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102722759A (en) * | 2012-05-17 | 2012-10-10 | 河海大学 | Method for predicting power supply reliability of power grid based on BP neural network |
| US20160142894A1 (en) * | 2014-11-13 | 2016-05-19 | Mobiltron, Inc. | Systems and methods for real time detection and reporting of personal emergencies |
| CN106549375A (en) * | 2016-12-06 | 2017-03-29 | 国网浙江慈溪市供电公司 | Power distribution network frequently power failure management-control method and device |
| CN108734381A (en) * | 2018-04-11 | 2018-11-02 | 国网山东省电力公司 | A kind of outage information collaborative management method, apparatus and system |
| CN108805259A (en) * | 2018-05-23 | 2018-11-13 | 北京达佳互联信息技术有限公司 | neural network model training method, device, storage medium and terminal device |
| CN108830406A (en) * | 2018-05-29 | 2018-11-16 | 贵州黔驰信息股份有限公司 | A kind of main distribution based on data mining has a power failure configuration method in advance |
| CN109272125A (en) * | 2018-10-30 | 2019-01-25 | 深圳供电局有限公司 | A kind of frequent power failure discriminance analysis processing method and system |
| CN109492814A (en) * | 2018-11-15 | 2019-03-19 | 中国科学院深圳先进技术研究院 | A kind of Forecast of Urban Traffic Flow prediction technique, system and electronic equipment |
| CN109858663A (en) * | 2018-11-19 | 2019-06-07 | 中国农业大学 | Distribution network failure power failure INFLUENCING FACTORS analysis based on BP neural network |
| CN110210686A (en) * | 2019-06-13 | 2019-09-06 | 郑州轻工业学院 | A kind of electricity charge risk model construction method of electric power big data |
| CN110490359A (en) * | 2019-07-04 | 2019-11-22 | 广州供电局有限公司 | Consider extreme meteorological dynamic power distribution network scope of power outage prediction technique and system |
| CN110929226A (en) * | 2019-11-26 | 2020-03-27 | 广州供电局有限公司 | Power distribution network power failure prediction method, device and system |
Non-Patent Citations (5)
| Title |
|---|
| ABDERRAZAK KHEDIRI 等: "Deep-Belief Network Based Prediction Model for Power Outage in Smart Grid", 《ICCES"18: PROCEEDINGS OF THE 4TH ACM INTERNATIONAL CONFERENCE OF COMPUTING FOR ENGINEERING AND SCIENCES》 * |
| LIPING FAN 等: "Research and Application of Smart Grid Early Warning Decision Platform Based on Big Data Analysis", 《2019 4TH INTERNATIONAL CONFERENCE ON INTELLIGENT GREEN BUILDING AND SMART GRID (IGBSG)》 * |
| 李虹 等: "基于大数据的频繁停电预警督办机制研究", 《科学与财富》 * |
| 许鑫 等: "一种基于数据挖掘的频繁停电投诉预警模型", 《信息记录材料》 * |
| 陶鸿飞 等: "配电网频繁停电控制策略研究", 《农村电工》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113112064A (en) * | 2021-04-09 | 2021-07-13 | 浙江和达科技股份有限公司 | Method and system for early warning water supply pipe network penetration, electronic equipment and storage medium |
| CN113112064B (en) * | 2021-04-09 | 2023-10-31 | 浙江和达科技股份有限公司 | Water supply network permeation early warning method, system, electronic equipment and storage medium |
| CN115276190A (en) * | 2022-09-22 | 2022-11-01 | 荣耀终端有限公司 | Charging reminding method, electronic device and storage medium |
| CN115276190B (en) * | 2022-09-22 | 2023-02-21 | 荣耀终端有限公司 | Charging reminding method, electronic device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11967823B2 (en) | Method for monitoring short-term voltage stability of power system | |
| CN110807550B (en) | Distribution transformer overload recognition and early warning method based on neural network and terminal equipment | |
| CN112036682A (en) | Early warning method and device for frequent power failure | |
| CN110794308A (en) | Method and device for predicting train battery capacity | |
| CN110309979A (en) | Methods of electric load forecasting, device and equipment based on echo state network | |
| CN110210670A (en) | A kind of prediction technique based on power-system short-term load | |
| CN107527121A (en) | A kind of method of the information system running status diagnosis prediction of power network | |
| CN116090821A (en) | Power distribution network line security risk assessment method considering extreme weather | |
| CN114493238A (en) | Power supply service risk prediction method, system, storage medium and computer equipment | |
| CN108596450B (en) | Power grid risk early warning method and system | |
| CN117674119A (en) | Power grid operation risk assessment method, device, computer equipment and storage medium | |
| JP2003090887A (en) | Predication system and prediction method of instantaneous voltage drop by thunderbolt | |
| CN116502771B (en) | Power distribution method and system based on electric power material prediction | |
| Sun et al. | A multi-model-integration-based prediction methodology for the spatiotemporal distribution of vulnerabilities in integrated energy systems under the multi-type, imbalanced, and dependent input data scenarios | |
| CN112836843A (en) | Method and device for predicting base station out-of-service alarm | |
| Kosterev et al. | Preventive risk-management of power system for its reliability increasing | |
| CN113393102B (en) | Distribution transformer operation state trend prediction method based on data driving | |
| CN115204615A (en) | Disaster grading method and device based on Euclidean distance and grey correlation degree | |
| CN109063922A (en) | A kind of distribution transformer heavy-overload prediction technique based on cell occupancy rate | |
| CN114925900A (en) | Distribution transformer weight and overload risk early warning analysis method | |
| CN115034586A (en) | Method and device for predicting power failure risk of overhead line in strong convection weather | |
| CN114118759A (en) | Distribution transformer area load overload state assessment method and device | |
| Faghihi et al. | An efficient probabilistic approach to dynamic resilience assessment of power systems | |
| Huang et al. | Data-driven fault risk warning method for distribution system | |
| CN117575291B (en) | Federal learning data collaborative management method based on edge parameter entropy |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201204 |