CN117791597A - Power distribution network fault self-healing method and system based on machine learning - Google Patents
Power distribution network fault self-healing method and system based on machine learning Download PDFInfo
- Publication number
- CN117791597A CN117791597A CN202410199486.XA CN202410199486A CN117791597A CN 117791597 A CN117791597 A CN 117791597A CN 202410199486 A CN202410199486 A CN 202410199486A CN 117791597 A CN117791597 A CN 117791597A
- Authority
- CN
- China
- Prior art keywords
- distribution network
- power distribution
- data
- healing
- power
- 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
- 238000009826 distribution Methods 0.000 title claims abstract description 171
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000010801 machine learning Methods 0.000 title claims abstract description 33
- 238000012549 training Methods 0.000 claims abstract description 46
- 238000012163 sequencing technique Methods 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 69
- 238000011084 recovery Methods 0.000 claims description 32
- 238000007781 pre-processing Methods 0.000 claims description 23
- 238000012546 transfer Methods 0.000 claims description 15
- 210000002569 neuron Anatomy 0.000 claims description 11
- 238000013480 data collection Methods 0.000 claims description 9
- 230000005611 electricity Effects 0.000 claims description 7
- 230000009471 action Effects 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 6
- 238000004891 communication Methods 0.000 claims description 5
- 230000007613 environmental effect Effects 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 3
- 230000010355 oscillation Effects 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 abstract description 7
- 230000008569 process Effects 0.000 abstract description 7
- 238000012545 processing Methods 0.000 abstract description 4
- 238000004364 calculation method Methods 0.000 abstract description 3
- 238000003860 storage Methods 0.000 description 7
- 238000003745 diagnosis Methods 0.000 description 4
- 238000002955 isolation Methods 0.000 description 4
- 238000007726 management method Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000004140 cleaning Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 235000008113 selfheal Nutrition 0.000 description 2
- 241001529739 Prunella <angiosperm> Species 0.000 description 1
- 238000012952 Resampling Methods 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000012550 audit Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 239000003990 capacitor Substances 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 239000002243 precursor Substances 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 238000011144 upstream manufacturing Methods 0.000 description 1
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention provides a power distribution network fault self-healing method and system based on machine learning, and relates to the technical field of power distribution network fault self-healing. The power distribution network fault self-healing model is trained to output a power restoration scheme, and the switch is automatically controlled according to the power restoration scheme, so that manual analysis and intervention in the power distribution network fault self-healing process are greatly reduced, the power distribution network fault self-healing efficiency and reliability are improved, and the machine learning and the power distribution network fault self-healing are combined when a power distribution network self-healing system is designed and operated, so that the automation level of the system is improved. When training data of the power distribution network fault self-healing model is obtained, characteristic selection and importance sequencing are carried out on historical operation data, and the parameters of the N arranged parameters are obtained, so that the calculation scale of the model is greatly reduced, and the efficiency and accuracy of fault processing are improved.
Description
Technical Field
The invention relates to the technical field of power distribution network fault self-healing, in particular to a power distribution network fault self-healing method and system based on machine learning.
Background
Self-healing of a power distribution network refers to the capability of self-prevention and self-recovery of the power distribution network, and the capability is derived from detection of important parameters of the power distribution network and an effective control strategy. When the power distribution network breaks down, the main station system and the power distribution automatic switch are used for realizing the rapid isolation of a fault line and the rapid power recovery of a non-fault line, so that the power failure range and the power recovery time are reduced, and the power supply reliability is improved.
The existing distribution network self-healing implementation mainly depends on Feeder Terminal Units (FTUs) and distribution network main station systems. When the feeder line fails, the main station system performs fault diagnosis and positioning according to the tripping signal of the feeder line outlet switch, the protection action signal and the fault signal sent by the distribution terminal, automatically executes a fault isolation scheme, a fault upstream recovery scheme and a fault downstream optimal recovery scheme, and automatically completes fault region positioning, isolation and non-fault region load transfer. In a self-healing system of a power distribution network, although most links are automated, some links still need human participation, for example:
fault diagnosis and analysis: although the power distribution network self-healing system is capable of automatically detecting power grid faults, in some cases, fault diagnosis and analysis may require human intervention, for example, when the system cannot accurately identify fault types or fault positions, deep analysis and judgment of fault data may need to be performed manually;
system maintenance and upgrades: to ensure proper operation of the self-healing system of the power distribution network, system maintenance and upgrade work including checking the state of system hardware and software, updating system patches, optimizing system performance, etc. are required to be performed regularly, and may be required to be completed by a professional maintenance engineer;
although human participation is necessary in certain links, excessive intervention may reduce the efficiency and reliability of the self-healing system of the power distribution network. The development of machine learning is now becoming more mature, and research is necessary to combine machine learning with self-healing of power distribution network faults thanks to the advantages of machine learning in terms of handling large amounts of data, predictive and identifying capabilities, self-adaption and learning capabilities, and reduced human intervention.
Disclosure of Invention
The invention provides a power distribution network fault self-healing method and a power distribution network fault self-healing system based on machine learning, which are used for combining the machine learning with the power distribution network fault self-healing when designing and running the power distribution network self-healing system, so that the automation level of the system is improved.
A power distribution network fault self-healing method based on machine learning comprises the following steps:
step S1, collecting power distribution network data: collecting historical operation data of a power distribution network, wherein the historical operation data comprise power grid state information of normal operation of the power distribution network, power grid state information when faults occur, and a corresponding re-electricity strategy;
step S2, data preprocessing: preprocessing the historical operation data;
step S3, data selection: comparing the power grid state information when the fault occurs with the power grid state information of normal operation to obtain parameters related to the fault in the preprocessed data;
step S4, training set acquisition: the importance ranking is carried out on the parameters related to the faults, the parameters of the N arranged in front are obtained, and a training set is formed by the data corresponding to the parameters of the N arranged in front and the corresponding complex electricity strategies;
step S5, model construction and training: constructing a power distribution network fault self-healing model, and training the power distribution network fault self-healing model by adopting the training set to obtain a trained power distribution network fault self-healing model; the objective function of the power distribution network fault self-healing model comprises the following steps: recovering the function with the maximum total power loss load, the minimum system active loss function, the minimum switching operation frequency function and the minimum load transfer function;
step S6, fault identification and self-healing: inputting the operation data of the power distribution network to be tested into the power distribution network fault self-healing model, outputting a power restoration scheme, and automatically controlling a switch by a system according to the power restoration scheme;
step S7, updating historical operation data: and storing the related data of the fault self-healing of the power distribution network to historical operation data.
Further, in the step S1, the grid state information includes: voltage and current data, power and frequency data, grid topology information, protection device action information, device status information, communication data, weather and environmental data.
Further, in the step S2, the data preprocessing includes: data cleansing, data normalization, data balancing, noise and outlier handling.
Further, in the step S4, the importance ranking is performed on the parameters related to the fault, so as to obtain the parameters ranked in the top N, which specifically are:
step S41, acquiring parameters affected when the fault occurs according to the power grid state information when the fault occurs;
and S42, counting the occurrence probability of all affected parameters, and sequencing the affected parameters according to the probability from large to small to obtain the parameters ranked in the previous N.
Further, the power distribution network fault self-healing model is composed of an RNN model, the input layer is 1 layer, the number of neurons is 5, the hidden layer is LSTM, the number of layers is 2, the number of neurons in each layer is 128, the output layer is 1 layer, the number of neurons is 3, and the activation function is a softmax function.
Further, the maximum function formula of the total load of recovery power failure is as follows:
(1)
wherein F (L) is a total load function for recovering power loss, L k For the load of the k-th bus connected during power failure, y k Representing the distribution state change coefficient, y k =1 represents that the power loss has recovered, y k =0 represents power lossWithout restoration, J represents the total number of bus bars.
Further, the minimum function formula of the system active loss is as follows:
(2)
wherein,is the active power consumption after the power supply of the distribution network is recovered, I r For the working current oscillation amplitude of the R branch, R r And h is the total branch quantity of the power distribution network system.
Further, the minimum function formula of the switching operation times is as follows:
(3)
wherein F (C, D) is a function of the number of switching operations, m represents an undrawn disconnector in the distribution network, n represents an undrawn disconnector in the distribution network, C i 1 represents that the isolating switch i maintains a closed state in recovery, C i If 0, the isolating switch i is turned on in fault recovery, D j 1 indicates that the tie control switch j is turned on to off in recovery, D j A value of 0 indicates that the tie control switch j remains on during recovery.
Further, the load transfer minimum function formula is:
(4)
wherein Qs is a load transfer function, T is all connection busbar sets downstream of the opened sectional control switch, Q q The load transferred for the q-th busbar.
The power distribution network fault self-healing system based on machine learning, which uses the power distribution network fault self-healing method based on machine learning as described in any one of the above, comprises the following parts:
the power distribution network data collection module: the historical operation data of the power distribution network are collected, and the historical operation data comprise power grid state information of normal operation of the power distribution network, power grid state information when faults occur and corresponding re-electricity strategies;
and a data preprocessing module: the power distribution network data collection module is connected with the power distribution network data collection module and used for preprocessing the historical operation data;
and a data selection module: the data preprocessing module is connected with the data preprocessing module and is used for acquiring parameters related to faults in preprocessed data by comparing the power grid state information when the faults occur with the power grid state information of normal operation;
training set acquisition module: the data selection module is connected with the data selection module and is used for carrying out importance ranking on parameters related to faults, acquiring the parameters of the N arranged in front, and forming a training set by the data corresponding to the parameters of the N arranged in front and the corresponding complex electricity strategy;
model construction and training module: the power distribution network fault self-healing module is connected with the training set acquisition module and used for constructing a power distribution network fault self-healing model, and training the power distribution network fault self-healing model by adopting the training set to acquire a trained power distribution network fault self-healing model; the objective function of the power distribution network fault self-healing model comprises the following steps: recovering the function with the maximum total power loss load, the minimum system active loss function, the minimum switching operation frequency function and the minimum load transfer function;
and a fault identification and self-healing module: the power distribution network fault self-healing module is connected with the model building and training module and is used for inputting power distribution network data to be tested into the power distribution network fault self-healing model, outputting a power restoration scheme and automatically controlling a switch according to the power restoration scheme by the system;
historical operating data updating module: and the fault identification and self-healing module is connected with the fault identification and self-healing module and is used for storing relevant data of the fault self-healing of the power distribution network to historical operation data.
Compared with the prior art, the invention has the beneficial effects that:
firstly, the invention outputs a re-electricity scheme through the trained power distribution network fault self-healing model, and automatically controls the switch according to the re-electricity scheme, thereby greatly reducing manual analysis and intervention in the power distribution network fault self-healing process and improving the power distribution network fault self-healing efficiency and reliability;
secondly, when training data of the power distribution network fault self-healing model is acquired, firstly, characteristic selection is carried out on historical operation data, parameters irrelevant to faults are removed, importance ranking is carried out on parameters relevant to the faults, and the parameters ranked in the front N are acquired, so that the calculation scale of the model is greatly reduced, and the efficiency and accuracy of fault processing are improved;
thirdly, the invention determines an objective function with larger influence on the power distribution network fault self-healing model, and the objective function comprises the following steps: the method comprises the steps of recovering the function with the maximum total power loss, the function with the minimum active loss, the function with the minimum switching operation times and the function with the minimum load transfer, solving, outputting a complex power scheme for controlling the switch, and providing a clear optimization direction for a self-healing model, so that the model can quickly and accurately make a decision when a fault occurs, and the quick recovery and power supply guarantee of a power grid are realized.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow diagram of a power distribution network fault self-healing method based on machine learning;
fig. 2 is a block diagram of a power distribution network fault self-healing system based on machine learning.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below. It will be apparent that the described embodiments are only some, but not all, embodiments of the invention. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the invention, are within the scope of the invention.
The following describes specific embodiments of the present invention with reference to the drawings.
According to the power distribution network fault self-healing system and method based on machine learning, a re-electricity scheme is output through a trained power distribution network fault self-healing model, and a switch is automatically controlled according to the re-electricity scheme, so that manual analysis and intervention in a power distribution network fault self-healing process are greatly reduced, the power distribution network fault self-healing efficiency and reliability are improved, and the automatic level of the system is improved by combining machine learning with power distribution network fault self-healing when the power distribution network self-healing system is designed and operated.
Example 1: as shown in fig. 1, the invention provides a power distribution network fault self-healing method based on machine learning, which comprises the following steps:
step S1, collecting power distribution network data: and collecting historical operation data of the power distribution network, wherein the historical operation data comprise power grid state information of normal operation of the power distribution network, power grid state information when faults occur, and a corresponding re-electricity strategy.
The grid state information includes: voltage and current data, power and frequency data, grid topology information, protection device action information, device status information, communication data, weather and environmental data.
Voltage and current data: the voltage and current values of all nodes are monitored in real time, and whether the power grid has abnormal conditions such as overvoltage, undervoltage and overcurrent can be judged. These anomalies are often precursors or direct manifestations of the failure.
Power and frequency data: monitoring the power flow and frequency changes of the power grid helps to identify if the power grid is out of balance or frequency offset occurs. This is useful for diagnosing problems such as generator failure, load imbalance, etc.
Grid topology information: knowledge of the topology of the grid, including the switching state, line connection relationships, etc., is critical to fault location and isolation. When a fault occurs, the fault area can be rapidly positioned by comparing the topological structures in the normal state and the fault state.
Protection device action information: protection devices (such as circuit breakers, relays and the like) in the power grid automatically act when faults occur, and fault currents are cut off. The action information of the protection devices is collected, so that the type and the position of the fault can be known.
Device status information: the running state of equipment in the power grid, such as parameters of temperature, load rate and the like of a transformer, a switch, a capacitor and the like, is monitored in real time, and the fault or abnormality of the equipment can be found in time.
Communication data: the operation state of the power grid can be monitored in real time through the data collected by the communication system, wherein the operation state comprises switch positions, telemetering data and the like. These data are very important for fault diagnosis and self-healing.
Weather and environmental data: meteorological and environmental factors (such as wind speed, temperature, humidity, etc.) may have an impact on grid operation, resulting in faults. Collecting this data helps analyze the cause of the fault and predict possible future faults.
Step S2, data preprocessing: and preprocessing the historical operation data.
The data preprocessing comprises the following steps: data cleansing, data normalization, data balancing, noise and outlier handling.
Data cleaning: there may be errors, anomalies, or duplicates in the historical fault data that, if not cleaned, may interfere with subsequent data analysis and model training. The data cleaning comprises the steps of removing repeated data, correcting error data, processing missing values and the like, and the accuracy and consistency of the data are ensured.
Data normalization: since distribution network fault data may come from different data sources or collection devices, there may be differences in the dimension and range of the data. The data can be converted into the same dimension and range by data standardization, the dimension influence among different features is eliminated, and the comparability of the data and the performance of the model are improved.
Feature extraction: the raw power distribution network fault data may contain a large number of features, but not all features have an important role in fault classification and prediction. The feature extraction aims to extract features which have important influences on fault classification and prediction from the original data, reduce the dimension of the data and improve the efficiency and performance of the model.
Data balancing: in practical applications, distribution of distribution network fault data may be unbalanced, i.e., the number of samples for certain fault types may be far greater than for other types. Data imbalance may cause the model to bias towards most classes of samples during training, affecting the generalization performance of the model. Through data preprocessing, resampling, oversampling or undersampling methods can be adopted to balance data distribution, and the performance of the model is improved.
Noise and outlier handling: noise and outliers may be present in the distribution network fault data, which may be due to errors in the data acquisition equipment, disturbances in the transmission process, etc. Noise and outliers can negatively impact model training and therefore require preprocessing to identify and process these values, improving the quality of the data and stability of the model.
The collected historical fault data of the power distribution network is preprocessed to ensure the accuracy, consistency, comparability and reliability of the data, and a high-quality data base is provided for subsequent fault classification, prediction and self-healing.
Step S3, data selection: and comparing the power grid state information when the fault occurs with the power grid state information of normal operation to acquire parameters related to the fault in the preprocessed data.
And comparing the power grid state information of the normal operation of the power distribution network with the power grid state information when faults occur according to the preprocessed historical operation data, and if the variation amplitude of certain parameters exceeds a set threshold value, considering that the variation of the parameters is related to the faults, and screening out the parameters as parameters related to the faults.
Through this step, all fault-related parameters in the historical operating data can be obtained.
Step S4, training set acquisition: and carrying out importance ranking on the parameters related to the faults, obtaining the parameters ranked in the front N, and forming a training set by the data corresponding to the parameters ranked in the front N and the corresponding complex electricity strategies.
The importance ranking is carried out on the parameters related to the faults, and the parameters ranked in the front N are obtained, specifically:
step S41, acquiring parameters affected when the fault occurs according to the power grid state information when the fault occurs;
and S42, counting the occurrence probability of all affected parameters, and sequencing the affected parameters according to the probability from large to small to obtain the parameters ranked in the previous N.
Specifically, parameters related to faults in the first historical data are obtained, wherein the parameters comprise E1, E2, E3 and E4; acquiring parameters related to faults in the second historical data, wherein the parameters comprise E1, E3, E5 and E6; and the like, until the parameters related to the faults in the last historical data are obtained, wherein the parameters comprise E1, E6, E7 and E8; counting all fault parameters, wherein the occurrence number of E1 is n1, the occurrence number of E2 is n2, … …, the occurrence number of E8 is n8, and the occurrence probability of E1 is P1=n1/(n1+n2+ … … +n8), … … and P8=n1/(n1+n2+ … … +n8), and then sorting P1, P2, … … and P8 in the order from large to small: and P1> P3> P5> P6> P8> P4> P2> P7, acquiring parameters corresponding to the probabilities of the top N rows, wherein if N=5, the acquired parameters are E1, E3, E5, E6 and E8. The first piece of data of the training set is: e1, E3, the duplicate strategy of this fault, the second piece of data of the training set is: e1, E3, E5, E6, the power recovery strategy for this failure, … …, and so on, the last piece of data for the training set is: e1, E6, E8, the return to electricity strategy for this failure.
Only a small number of faults are pointed out, in practical application, the number of parameters affected by the faults can be greatly increased, and balance between parameter types and training efficiency is achieved by reasonably setting the value of N.
When the training data of the power distribution network fault self-healing model is obtained, the characteristic selection is firstly carried out on the historical operation data, parameters irrelevant to faults are removed, importance ranking is carried out on the parameters relevant to the faults, and the parameters ranked in the front N are obtained, so that the calculation scale of the model is greatly reduced, and the efficiency and accuracy of fault processing are improved.
Step S5, model construction and training: constructing a power distribution network fault self-healing model, and training the power distribution network fault self-healing model by adopting the training set to obtain a trained power distribution network fault self-healing model; the objective function of the power distribution network fault self-healing model comprises the following steps: the maximum function of the total power loss load, the minimum function of the active loss of the system, the minimum function of the switching operation times and the minimum function of load transfer are recovered.
The power distribution network fault self-healing model is composed of an RNN model, an input layer is responsible for receiving power grid state data, the number of layers is 1 layer, the number of neurons is 5, and each neuron corresponds to one input characteristic, such as 5 key parameters of voltage, current, active power, reactive power and frequency;
the hidden layer is responsible for capturing time dependence and modes in sequence data, LSTM (long-short-term memory) is selected as a circulating neural network layer, the number of layers is 2, and the number of neurons in each layer is 128;
the output layer is responsible for predicting recovery strategies after power grid faults, the number of layers is 1, the number of neurons is 3, each neuron corresponds to one possible recovery strategy, the predicted future power grid state, fault type and recovery strategy are output, and the activation function is a softmax function.
In practice, the fault self-healing of the power distribution network is realized by constructing a network again after the fault, the process does not meet the linear relation, and the influence factors such as load recovery amount, control switch operation management number, network constraint, user priority distribution level and the like need to be comprehensively considered, so that under the condition of meeting different fault recovery requirements, more combination possibilities need to be provided to achieve the optimal selection of fault recovery, and an objective function needs to be determined first.
When the power distribution network faults self-heal, the power supply restoration decision is required to restore the power supply of the power consumer in the power loss area to the greatest extent. The maximum function formula for recovering the total power loss load is as follows:
(1)
wherein F (L) is a total load function for recovering power loss, L k For the load of the k-th bus connected during power failure, y k Representing the distribution state change coefficient, y k =1 represents that the power loss has recovered, y k =0 represents that no power loss is recovered, J represents the total number of bus bars.
The minimum function formula of the system active loss is as follows:
(2)
wherein,is the active power consumption after the power supply of the distribution network is recovered, I r For the working current oscillation amplitude of the R branch, R r And h is the total branch quantity of the power distribution network system.
When the power distribution network faults self-heals, the economic cost of the control management of the sectional control switches required by the recovery strategy is smaller, namely the fewer the number of the control management sectional control switches is, the less the time spent for the control management is, and the faster the recovery of the supplied power supply in the power failure area can be supported. The minimum function formula of the switching operation times is as follows:
(3)
wherein F (C, D) is a function of the number of switching operations, m represents an undrawn disconnector in the distribution network, n represents an undrawn disconnector in the distribution network, C i 1 represents that the isolating switch i maintains a closed state in recovery, C i If 0, the isolating switch i is turned on in fault recovery, D j 1 indicates that the tie control switch j is turned on to off in recovery, D j A value of 0 indicates that the tie control switch j remains on during recovery.
When the fault self-healing of the power distribution network occurs, when the replacement use volume of the working power supply close to the power-off area cannot recover all power-off loads, some loads close to the working power supply in the power-off section need to be converted to other more distant power supply areas so as to increase the replacement use volume of the working power supply close to the power-off area to recover a great amount of power-off loads. In the recovery full process, it is desirable that the effect on other power supply areas be smaller and better. The load transfer minimum function formula is:
(4)
wherein Qs is a load transfer function, T is all connection busbar sets at the downstream of the opened sectional control switch, and Qq is the load transferred by the q-th busbar.
The invention determines an objective function with larger influence on a power distribution network fault self-healing model, and comprises the following steps: the method comprises the steps of recovering the function with the maximum total power loss, the function with the minimum active loss, the function with the minimum switching operation times and the function with the minimum load transfer, solving, outputting a complex power scheme for controlling the switch, and providing a clear optimization direction for a self-healing model, so that the model can quickly and accurately make a decision when a fault occurs, and the quick recovery and power supply guarantee of a power grid are realized.
Step S6, fault identification and self-healing: inputting to-be-detected power distribution network data into the power distribution network fault self-healing model, outputting a power restoration scheme, and automatically controlling a switch by a system according to the power restoration scheme;
according to the power distribution network fault self-healing method, the trained power distribution network fault self-healing model is used for outputting the power restoration scheme, and the switch is automatically controlled according to the power restoration scheme, so that manual analysis and intervention in the power distribution network fault self-healing process are greatly reduced, and the power distribution network fault self-healing efficiency and reliability are improved.
Step S7, updating historical operation data: and storing the related data of the fault self-healing of the power distribution network to historical operation data.
By supplementing the historical operation data, the reliability, economy and operation efficiency of the power distribution network can be improved in the aspects of data analysis and improvement, model optimization, decision support, fault prediction and prevention, compliance, audit and the like.
Example 2: as shown in fig. 2, the invention also provides a self-healing system for power distribution network faults based on machine learning, which uses the self-healing method for power distribution network faults based on machine learning as described in any one of embodiment 1, and comprises the following parts:
the power distribution network data collection module: the historical operation data of the power distribution network are collected, and the historical operation data comprise power grid state information of normal operation of the power distribution network, power grid state information when faults occur and corresponding re-electricity strategies;
and a data preprocessing module: the power distribution network data collection module is connected with the power distribution network data collection module and used for preprocessing the historical operation data;
and a data selection module: the data preprocessing module is connected with the data preprocessing module and is used for acquiring parameters related to faults in preprocessed data by comparing the power grid state information when the faults occur with the power grid state information of normal operation;
training set acquisition module: the data selection module is connected with the data selection module and is used for carrying out importance ranking on parameters related to faults, acquiring the parameters of the N arranged in front, and forming a training set by the data corresponding to the parameters of the N arranged in front and the corresponding complex electricity strategy;
model construction and training module: the power distribution network fault self-healing module is connected with the training set acquisition module and used for constructing a power distribution network fault self-healing model, and training the power distribution network fault self-healing model by adopting the training set to acquire a trained power distribution network fault self-healing model; the objective function of the power distribution network fault self-healing model comprises the following steps: recovering the function with the maximum total power loss load, the minimum system active loss function, the minimum switching operation frequency function and the minimum load transfer function;
and a fault identification and self-healing module: the power distribution network fault self-healing module is connected with the model building and training module and is used for inputting power distribution network data to be tested into the power distribution network fault self-healing model, outputting a power restoration scheme and automatically controlling a switch according to the power restoration scheme by the system;
historical operating data updating module: and the fault identification and self-healing module is connected with the fault identification and self-healing module and is used for storing relevant data of the fault self-healing of the power distribution network to historical operation data.
Example 3: an electronic device, the electronic device comprising:
a processor and a memory;
the processor is configured to execute the steps of a power distribution network fault self-healing method based on machine learning according to any one of embodiment 1 by calling a program or instructions stored in the memory.
Example 4: a computer readable storage medium comprising computer program instructions for causing a computer to perform the steps of a machine learning based power distribution network fault self-healing method according to any one of embodiment 1.
The computer readable storage medium may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium would include the following: an electrical connection having one or more wires, a portable disk, a hard disk, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.
Claims (10)
1. The power distribution network fault self-healing method based on machine learning is characterized by comprising the following steps of:
step S1, collecting power distribution network data: collecting historical operation data of a power distribution network, wherein the historical operation data comprise power grid state information of normal operation of the power distribution network, power grid state information when faults occur, and a corresponding re-electricity strategy;
step S2, data preprocessing: preprocessing the historical operation data;
step S3, data selection: comparing the power grid state information when the fault occurs with the power grid state information of normal operation to obtain parameters related to the fault in the preprocessed data;
step S4, training set acquisition: the importance ranking is carried out on the parameters related to the faults, the parameters of the N arranged in front are obtained, and a training set is formed by the data corresponding to the parameters of the N arranged in front and the corresponding complex electricity strategies;
step S5, model construction and training: constructing a power distribution network fault self-healing model, and training the power distribution network fault self-healing model by adopting the training set to obtain a trained power distribution network fault self-healing model; the objective function of the power distribution network fault self-healing model comprises the following steps: recovering the function with the maximum total power loss load, the minimum system active loss function, the minimum switching operation frequency function and the minimum load transfer function;
step S6, fault identification and self-healing: inputting the operation data of the power distribution network to be tested into the power distribution network fault self-healing model, outputting a power restoration scheme, and automatically controlling a switch by a system according to the power restoration scheme;
step S7, updating historical operation data: and storing the related data of the fault self-healing of the power distribution network to historical operation data.
2. The machine learning based power distribution network fault self-healing method according to claim 1, wherein in the step S1, the power grid state information includes: voltage and current data, power and frequency data, grid topology information, protection device action information, device status information, communication data, weather and environmental data.
3. The machine learning based power distribution network fault self-healing method according to claim 1, wherein in the step S2, the data preprocessing includes: data cleansing, data normalization, data balancing, noise and outlier handling.
4. The machine learning based power distribution network fault self-healing method according to claim 1, wherein in step S4, the importance ranking is performed on the parameters related to the fault, and the parameters ranked in the top N are obtained, which specifically are:
step S41, acquiring parameters affected when the fault occurs according to the power grid state information when the fault occurs;
and S42, counting the occurrence probability of all affected parameters, and sequencing the affected parameters according to the probability from large to small to obtain the parameters ranked in the previous N.
5. The machine learning-based power distribution network fault self-healing method according to claim 1, wherein the power distribution network fault self-healing model is composed of an RNN model, the input layer is 1 layer, the number of neurons is 5, the hidden layer is LSTM, the number of layers is 2, the number of neurons in each layer is 128, the output layer is 1 layer, the number of neurons is 3, and the activation function is a softmax function.
6. The machine learning-based power distribution network fault self-healing method according to claim 5, wherein the maximum function formula of the total load of recovery power loss is:
(1)
wherein F (L) is a total load function for recovering power loss, L k For the load of the k-th bus connected during power failure, y k Representing the distribution state change coefficient, y k =1 represents that the power loss has recovered, y k =0 represents no recovery from power lossJ represents the total number of bus bars.
7. The machine learning-based power distribution network fault self-healing method according to claim 1, wherein the system active loss minimum function formula is:
(2)
wherein,is the active power consumption after the power supply of the distribution network is recovered, I r For the working current oscillation amplitude of the R branch, R r And h is the total branch quantity of the power distribution network system.
8. The machine learning based power distribution network fault self-healing method according to claim 1, wherein the minimum function formula of the number of switching operations is:
(3)
wherein F (C, D) is a function of the number of switching operations, m represents an undrawn disconnector in the distribution network, n represents an undrawn disconnector in the distribution network, C i 1 represents that the isolating switch i maintains a closed state in recovery, C i If 0, the isolating switch i is turned on in fault recovery, D j 1 indicates that the tie control switch j is turned on to off in recovery, D j A value of 0 indicates that the tie control switch j remains on during recovery.
9. The machine learning based power distribution network fault self-healing method according to claim 1, wherein the load transfer minimum function formula is:
(4)
wherein Qs is a load transfer function, T is all connection busbar sets downstream of the opened sectional control switch, Q q The load transferred for the q-th busbar.
10. Machine learning-based power distribution network fault self-healing system, using the machine learning-based power distribution network fault self-healing method according to any one of claims 1 to 9, characterized by comprising the following parts:
the power distribution network data collection module: the historical operation data of the power distribution network are collected, and the historical operation data comprise power grid state information of normal operation of the power distribution network, power grid state information when faults occur and corresponding re-electricity strategies;
and a data preprocessing module: the power distribution network data collection module is connected with the power distribution network data collection module and used for preprocessing the historical operation data;
and a data selection module: the data preprocessing module is connected with the data preprocessing module and is used for acquiring parameters related to faults in preprocessed data by comparing the power grid state information when the faults occur with the power grid state information of normal operation;
training set acquisition module: the data selection module is connected with the data selection module and is used for carrying out importance ranking on parameters related to faults, acquiring the parameters of the N arranged in front, and forming a training set by the data corresponding to the parameters of the N arranged in front and the corresponding complex electricity strategy;
model construction and training module: the power distribution network fault self-healing module is connected with the training set acquisition module and used for constructing a power distribution network fault self-healing model, and training the power distribution network fault self-healing model by adopting the training set to acquire a trained power distribution network fault self-healing model; the objective function of the power distribution network fault self-healing model comprises the following steps: recovering the function with the maximum total power loss load, the minimum system active loss function, the minimum switching operation frequency function and the minimum load transfer function;
and a fault identification and self-healing module: the power distribution network fault self-healing module is connected with the model building and training module and is used for inputting power distribution network data to be tested into the power distribution network fault self-healing model, outputting a power restoration scheme and automatically controlling a switch according to the power restoration scheme by the system;
historical operating data updating module: and the fault identification and self-healing module is connected with the fault identification and self-healing module and is used for storing relevant data of the fault self-healing of the power distribution network to historical operation data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410199486.XA CN117791597A (en) | 2024-02-23 | 2024-02-23 | Power distribution network fault self-healing method and system based on machine learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410199486.XA CN117791597A (en) | 2024-02-23 | 2024-02-23 | Power distribution network fault self-healing method and system based on machine learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117791597A true CN117791597A (en) | 2024-03-29 |
Family
ID=90387450
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410199486.XA Pending CN117791597A (en) | 2024-02-23 | 2024-02-23 | Power distribution network fault self-healing method and system based on machine learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117791597A (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120150361A1 (en) * | 2012-02-16 | 2012-06-14 | Lazaris Spyros J | Renewable energy-based electricity grid infrastructure and method of grid infrastructure automation and operation |
CN115495986A (en) * | 2022-09-24 | 2022-12-20 | 国网浙江省电力有限公司经济技术研究院 | Self-healing distributed network reconstruction method for power distribution network |
CN115800274A (en) * | 2023-02-08 | 2023-03-14 | 国网天津市电力公司电力科学研究院 | Automatic self-adaption method and device for feeder of 5G power distribution network and storage medium |
CN116975766A (en) * | 2023-08-01 | 2023-10-31 | 广东电网有限责任公司 | Power distribution network fault self-healing method and system based on machine learning |
CN116990633A (en) * | 2023-06-25 | 2023-11-03 | 浙江华电器材检测研究院有限公司 | Fault studying and judging method based on multiple characteristic quantities |
-
2024
- 2024-02-23 CN CN202410199486.XA patent/CN117791597A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120150361A1 (en) * | 2012-02-16 | 2012-06-14 | Lazaris Spyros J | Renewable energy-based electricity grid infrastructure and method of grid infrastructure automation and operation |
CN115495986A (en) * | 2022-09-24 | 2022-12-20 | 国网浙江省电力有限公司经济技术研究院 | Self-healing distributed network reconstruction method for power distribution network |
CN115800274A (en) * | 2023-02-08 | 2023-03-14 | 国网天津市电力公司电力科学研究院 | Automatic self-adaption method and device for feeder of 5G power distribution network and storage medium |
CN116990633A (en) * | 2023-06-25 | 2023-11-03 | 浙江华电器材检测研究院有限公司 | Fault studying and judging method based on multiple characteristic quantities |
CN116975766A (en) * | 2023-08-01 | 2023-10-31 | 广东电网有限责任公司 | Power distribution network fault self-healing method and system based on machine learning |
Non-Patent Citations (2)
Title |
---|
刘凯等: "《基于启发式规则和人工免疫的智能配电网自愈》", 《武汉大学学报(工学版)》, 1 April 2016 (2016-04-01), pages 218 - 222 * |
闫颖;何云瑞;陈亮;王宁;李扬;郑思远;: "基于CMDB的信息系统故障根因定位技术的研究", 通信电源技术, no. 03, 10 February 2020 (2020-02-10) * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Haque et al. | Application of neural networks in power systems; a review | |
RU2525859C2 (en) | Controlling power outage and faulty state of electrical power system | |
CN115271263B (en) | Power equipment defect early warning method, system and medium based on improved association rule | |
CN109949178B (en) | Method for judging and complementing power failure event of medium-voltage distribution network based on support vector machine | |
CN113328437B (en) | Intelligent power distribution network CPS topology construction method and fault recovery method | |
CN101609987B (en) | Joint coordination method for master-station automatic voltage control and plant-station voltage reactive power control | |
CN112485597A (en) | Power system transmission line fault diagnosis method and system based on multi-source data | |
CN117374978B (en) | Grid-connected scheduling management method and system constructed by combining knowledge graph | |
CN111049131A (en) | Method and system for generating regional power grid online fault disposal plan | |
CN113344364B (en) | Power failure plan risk analysis method and device, electronic equipment and storage medium | |
Sarathkumar et al. | Review on power restoration techniques for smart power distribution systems | |
CN113964816A (en) | Analysis method for distribution network feeder line fault self-healing rate | |
CN117791597A (en) | Power distribution network fault self-healing method and system based on machine learning | |
CN112467686B (en) | Power distribution network fault recovery method, device, control equipment and storage medium | |
Yang et al. | Data domain adaptation for voltage stability evaluation considering topology changes | |
CN111751655A (en) | Fault self-healing method and device for distribution line, computer equipment and storage medium | |
CN113890020B (en) | Hybrid optimization configuration method for FRTU and FI | |
CN117335578B (en) | Detection and regulation system for loop closing power conversion of low-voltage power distribution | |
McArthur et al. | Automated post-fault diagnosis of power system disturbances | |
Rajesh et al. | ENHANCING POWER SYSTEM STABILITY WITH AI-BASED RELAYING ALGORITHMS–A REVIEW. | |
CN115357462A (en) | Power dispatching abnormal data rapid detection method based on deep learning | |
Bai et al. | A micro grid fault diagnosis method based on redundant Petri net considering temporal constraints | |
Dong et al. | GIS-integrated analytical model for fault diagnosis of power distribution network | |
CN117744020A (en) | Power distribution network power outage range identification method based on multi-source heterogeneous data fusion | |
WO2024003367A1 (en) | Method and device for forecasting upcoming faults in power systems |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination |