CN117590158B - Abnormal state identification method, device and control system of power distribution network - Google Patents
Abnormal state identification method, device and control system of power distribution network Download PDFInfo
- Publication number
- CN117590158B CN117590158B CN202410075689.8A CN202410075689A CN117590158B CN 117590158 B CN117590158 B CN 117590158B CN 202410075689 A CN202410075689 A CN 202410075689A CN 117590158 B CN117590158 B CN 117590158B
- Authority
- CN
- China
- Prior art keywords
- current
- zero sequence
- distribution network
- power distribution
- low frequency
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 70
- 238000000034 method Methods 0.000 title claims abstract description 37
- 230000008859 change Effects 0.000 claims abstract description 44
- 238000012549 training Methods 0.000 claims description 40
- 230000035772 mutation Effects 0.000 claims description 18
- 230000007935 neutral effect Effects 0.000 claims description 17
- 230000001629 suppression Effects 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 5
- NCGICGYLBXGBGN-UHFFFAOYSA-N 3-morpholin-4-yl-1-oxa-3-azonia-2-azanidacyclopent-3-en-5-imine;hydrochloride Chemical compound Cl.[N-]1OC(=N)C=[N+]1N1CCOCC1 NCGICGYLBXGBGN-UHFFFAOYSA-N 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000010187 selection method Methods 0.000 description 2
- 239000000243 solution Substances 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- AYFVYJQAPQTCCC-GBXIJSLDSA-N L-threonine Chemical compound C[C@@H](O)[C@H](N)C(O)=O AYFVYJQAPQTCCC-GBXIJSLDSA-N 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000002347 injection Methods 0.000 description 1
- 239000007924 injection Substances 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
- G01R31/086—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
- Y04S10/52—Outage or fault management, e.g. fault detection or location
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a method, a device and a control system for identifying abnormal states of a power distribution network, belonging to the technical field of power grid fault detection, wherein the method comprises the following steps: by measuring the extremely low frequency zero sequence current signals in each line and the power frequency zero sequence voltage of each bus when the state of the power distribution network is changed, the extremely low frequency current is shunted according to the ground impedance of each line, and the extremely low frequency signals are selected as characteristic signals for identifying fault lines, so that the accuracy of identifying the fault lines in the power distribution network can be effectively improved; and finally, determining a current sample by using the power frequency zero sequence voltage of each bus and the target very low frequency current of each line at the abrupt change moment, inputting a trained CNN abnormal recognition model to obtain a classification result, and improving the accuracy of power distribution network fault detection.
Description
Technical Field
The invention belongs to the technical field of power grid fault detection, and particularly relates to an abnormal state identification method, device and control system of a power distribution network.
Background
The distribution network is used as an important component of a power system, has the advantages of more power equipment, wide coverage area and complex load types, and is easy to generate various faults, wherein single-phase earth faults of the line are the most common faults of the distribution network line. After the single-phase earth fault occurs to the line, the fault phase voltage is reduced, the normal phase voltage is increased, and the zero sequence voltage is increased, but the line voltage is balanced, so that the system can operate for a period of time when the system has a stable earth fault. In order to prevent serious personal safety accidents and property loss caused by further expansion of faults during the period, a method needs to be designed for rapidly identifying fault lines, namely fault line selection, and timely cutting off, so that the power supply reliability of the system is ensured.
The medium-low voltage distribution network generally adopts a neutral point non-effective grounding mode, and in a system with more overhead lines, the neutral point non-grounding mode is generally adopted, when a single-phase grounding fault occurs in the system adopting the mode, the power frequency zero sequence current amplitude of a fault line is large and is approximately equal to the sum of all other line zero sequence currents. Thus, faulty wires of such a system are relatively easy to identify. However, if the fault line zero sequence current is too large, the fault ground point is prone to arcing.
At present, a system with a neutral point grounded through an arc suppression coil does not have a very effective method for accurately identifying a fault line, and the arc suppression coil reduces fault characteristics while improving the safety of the system. It has been proposed to identify a line by using a signal of a higher harmonic, for example, a fifth harmonic method, a higher harmonic injection method, etc., but the capacitance to ground exhibits low impedance at a high frequency voltage, and the identification effect is not ideal for a high-resistance ground fault; there are also papers that propose to inject half-wave direct current signals into the system by arc suppression coil parallel diodes, and judge the fault line according to the superposition signal of half-wave direct current signals and zero sequence current, but when the alternating voltage is a certain half period, the system is equivalent to a mode that the neutral point is grounded through a small resistor, and the safety is possibly affected. On the one hand, the fault line diagnosis method has a certain limitation on accuracy, on the other hand, the current mainstream method is very dependent on zero sequence voltage threshold starting, namely, once the zero sequence voltage exceeds a set threshold value, fault judgment is automatically carried out, but according to the fact that the operation overvoltage under the unbalanced condition frequently occurs in a system, the zero sequence voltage has larger abrupt change in a short time, however, the situation is very likely to cause the starting of a more sensitive fault diagnosis algorithm, so that the instantaneous interference is identified as the error judgment caused by the fault, and therefore, the problem that the instantaneous interference is urgently needed to be solved is identified without the help of the zero sequence voltage starting algorithm. In general, the accuracy of the existing power distribution network fault and interference state identification method is to be improved.
Disclosure of Invention
Aiming at the defects or improvement demands of the prior art, the invention provides an abnormal state identification method, device and control system of a power distribution network, which aim to construct a current sample by utilizing the power frequency zero sequence voltage of each busbar and the target very low frequency current of each line at the moment of abrupt change; the current sample is input into a trained CNN abnormal recognition model so that the CNN abnormal recognition model outputs a classification result, abnormal state recognition of the power distribution network is realized, whether the power distribution network is line fault or instant interference is recognized, and therefore the technical problem of low accuracy of the existing fault recognition is solved.
To achieve the above object, according to one aspect of the present invention, there is provided a method for identifying an abnormal state of a power distribution network, including:
S1: when the current abrupt change value on any line exceeds a set threshold value, intercepting the extremely low frequency zero sequence current of all lines and the power frequency zero sequence voltage of each bus in a preset period before and after the abrupt change moment; the current abrupt change value is the difference value between the very low frequency zero sequence current at the current moment and the previous moment;
S2: calculating a current compensation coefficient according to the change amount of the power frequency zero sequence voltage of each bus in the preset period;
s3: compensating the very low frequency zero sequence current of each line under each frequency by using the current compensation coefficient to obtain the target very low frequency current of each line under each frequency band at the abrupt moment;
S4: constructing a training sample by utilizing the power frequency zero sequence voltage of each busbar and the target very low frequency current of each line at the abrupt change moment;
S5: training the CNN abnormal recognition model by using a plurality of training samples;
S6: when the current mutation value of any line exceeds the set threshold, a corresponding current sample is obtained according to S1-S4, and the current sample is input into a trained CNN abnormal recognition model to output a classification result, so that abnormal state recognition of the power distribution network is realized; the classification result comprises: and judging the abnormal state of the power distribution network as a line fault or instantaneous interference of the whole power distribution network.
In one embodiment, the S2 includes:
S21: determining a busbar power frequency zero sequence voltage mutation value corresponding to the mutation moment from the power frequency zero sequence voltage of each busbar in the preset period;
s22: and calculating the current compensation coefficient according to the relation between the bus power frequency zero sequence voltage abrupt change value and the maximum value of the bus power frequency zero sequence voltage in the preset period.
In one embodiment, the step S22 includes:
calculating the current compensation coefficient alpha by using the formula alpha=sin -1[U(T0)/Umax;
wherein U (T 0) is a busbar power frequency zero sequence voltage mutation value corresponding to the mutation time T 0; u max is the maximum value of the bus power frequency zero sequence voltage in the preset period.
In one embodiment, the step S3 includes:
Calculating a target very low frequency current I' n (omega) of the ith line at the frequency omega at the abrupt moment by using a formula I'i(ω)=Ii(ω)×{(ω0 2-ω2)/[(ω0cosα)2+(ωsinα)2]1/2};
Wherein, I i (omega) and I' i (omega) are respectively the very low frequency zero sequence current before and after the I line compensates in frequency omega, omega 0 is power frequency, and alpha is the current compensation coefficient.
In one embodiment, the S4 includes:
Combining the target extremely low frequency current of each line at the abrupt change moment and the power frequency zero sequence voltage sequence of each bus into a multidimensional matrix with a row vector as a time sequence and a column vector as a characteristic quantity;
mapping the data of each row in the multidimensional matrix to [0,1] for normalization to obtain a feature vector, converting the feature vector into an electrical feature pseudo-graph, and taking the electrical feature pseudo-graph as one training sample.
In one embodiment, the step S5 includes:
Taking each training sample as input, taking a classification result as output, and training the CNN abnormal recognition model; the CNN anomaly identification model comprises an input layer, a convolution layer, a batch normalization layer, a ReLU activation layer, a maximum pooling layer, a full connection layer and a classification layer.
In one embodiment, the method for identifying abnormal states of the power distribution network further includes:
When the classification result is a fault of a certain line in the power distribution network, dividing the line into a plurality of sections, calculating the difference value of the very low frequency zero sequence currents at the head end and the tail end of each section, and determining the section with the largest difference value as the fault section.
In one embodiment, the neutral point of the distribution network is not grounded, or the neutral point is grounded through an arc suppression coil, or the neutral point is grounded through a small resistor.
According to another aspect of the present invention, there is provided an abnormal state identification apparatus for a power distribution network, including:
The intercepting module is used for intercepting the extremely low-frequency zero sequence current of all lines and the power frequency zero sequence voltage of each bus in a preset period before and after the abrupt change moment when the abrupt change value of the current on any line exceeds a set threshold value; the current abrupt change value is the difference value between the very low frequency zero sequence current at the current moment and the previous moment;
the calculation module is used for calculating a current compensation coefficient according to the change amount of the power frequency zero sequence voltage of each bus in the preset period;
the compensation module is used for compensating the very low frequency zero sequence current of each line under each frequency by utilizing the current compensation coefficient to obtain the target very low frequency current of each line under each frequency band at the abrupt moment;
The construction module is used for constructing a training sample by utilizing the power frequency zero sequence voltage of each busbar at the abrupt change moment and the target very low frequency current of each line;
the training model is used for training the CNN abnormal recognition model by utilizing a plurality of training samples;
The identification module is used for acquiring a corresponding current sample when the current mutation value of any line exceeds the set threshold value, inputting the current sample into a trained CNN abnormal identification model to enable the current sample to output a classification result, and realizing abnormal state identification of the power distribution network; the classification result comprises: and judging the abnormal state of the power distribution network as a line fault or instantaneous interference of the whole power distribution network.
According to another aspect of the present invention, there is provided a power distribution network system comprising a memory storing a computer program and a processor implementing the steps of the method for identifying abnormal states of a power distribution network when the computer program is executed by the processor.
According to another aspect of the present invention, a computer readable storage medium is provided, on which a computer program is stored, which computer program, when being executed by a processor, implements the steps of a method for identifying an abnormal state of a power distribution network.
In general, the above technical solutions conceived by the present invention, compared with the prior art, enable the following beneficial effects to be obtained:
(1) According to the invention, by measuring the extremely low frequency zero sequence current signals in each line and the power frequency zero sequence voltage of each bus when the state of the power distribution network is changed, the extremely low frequency current is split according to the ground impedance of each line, and the extremely low frequency signals are selected as the characteristic signals for identifying the fault line, so that the problem that the fault characteristics are not outstanding due to resonance between the arc suppression coil and the line capacitance in the existing system can be effectively avoided, and the accuracy of identifying the fault line in the power distribution network is effectively improved; the power frequency zero sequence voltage variation amount of each busbar is further utilized to calculate a current compensation coefficient, the extremely low frequency zero sequence current of each line under each frequency is compensated, the influence of an initial fault angle on the amplitude attenuation of the extremely low frequency zero sequence current is reduced, finally, the power frequency zero sequence voltage of each busbar and the target extremely low frequency current of each line at the abrupt change moment are utilized to determine a current sample, a trained CNN abnormal recognition model is input to obtain an abnormal recognition result, a threshold value of the zero sequence voltage is not required to be set as a rigidity criterion, the CNN abnormal recognition model is combined, a flexible classification criterion is adopted, the interference of the zero sequence voltage disturbance to fault starting during normal operation of the system is avoided, the misjudgment of the power distribution network state is caused, and the accuracy of power distribution network fault detection is improved.
(2) According to the scheme, the current compensation coefficient is calculated according to the relation between the power frequency zero sequence voltage abrupt change value of the bus and the maximum value of the power frequency zero sequence voltage of the bus in the preset period, and the initial fault angle can be simply estimated only by means of the zero sequence voltage, so that the low-frequency zero sequence current of the positive electrode is corrected.
(3) The scheme calculates the current compensation coefficient alpha by utilizing alpha=sin -1[U(T0)/Umax ]; and calculating the target very low frequency current I' n (omega) of the ith line at the frequency omega at the abrupt change moment by using a formula I'i(ω)=Ii(ω)×{(ω0 2-ω2)/[(ω0cosα)2+(ωsinα)2]1/2}, so that the attenuation of the very low frequency currents with different frequencies caused by different initial fault angles can be compensated, and the influence of the initial fault angles on the attenuation of the very low frequency currents is reduced.
(4) The scheme combines the target very low frequency current of each line at the abrupt change moment and the power frequency zero sequence voltage sequence of each bus into a multidimensional matrix with a row vector as a time sequence and a column vector as a characteristic quantity; and mapping the data of each row in the multidimensional matrix to [0,1] for normalization to obtain a feature vector, and converting the feature vector into an electrical feature pseudo-graph, so that electrical information can be converted into an image recognition problem of CNN good treatment, and the electrical information processing efficiency is improved.
(5) The CNN abnormal recognition model comprises an input layer, a convolution layer, a batch normalization layer, a ReLU activation layer, a maximum pooling layer, a full connection layer and a classification layer, a specific model with complete structure for solving the problem of recognition of abnormal states of the power distribution network is constructed, and the capability of the CNN model for processing specific problems is improved.
(6) According to the scheme, when the classification result is that a certain line in the power distribution network fails, the line is divided into a plurality of sections, the difference value of the very low frequency zero sequence currents at the head end and the tail end of each section is calculated, the section with the largest difference value is determined as the failure section, and the failure position can be further accurately judged. In addition, specific basis is provided for taking measures to eliminate faults by determining fault intervals on fault lines, and the power supply reliability of the system is guaranteed.
(7) The scheme can be applied to a power distribution network with a neutral point which is not grounded, the neutral point is grounded through the arc suppression coil, or the neutral point is grounded through a small resistor, and the application range is wider. The fault identification process does not change the original circuit, so that potential influence of the changed circuit on stable power supply of the power system can be avoided, and abnormal operation of other equipment in the system is prevented.
Drawings
Fig. 1 is a schematic diagram of a power distribution network according to an embodiment of the present invention.
Fig. 2 is a flowchart of a method for identifying abnormal states of a power distribution network according to an embodiment of the present invention.
Fig. 3 is a schematic network structure diagram of a CNN anomaly identification model according to an embodiment of the present invention.
Fig. 4a and fig. 4b are diagrams illustrating the result of the method for identifying abnormal states of the power distribution network according to the embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. In addition, the technical features of the embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
In a general power distribution network, when a single-phase earth fault occurs to a line, a system power supply is equivalent to a zero sequence voltage signal due to abrupt change of the system state, a step signal is superimposed on the zero sequence voltage signal, the step signal contains rich frequency spectrums, a transient state-decayable very low frequency current signal is generated on a zero sequence circuit of the power distribution network, the frequency of the signal is lower than a 50Hz power frequency signal of a normal system, the energy is strongest in a frequency band close to direct current, and the current of the frequency cannot be compensated by the inductance of an arc suppression coil; for a normal line, the low-frequency impedance of the capacitance to the ground is very large, low-frequency current is difficult to pass, but for a fault line, a fault point generally has a resistive component, so that most of the low-frequency current flows through the fault line, and whether the extremely low-frequency current exists on a detection line can judge the fault line; the invention selects the extremely low frequency signal as the characteristic signal for identifying the fault line, can effectively avoid the problem of unobtrusive fault characteristics caused by resonance between the arc suppression coil and the line capacitance in the existing system, and effectively improves the accuracy of fault line identification in the power distribution network.
The extremely low frequency signal is less influenced by the capacitance and inductance devices in the power distribution network, and primary circuit connection does not need to be changed, so that the abnormal state identification device of the power distribution network carries out fault identification according to the extremely low frequency signal, and the accuracy is higher than that of the existing high frequency line selection methods such as the fifth harmonic component line selection method. The power frequency zero sequence transformer is only installed in a general power distribution network, a sensor for measuring extremely low frequency is not arranged, the response of the power frequency zero sequence transformer to power frequency and frequency doubling signals is good, the response to low frequency is poor, and particularly when the low frequency current content is extremely low, the measurement precision is difficult to reach the required requirement only by means of the power frequency transformer, and the extremely low frequency sensor is additionally arranged. The very low frequency current sensor can measure very low frequency zero sequence current with very small content under the interference of relatively large power frequency current, and the very low frequency current sensor is used as a measuring device for measuring very low frequency signals of a line, so that the accuracy of very low frequency signal measurement can be ensured, and the accuracy of fault line identification in a power distribution network is improved.
The power frequency of the power distribution network is 50Hz, because the current frequency band concerned by the invention is extremely low frequency, the interference of 50Hz is required to be eliminated, the extremely low frequency current close to the direct current frequency band is measured, correspondingly, the measuring frequency band is concentrated in the low frequency band far lower than the power frequency signal, so that the extremely low frequency signal close to the direct current in the power frequency signal can be accurately measured, and the power frequency current and the extremely low frequency current are distinguished.
In one embodiment, a method for identifying an abnormal state of a power distribution network is provided, including: s1: when the current abrupt change value on any line exceeds a set threshold value, intercepting the extremely low frequency zero sequence current of all lines and the power frequency zero sequence voltage of each bus in a preset period before and after the abrupt change moment; the current abrupt change value is the difference value between the very low frequency zero sequence current at the current moment and the previous moment; s2: calculating a current compensation coefficient according to the variation amount of the power frequency zero sequence voltage of each bus in a preset period; s3: compensating the very low frequency zero sequence current of each line under each frequency by using a current compensation coefficient to obtain the target very low frequency current of each line under each frequency band at the abrupt moment; s4: constructing a training sample by using the power frequency zero sequence voltage of each busbar and the target very low frequency current of each line at the abrupt moment; s5: training the CNN abnormal recognition model by utilizing a plurality of training samples; s6: when the current mutation value of any line exceeds a set threshold, a corresponding current sample is obtained according to S1-S4, and the current sample is input into a trained CNN abnormal recognition model to enable the CNN abnormal recognition model to output a classification result, so that abnormal state recognition of the power distribution network is realized; the classification result includes: and judging the abnormal state of the power distribution network as a line fault or instantaneous interference of the whole power distribution network.
For example, a certain domestic transformer substation history data with 4 lines and 1 bus is used as training set data to form an initial data set, and a single sample in the data set is characterized by an extremely low frequency zero sequence current I n (n=1-4) of each line measured by a measuring device and a bus power frequency zero sequence voltage U, and the tag is known. Regarding all lines, taking the difference between the current time T 0 and the extremely low frequency zero sequence current of the previous time T 0 -DeltaT as a current abrupt change value DeltaI n(T0), and intercepting the extremely low frequency zero sequence current and bus power frequency zero sequence voltage data of all lines 20 seconds before and after the abrupt change time when DeltaI n(T0) exceeds a set threshold value I thre. Further, the zero sequence current data with the frequency below 1Hz, which is 20 seconds before and after, can be selected, and meanwhile, the waveform data of the bus power frequency zero sequence voltage at the same moment can be intercepted.
As shown in fig. 1, lines 1-4 respectively represent lines in a power distribution network, wherein a neutral point is grounded through an arc suppression coil, a transformer is in a star-delta connection method, an extremely low frequency current sensor for measuring extremely low frequency signals can be installed on each line, a power frequency zero sequence voltage sensor is also installed on a bus, and a computer terminal of an abnormal state identification device of the power distribution network is connected with the extremely low frequency current sensor and the power frequency zero sequence voltage sensor. For example, the time sequence of the very low frequency zero sequence current obtained on each line is denoted as I n (n is the line number, the time sequence of the current of the line 1 is denoted as I 1, and so on), the time sequence of the power frequency zero sequence voltage obtained on the bus is denoted as U m (m is the bus number, if there is only one bus, the time sequence of the voltage of the bus is denoted as U 1, and so on); respectively acquiring waveforms before and after the moment generated by the maximum value of the extremely low frequency zero sequence current of each line through signal processing, and simultaneously acquiring waveforms before and after the maximum value of the zero sequence voltage of the bus; and respectively finding out the specific time t 1、t2 of the circuit where the maximum value of the extremely low frequency zero sequence current is located and the specific time t 1、t2 of the bus where the maximum value of the power frequency zero sequence voltage is located, shifting the fault voltage waveform forward or backward by the number of data points corresponding to the time difference according to the time difference of the specific time t 1、t2 and the specific time t 1、t2, and supplementing the rest part by a linear interpolation method, thereby realizing the alignment of the fault voltage current waveform. The aligned time sequence of the very low frequency zero sequence current and the power frequency zero sequence voltage is respectively recorded as I an、Uan; obtaining an electrical characteristic pseudo graph P according to P= [ I a1;Ia2;Ia3;Ia4;Ua1 ] (taking n=4 and m=1 as an example); inputting the normalized electrical characteristic pseudo-graph into a neural network, and identifying one with the largest line weight as a fault line or instantaneous interference according to a classification result, wherein a CNN (computerized numerical control) very low frequency zero sequence current state identification model is shown in a specific flow chart in fig. 2, and a structural schematic diagram of the CNN abnormal identification model is shown in fig. 3; the training set data are used for training the model, the accuracy of model classification is checked by the test set data, and according to the classification result and the confusion matrix, the method can be used for improving the accuracy of line fault identification in the power distribution network, as shown in fig. 4a and 4b, and the method can be suitable for power systems of different scales and has strong generalization capability.
Assuming that a single-phase ground fault occurs in the line 1 at a certain moment, the waveforms measured by the very low frequency current sensors and the zero sequence voltage waveforms of the buses are recorded and transmitted to a computer terminal; the computer terminal processes the acquired ultra-low frequency zero sequence current and bus power frequency zero sequence voltage steady-state waveforms to give the ultra-low frequency zero sequence current value of each line and the bus power frequency zero sequence voltage value; and aligning the data according to the time difference of the time at which the maximum value of the two is positioned. Taking data collected for 20s as an event, collecting 207 groups of single-phase earth faults and instantaneous interference data of different lines of a certain transformer substation in China, forming an electrical characteristic pseudo-graph P according to P= [ I a1;Ia2;Ia3;Ia4;Ua1 ] and normalizing, inputting the electrical characteristic pseudo-graph of a normalized training set into a CNN abnormal recognition model to complete training of the model, inputting a test set into the trained CNN abnormal recognition model to test classification effects of the test set, wherein as shown in a table 1, reference numerals 1-4 respectively represent fault lines as lines 1-4, reference numeral 5 represents instantaneous interference, and the classification result shows that the accuracy of the invention on line fault recognition of the power distribution network is extremely high;
；
In one embodiment, S2 comprises: s21: determining a busbar power frequency zero sequence voltage mutation value corresponding to the mutation moment from the power frequency zero sequence voltage of each busbar in a preset period; s22: and calculating a current compensation coefficient according to the relation between the bus power frequency zero sequence voltage abrupt change value and the maximum value of the bus power frequency zero sequence voltage in a preset period.
In one embodiment, S22 includes: calculating a current compensation coefficient alpha by using the formula alpha=sin -1[U(T0)/Umax ]; u (T 0) is a busbar power frequency zero sequence voltage mutation value corresponding to a mutation time T 0; u max is the maximum value of the power frequency zero sequence voltage of the bus within a preset period.
In one embodiment, S3 comprises: calculating a target very low frequency current I' n (omega) of the ith line at the frequency omega at the abrupt moment by using a formula I'i(ω)=Ii(ω)×{(ω0 2-ω2)/[(ω0cosα)2+(ωsinα)2]1/2}; wherein, I i (omega) and I' i (omega) are respectively the very low frequency zero sequence current before and after the I line compensates on the frequency omega, omega 0 is the power frequency, omega < < omega 0, and alpha is the current compensation coefficient.
In one embodiment, S4 comprises: combining the target very low frequency current of each line at the abrupt change moment and the power frequency zero sequence voltage sequence of each bus into a multidimensional matrix with a row vector as a time sequence and a column vector as a characteristic quantity; mapping the data of each row in the multidimensional matrix to [0,1] for normalization to obtain a feature vector, converting the feature vector into an electrical feature pseudo-graph, and taking the electrical feature pseudo-graph as a training sample.
In one embodiment, S5 comprises: taking each training sample as input, taking a classification result as output, and training a CNN abnormal recognition model; the CNN anomaly identification model comprises an input layer, a convolution layer, a batch normalization layer, a ReLU activation layer, a maximum pooling layer, a full connection layer and a classification layer. Mapping the output of the CNN output layer to the [0,1] interval through a Softmax function, translating the output into class probability, measuring the difference between the output of the model and the real label by using a cross entropy loss function, improving the generalization capability of the model, and storing the CNN abnormal recognition model with the minimum cross entropy as a trained CNN abnormal recognition model after repeated iterative training.
It should be noted that, the network structure of the CNN anomaly identification model is set as follows: the input layer is 105Two hidden layers are 16 convolution kernels with a size/>The activation layer is a ReLU linear correction unit, and the maximum pooling layer is the size/>A convolution layer with a step length of 2; and 32 convolution kernels of size/>The activation layer is a ReLU linear correction unit, and the maximum pooling layer is the size/>A convolution layer with a step size of 2. The super parameters are set as follows: adopting Adam gradient descent algorithm, the maximum training times is 100 times, the batch size of each training treatment is 128, the initial learning rate is 1×10 -3, the L2 regularization parameter is 1×10 -4, the learning rate descent factor is 0.1, the learning rate is 1×10 -4 after 50 training, and each training breaks up the data set.
In one embodiment, the method for identifying abnormal states of the power distribution network further includes: when the classification result is that a certain line in the power distribution network has faults, the line is divided into a plurality of sections, the difference value of the very low frequency zero sequence currents at the head end and the tail end of each section is calculated, and the section with the largest difference value is determined as the fault section. The embodiment combines the installation position to determine the fault interval on the fault line, provides specific basis for taking measures to eliminate faults, and is beneficial to ensuring the power supply reliability of the system; the distribution network generally has multiple branch lines, the measuring device can be selectively installed on branch nodes, and the extremely low frequency zero sequence current measured before and after the fault point is greatly different because the extremely low frequency zero sequence current is generally discharged through the ground at the fault point, so that the estimation of the fault interval is realized.
In one embodiment, the neutral point of the distribution network is not grounded, or its neutral point is grounded via an arc suppression coil, or its neutral point is grounded via a small resistor.
In one embodiment, an abnormal state identification device of a power distribution network is provided, including:
The intercepting module is used for intercepting the extremely low-frequency zero sequence current of all lines and the power frequency zero sequence voltage of each bus in a preset period before and after the abrupt change moment when the abrupt change value of the current on any line exceeds a set threshold value; the current abrupt change value is the difference value between the very low frequency zero sequence current at the current moment and the previous moment;
the calculation module is used for calculating a current compensation coefficient according to the change amount of the power frequency zero sequence voltage of each bus in a preset period;
the compensation module is used for compensating the very low frequency zero sequence current of each line under each frequency by utilizing the current compensation coefficient to obtain the target very low frequency current of each line under each frequency band at the abrupt moment;
The construction module is used for constructing a training sample by utilizing the power frequency zero sequence voltage of each busbar at the abrupt moment and the target very low frequency current of each line;
the training model is used for training the CNN abnormal recognition model by utilizing a plurality of training samples;
The identification module is used for acquiring a corresponding current sample when the current mutation value of any line exceeds a set threshold value, inputting the current sample into a trained CNN abnormal identification model so as to enable the current sample to output a classification result, and realizing the abnormal state identification of the power distribution network; the classification result includes: and judging the abnormal state of the power distribution network as a line fault or instantaneous interference of the whole power distribution network.
In one embodiment, a power distribution network system is provided, including a memory storing a computer program and a processor implementing the steps of a method for identifying an abnormal state of a power distribution network when the processor executes the computer program.
In one embodiment, a computer readable storage medium is provided having a computer program stored thereon, which when executed by a processor, implements the steps of a method for identifying an abnormal state of a power distribution network.
It will be readily appreciated by those skilled in the art that the foregoing description is merely a preferred embodiment of the invention and is not intended to limit the invention, but any modifications, equivalents, improvements or alternatives falling within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (7)
1. The abnormal state identification method of the power distribution network is characterized by comprising the following steps of:
S1: when the current abrupt change value on any line exceeds a set threshold value, intercepting the extremely low frequency zero sequence current of all lines and the power frequency zero sequence voltage of each bus in a preset period before and after the abrupt change moment; the current abrupt change value is the difference value between the very low frequency zero sequence current at the current moment and the previous moment; the extremely low frequency zero sequence current is zero sequence current data within 1 hz;
S2: calculating a current compensation coefficient according to the change amount of the power frequency zero sequence voltage of each bus in the preset period; the step S2 comprises the following steps: s21: determining a busbar power frequency zero sequence voltage mutation value corresponding to the mutation moment from the power frequency zero sequence voltage of each busbar in the preset period; s22: calculating the current compensation coefficient according to the relation between the bus power frequency zero sequence voltage abrupt change value and the maximum value of the bus power frequency zero sequence voltage in the preset period; the S22 includes: using the formula
Α=sin -1[U(T0)/Umax ] calculating the current compensation coefficient α; u (T 0) is a busbar power frequency zero sequence voltage mutation value corresponding to the mutation time T 0; u max is the maximum value of the bus power frequency zero sequence voltage in the preset period;
s3: compensating the very low frequency zero sequence current of each line under each frequency by using the current compensation coefficient to obtain the target very low frequency current of each line under each frequency band at the abrupt moment; the step S3 comprises the following steps: using the formula
I'i(ω)=Ii(ω)×{(ω0 2-ω2)/[(ω0cosα)2+(ωsinα)2]1/2} Calculating the target very low frequency current I 'n(ω);Ii (omega) and I' i (omega) of the ith line at the frequency omega at the abrupt moment, wherein the target very low frequency zero sequence current I 'n(ω);Ii (omega) and the target very low frequency zero sequence current I' i (omega) are respectively the very low frequency zero sequence currents before and after the ith line compensates at the frequency omega, omega 0 is power frequency, and alpha is the current compensation coefficient;
S4: constructing a training sample by utilizing the power frequency zero sequence voltage of each busbar and the target very low frequency current of each line at the abrupt change moment;
S5: training the CNN abnormal recognition model by using a plurality of training samples;
S6: when the current mutation value of any line exceeds the set threshold, a corresponding current sample is obtained according to S1-S4, and the current sample is input into a trained CNN abnormal recognition model to output a classification result, so that abnormal state recognition of the power distribution network is realized; the classification result comprises: and judging the abnormal state of the power distribution network as a line fault or instantaneous interference of the whole power distribution network.
2. The method for identifying abnormal states of a power distribution network according to claim 1, wherein S4 comprises:
Combining the target extremely low frequency current of each line at the abrupt change moment and the power frequency zero sequence voltage sequence of each bus into a multidimensional matrix with a row vector as a time sequence and a column vector as a characteristic quantity;
mapping the data of each row in the multidimensional matrix to [0,1] for normalization to obtain a feature vector, converting the feature vector into an electrical feature pseudo-graph, and taking the electrical feature pseudo-graph as one training sample.
3. The method for identifying abnormal states of a power distribution network according to claim 2, wherein S5 comprises:
Taking each training sample as input, taking a classification result as output, and training the CNN abnormal recognition model; the CNN anomaly identification model comprises an input layer, a convolution layer, a batch normalization layer, a ReLU activation layer, a maximum pooling layer, a full connection layer and a classification layer.
4. A method of identifying an abnormal state of a power distribution network according to any one of claims 1 to 3, wherein the method of identifying an abnormal state of a power distribution network further comprises:
When the classification result is a fault of a certain line in the power distribution network, dividing the line into a plurality of sections, calculating the difference value of the very low frequency zero sequence currents at the head end and the tail end of each section, and determining the section with the largest difference value as the fault section.
5. A method for identifying abnormal states of an electrical distribution network according to any one of claims 1 to 3,
The neutral point of the power distribution network is not grounded, or the neutral point of the power distribution network is grounded through an arc suppression coil, or the neutral point of the power distribution network is grounded through a small resistor.
6. An abnormal state recognition device of a power distribution network, for executing the abnormal state recognition method of a power distribution network according to any one of claims 1 to 5, comprising:
The intercepting module is used for intercepting the extremely low-frequency zero sequence current of all lines and the power frequency zero sequence voltage of each bus in a preset period before and after the abrupt change moment when the abrupt change value of the current on any line exceeds a set threshold value; the current abrupt change value is the difference value between the very low frequency zero sequence current at the current moment and the previous moment;
the calculation module is used for calculating a current compensation coefficient according to the change amount of the power frequency zero sequence voltage of each bus in the preset period;
the compensation module is used for compensating the very low frequency zero sequence current of each line under each frequency by utilizing the current compensation coefficient to obtain the target very low frequency current of each line under each frequency band at the abrupt moment;
The construction module is used for constructing a training sample by utilizing the power frequency zero sequence voltage of each busbar at the abrupt change moment and the target very low frequency current of each line;
the training model is used for training the CNN abnormal recognition model by utilizing a plurality of training samples;
The identification module is used for acquiring a corresponding current sample when the current mutation value of any line exceeds the set threshold value, inputting the current sample into a trained CNN abnormal identification model to enable the current sample to output a classification result, and realizing abnormal state identification of the power distribution network; the classification result comprises: and judging the abnormal state of the power distribution network as a line fault or instantaneous interference of the whole power distribution network.
7. A power distribution network system comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410075689.8A CN117590158B (en) | 2024-01-18 | 2024-01-18 | Abnormal state identification method, device and control system of power distribution network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410075689.8A CN117590158B (en) | 2024-01-18 | 2024-01-18 | Abnormal state identification method, device and control system of power distribution network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117590158A CN117590158A (en) | 2024-02-23 |
| CN117590158B true CN117590158B (en) | 2024-04-19 |
Family
ID=89922353
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410075689.8A Active CN117590158B (en) | 2024-01-18 | 2024-01-18 | Abnormal state identification method, device and control system of power distribution network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117590158B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101223883B1 (en) * | 2012-11-09 | 2013-01-17 | 목포해양대학교 산학협력단 | Apparatus and method for diagnostic medium voltage cable status using the vlf td measured data |
| CN108279364A (en) * | 2018-01-30 | 2018-07-13 | 福州大学 | Wire selection method for power distribution network single phase earthing failure based on convolutional neural networks |
| CN110736899A (en) * | 2019-11-25 | 2020-01-31 | 深圳供电局有限公司 | Small current grounding fault positioning method and system, monitoring device, equipment and medium |
| CN111381129A (en) * | 2020-03-05 | 2020-07-07 | 华中科技大学 | Ground fault line and type identification method and device based on ultralow frequency signal |
| CN111398732A (en) * | 2020-03-14 | 2020-07-10 | 华中科技大学 | Power distribution network system based on active control of ground potential fluctuation and fault identification method thereof |
| RU2734164C1 (en) * | 2019-11-07 | 2020-10-13 | Федеральное государственное автономное образовательное учреждение высшего образования "Уральский федеральный университет имени первого Президента России Б.Н. Ельцина" | Method of detecting single-phase earth faults in distribution network connections |
| CN113933647A (en) * | 2021-09-23 | 2022-01-14 | 上海宏力达信息技术股份有限公司 | Fault line selection method of low-current grounding system based on first half-wave power direction |
| CN114784769A (en) * | 2022-05-19 | 2022-07-22 | 国网陕西省电力有限公司西安供电公司 | Distribution network fault partition isolation method based on single-phase earth fault zero-sequence current disturbance |
| CN116298665A (en) * | 2022-12-16 | 2023-06-23 | 国网新疆电力有限公司哈密供电公司 | Distribution cable arc light grounding fault judging method and system |
-
2024
- 2024-01-18 CN CN202410075689.8A patent/CN117590158B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101223883B1 (en) * | 2012-11-09 | 2013-01-17 | 목포해양대학교 산학협력단 | Apparatus and method for diagnostic medium voltage cable status using the vlf td measured data |
| CN108279364A (en) * | 2018-01-30 | 2018-07-13 | 福州大学 | Wire selection method for power distribution network single phase earthing failure based on convolutional neural networks |
| RU2734164C1 (en) * | 2019-11-07 | 2020-10-13 | Федеральное государственное автономное образовательное учреждение высшего образования "Уральский федеральный университет имени первого Президента России Б.Н. Ельцина" | Method of detecting single-phase earth faults in distribution network connections |
| CN110736899A (en) * | 2019-11-25 | 2020-01-31 | 深圳供电局有限公司 | Small current grounding fault positioning method and system, monitoring device, equipment and medium |
| CN111381129A (en) * | 2020-03-05 | 2020-07-07 | 华中科技大学 | Ground fault line and type identification method and device based on ultralow frequency signal |
| CN111398732A (en) * | 2020-03-14 | 2020-07-10 | 华中科技大学 | Power distribution network system based on active control of ground potential fluctuation and fault identification method thereof |
| CN113933647A (en) * | 2021-09-23 | 2022-01-14 | 上海宏力达信息技术股份有限公司 | Fault line selection method of low-current grounding system based on first half-wave power direction |
| CN114784769A (en) * | 2022-05-19 | 2022-07-22 | 国网陕西省电力有限公司西安供电公司 | Distribution network fault partition isolation method based on single-phase earth fault zero-sequence current disturbance |
| CN116298665A (en) * | 2022-12-16 | 2023-06-23 | 国网新疆电力有限公司哈密供电公司 | Distribution cable arc light grounding fault judging method and system |
Non-Patent Citations (3)
| Title |
|---|
| L. Hu.An Abnormal State Detection Method for Power Distribution Network Based on Big Data Technology.《2018 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery》.全文. * |
| 基于极低频信号的小电阻接地***单相接地故障识别方法研究;彭俊然;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20230115(第01期);全文 * |
| 基于空间多维度数据驱动的单相接地故障识别研究;杨世武;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20230115(第01期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117590158A (en) | 2024-02-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113219300B (en) | Power distribution network single-phase earth fault sensing method based on phase current transient state steady state | |
| CN106990324B (en) | Power distribution network ground fault detection and positioning method | |
| CN109001593B (en) | Fault recording control method for power distribution network | |
| CN109613374B (en) | Capacitor comprehensive online monitoring method based on redundant data | |
| CN111812451A (en) | Phase current transient fault component-based distributed line selection method for power distribution network | |
| CN112485595B (en) | Power distribution network ground fault line selection protection method and device | |
| CN110308369A (en) | A kind of the power distribution network intelligence sensor and fault detection algorithm of Convergence gateway function | |
| CN112485716B (en) | Line selection method based on zero-rest transient characteristic signal of ground fault arc current | |
| CN111398871A (en) | Device and method for checking polarity of zero sequence current transformer | |
| CN110398649B (en) | Method for online detecting transformer winding deformation based on voltage difference/current trace diagram | |
| CN115656702A (en) | Power distribution network single-phase earth fault positioning method and system based on edge calculation | |
| CN117590158B (en) | Abnormal state identification method, device and control system of power distribution network | |
| JP3550125B2 (en) | Transmission line fault monitoring device | |
| CN112034283B (en) | Device, system and process for detecting and positioning aluminum electrolysis cell ground fault | |
| CN113899980A (en) | Power distribution network single-phase earth fault section positioning method and system | |
| CN111398730A (en) | Power distribution network based on passive injection direct current signal and fault identification method thereof | |
| CN115598563A (en) | Power distribution network single-phase earth fault diagnosis method based on rough neural network | |
| CN113671315B (en) | ITn power supply insulation fault positioning method based on proportional differential principle | |
| CN115856505A (en) | Location positioning method and device for single-phase earth fault of active power distribution network | |
| CN112505475B (en) | Low-cost non-contact type overhead transmission line fault interval positioning method and system | |
| CN114252736A (en) | Active power distribution network single-phase fault line selection method based on background harmonic | |
| CN114865601A (en) | Fault judgment method and system based on variable quantity criterion | |
| CN113848439A (en) | Fault arc detection method and device, computer equipment and storage medium | |
| CN110888019B (en) | Power distribution network single-phase earth fault positioning method and system by utilizing line characteristic correction | |
| Liu et al. | Single-phase Grounding Fault Line Selection Method Based on the Difference of Electric Energy Information Between the Distribution End and the Load End |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |