CN118114019A - Automatic identification method and system for power distribution network topology based on data analysis - Google Patents
Automatic identification method and system for power distribution network topology based on data analysis Download PDFInfo
- Publication number
- CN118114019A CN118114019A CN202410224655.0A CN202410224655A CN118114019A CN 118114019 A CN118114019 A CN 118114019A CN 202410224655 A CN202410224655 A CN 202410224655A CN 118114019 A CN118114019 A CN 118114019A
- Authority
- CN
- China
- Prior art keywords
- frequency
- current signal
- data
- power distribution
- distribution network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000007405 data analysis Methods 0.000 title claims abstract description 22
- 238000005070 sampling Methods 0.000 claims abstract description 59
- 230000003595 spectral effect Effects 0.000 claims abstract description 32
- 238000012545 processing Methods 0.000 claims abstract description 24
- 238000005516 engineering process Methods 0.000 claims abstract description 5
- 238000004458 analytical method Methods 0.000 claims description 25
- 238000001228 spectrum Methods 0.000 claims description 25
- 238000004422 calculation algorithm Methods 0.000 claims description 23
- 230000002159 abnormal effect Effects 0.000 claims description 22
- 230000006870 function Effects 0.000 claims description 21
- 238000001514 detection method Methods 0.000 claims description 15
- 238000004891 communication Methods 0.000 claims description 9
- 238000012423 maintenance Methods 0.000 claims description 8
- 230000009466 transformation Effects 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 7
- 238000010183 spectrum analysis Methods 0.000 claims description 7
- 230000001052 transient effect Effects 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000013480 data collection Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 238000009434 installation Methods 0.000 claims description 3
- 238000010801 machine learning Methods 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 230000000737 periodic effect Effects 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 claims description 3
- 230000001360 synchronised effect Effects 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 claims description 3
- 238000012800 visualization Methods 0.000 claims description 3
- 230000002265 prevention Effects 0.000 abstract description 3
- 238000013136 deep learning model Methods 0.000 abstract 1
- 238000012544 monitoring process Methods 0.000 description 4
- 238000003745 diagnosis Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
Abstract
The invention relates to the technical field of power system automation, in particular to a power distribution network topology automatic identification method and system based on data analysis, comprising the following steps: s1: collecting current signal data in real time, and providing necessary original input for topology identification; s2: performing fast fourier transform on the current signal acquired in S1; s3: selecting proper sampling frequency and sampling point number for signal processing; s4: identifying and recording key spectral features of the current signal; s5: identifying different types of current signal characteristics; s6: and (5) constructing a low-voltage transformer area topological graph of the power distribution network based on the identification result in the step (S5). According to the invention, by combining real-time data acquisition, advanced signal processing technology and deep learning model, the automatic and accurate identification of the power distribution network topology is realized, the efficiency and the intelligent level of power distribution network management are obviously improved, and meanwhile, powerful technical guarantee is provided for the stable operation and fault prevention of a power system.
Description
Technical Field
The invention relates to the technical field of power system automation, in particular to a power distribution network topology automatic identification method and system based on data analysis.
Background
In the current technical field of power system automation, efficient management and operation and maintenance of a power distribution network are key to achieving reliability, efficiency and intelligence of the power system, the power distribution network is used as the endmost part of the power system which is directly connected with a user, and the running state of the power distribution network directly influences the stability and safety of power supply, so that the topology structure of the power distribution network is accurately identified, the running state of the power distribution network is monitored and analyzed in real time, and the method has important significance in optimizing management of the power distribution network, improving the reliability of power supply and preventing and rapidly responding to power grid faults.
However, one of the main challenges faced by the power distribution network is that the structure and the running state of the power distribution network are complex and changeable, the topology structure of the power distribution network is frequently changed due to new user access, equipment replacement or maintenance, emergency handling and other reasons, which brings great difficulty to real-time monitoring and management of the power distribution network, and in addition, the traditional power distribution network topology identification method is mostly dependent on manual operation and offline analysis, so that the efficiency is low, and quick response and accurate reflection of dynamic changes of the power distribution network are difficult to realize, and a technical scheme capable of automatically and real-time identifying the topology structure of the power distribution network is needed to face the challenges.
Disclosure of Invention
Based on the above purpose, the invention provides a method and a system for automatically identifying the topology of a power distribution network based on data analysis.
A power distribution network topology automatic identification method based on data analysis comprises the following steps:
s1: installing intelligent sensors at key nodes of the power distribution network, collecting current signal data in real time, and providing necessary original input for topology identification;
S2: performing fast fourier transform on the current signals acquired in S1, and converting each current signal from a time domain to a frequency domain;
s3: selecting proper sampling frequency and sampling point number to process signal according to the fast Fourier transform result of S2;
s4: analyzing the frequency domain information obtained in the step S3, and identifying and recording key frequency spectrum characteristics of the current signal, wherein the key frequency spectrum characteristics comprise frequency components and amplitude;
S5: utilizing the spectral characteristics obtained by analysis in the step S4, automatically classifying and identifying the current signals through a pre-trained machine learning model so as to identify current signal characteristics of different types, and associating the identified current signal characteristics with corresponding electric equipment or network paths;
S6: constructing a low-voltage transformer area topological graph of the power distribution network based on the identification result in the step S5, wherein the topological graph is used for reflecting the electric connection relation in the power distribution transformer area;
S7: synchronizing the low-voltage station topological graph generated in the step S6 with actual physical equipment and network states, and ensuring that the positions of each equipment and connection in the topological graph are consistent with the actual states.
Further, the S1 specifically includes:
S11: firstly, determining key nodes in a power distribution network, wherein the key nodes comprise sensitive areas of a transformer outlet, a branch line starting point and an important load access point;
S12: installing an intelligent sensor at the key node determined in the step S11, wherein the intelligent sensor comprises a current transformer and a voltage transformer, can measure the amplitude and the phase of current and voltage at the same time, and supports broadband operation to capture transient and harmonic components of a current signal;
S13: the intelligent sensor is connected with the data acquisition unit and is used for converting analog signals captured by the sensor into digital signals, and the data acquisition unit is provided with a high-speed analog-to-digital converter and is used for ensuring high-precision and high-speed acquisition of current and voltage signals;
s14: transmitting the current signal data acquired by the data acquisition unit in the step S13 to a central processing system through a preset communication module, wherein the communication module supports wired and wireless communication technologies so as to adapt to different power distribution network environments and installation conditions;
S15: after the central processing system receives the current signal data, preliminary processing including filtering, denoising and time synchronization processing of the signals is performed.
Further, the S2 specifically comprises
S21: receiving the current signal data processed in the step S15, and storing the received current signal data in a time sequence form, wherein the data stored in the time sequence form comprises the information of the amplitude and the phase of the current and the voltage, and the original input is prepared for Fourier transformation;
s22: preprocessing each piece of current signal data by using a window function, and specifically selecting a preset window function to window the data so as to reduce spectrum leakage caused by signal interception, wherein the preset window function comprises a hanning window or a hamming window;
S23: performing a fast fourier transform algorithm on the preprocessed current signal, and converting the signal from a time domain to a frequency domain, wherein the fast fourier transform algorithm has a formula as follows:
Where x (N) represents a current signal sample in the time domain, n=0, 1,..; x (k) represents the result in the frequency domain, k=0, 1,..n-1, corresponding to different frequency components; n is the number of sampling points, which determines the resolution of the frequency spectrum and the transformation range; w (n) is a window function, applied to each sampling point, for reducing the edge effect of the non-periodic signal fast fourier transform, improving the smoothness of the spectrum and the detection capability of the signal transient characteristic, e -j2πnk/N is a complex form twiddle factor, responsible for converting the time domain signal into the frequency domain;
s24: after the fast fourier transform is completed, the calculation result of the frequency domain signal is reserved, and the result comprises frequency components and corresponding amplitude and phase information.
Further, the step S3 specifically includes:
S31: analyzing the fast fourier transform result of step S2, in particular analyzing the frequency content in the frequency domain result X (k) to determine a highest frequency f max of the current signal, the analysis being used to ensure that the selected sampling frequency can meet the requirements of signal acquisition, the sampling frequency f s being at least twice the highest frequency, i.e. f s≥2fmax, according to the nyquist sampling theorem;
S32: selecting a sampling frequency f s to ensure that the sampling frequency f s meets the nyquist sampling theorem and leaves a proper margin to prevent the presence of unexpected high frequency components in the signal to ensure complete capture of the signal and allow for frequency fluctuations within a certain range;
S33: determining a sampling point number N, the selection of the sampling point number being based on the requirements of the fast fourier transform and the time length T of the signal to obtain a sufficient spectral resolution, the sampling point number N being selected according to a required minimum frequency resolution Δf, where Δf = 1/T, whereby for a given signal length and required spectral resolution the sampling point number N = f s x T;
s34: adjusting parameter settings of a fast fourier transform algorithm, wherein the parameter settings comprise a sampling frequency f s and a sampling point number N to optimize the execution of the fast fourier transform;
s35: the fast fourier transform of step S2 is re-performed using the adjusted sampling frequency and the number of sampling points to obtain an improved frequency domain signal analysis result.
Further, the step S4 specifically includes:
s41: extracting spectral data from the frequency domain information obtained in the step S3, wherein the spectral data is expressed in the form of a frequency component k and a corresponding amplitude X (k), and the spectral data reflects the energy distribution condition of the current signal under different frequencies;
s42: the extracted spectrum data is processed by applying a spectrum analysis algorithm to identify main frequency components and harmonics of the current signal, the specific spectrum analysis algorithm is a peak detection algorithm, and the peak detection algorithm is used for identifying frequency components with amplitude significantly higher than background noise, and has the formula:
P (k) = { k|x (k)) X (k-1) and X (k) > X (k+1) and X (k) > θ } where P (k) is a set of frequency components identified as peaks and θ is a threshold value for distinguishing effective signals from noise;
S43: carrying out amplitude analysis on each identified main frequency component and harmonic wave thereof, and determining the amplitude ratio of each identified main frequency component to the reference frequency;
S44: recording the identified frequency components and amplitude information thereof into a database, wherein each record comprises frequency values, amplitude values and time stamp information;
S45: the recorded spectral features are analyzed to identify typical patterns or anomalies of the current signal.
Further, the analyzing the recorded spectrum features in S45 specifically includes:
S451: the spectrum characteristic set of the current signal is defined as follows: f= { (F i,Ai) |i=1, 2, …, n }, where F i is the ith frequency component, a i is the corresponding amplitude value, n is the total number of frequency components, and by comparing the amplitude of each frequency component with its historical average amplitude or expected amplitude, to identify abnormal spectral features, the threshold for anomaly detection is set to τ, and the specific formula is: wherein/> The historical average amplitude value of the frequency component f i is the frequency component set identified as abnormal, and the threshold tau is determined according to actual conditions and empirical data and is used for controlling the sensitivity of abnormal detection;
S452: analyzing the relation between the abnormal frequency component E and the known typical mode, particularly by matching the known fault characteristic frequency or the harmonic mode, and marking the frequency component as abnormal of a corresponding type when f i is matched with the characteristic frequency in the known fault mode;
S453: finally, the identified abnormal situation and the typical pattern are recorded in a database.
Further, the step S5 specifically includes:
S51: constructing an input data set from the spectral features analyzed in step S4, each data point including a frequency component f i and its corresponding amplitude a i as a feature vector x= [ f 1,A1,f2,A2,…,fn,An ], where n is the total number of frequency components;
S52: designing a deep neural network DNN model, wherein the model structure comprises an input layer, a plurality of hidden layers and an output layer, the hidden layers use a ReLU activation function, the output layer uses a softmax function to carry out multi-classification, and the specific formula of the model is as follows:
y=softmax (W m·ReLU(Wm-1·(…ReLU(W1·X+b1)…)+bm-1)+bm), where W 1,...,Wm and b 1,...,bm are the weight and bias of the network layer, respectively, m is the number of hidden layers, Y is the prediction result of the output layer, representing the probability distribution of different types of current signals;
s53: pre-training a DNN model by using historical current signal data and known categories, and adjusting model parameters through supervised learning to maximize the classification accuracy of the model on a training set;
s54: inputting the feature vector X constructed in the step S51 by using the trained DNN model, outputting a prediction result Y, wherein the result Y is the probability of each category to which the current signal belongs, and determining the category of the current signal according to the category with the largest probability;
s55: according to the classification result of step S54, each current signal feature is associated with a corresponding powered device or network path.
Further, the step S6 specifically includes:
S61: collecting DNN model identification results in the step S5, wherein the DNN model identification results comprise the category of the current signal and the associated information of the corresponding electric equipment or network path, and the associated information is used as basic data for constructing a topological graph;
S62: determining all key equipment and network nodes in a low-voltage transformer area, including transformers, power distribution cabinets, branch lines and important loads, and allocating unique identifiers to each equipment and node so as to be represented in a topological graph;
S63: determining the electrical connection relation between each device and the node according to the association of the current signal characteristic and the device or the path identified in the step S61, and particularly when a certain current signal characteristic is matched with the fault type on a specific branch line, indicating that the branch line is electrically connected with the device related to the fault;
s64: and constructing a topological graph by using a graphical representation method, wherein the topological graph comprises nodes and edges, and particularly, drawing the edges in the graph to connect the related nodes by utilizing the connection relation determined in the step S63 so as to form a complete low-voltage area electric topological graph.
Further, the step S7 specifically includes:
S71: step S6, a generated topological graph of the low-voltage transformer area is obtained and is used for indicating all electric equipment and connection relations in the power distribution transformer area;
s72: collecting real-time data, including current, voltage, power and equipment state information, specifically through an intelligent sensor and a data collection system;
S73: analyzing the collected real-time data, and determining the running state of each electrical device and the actual load condition of the network;
s74: updating the device state and the network load condition obtained by analysis into a low-voltage station area topological graph in the step S71, wherein the topological graph reflects the current physical state of the power distribution network;
s75: and displaying the updated low-voltage area topological graph through a visualization tool, and providing visual network structure and equipment state information for operation and maintenance personnel.
The utility model provides a distribution network topology automatic identification system based on data analysis, is used for realizing the distribution network topology automatic identification method based on data analysis, includes:
And a data acquisition module: the method comprises the steps of acquiring real-time current signal data from key nodes of a power distribution network, and providing original input data for topology identification;
And a signal processing module: receiving the current signal data acquired by the data acquisition module, performing fast Fourier transform and window function processing on the current signal data, and converting the current signal from a time domain to a frequency domain;
and the feature extraction and classification module is used for: based on the frequency domain information obtained by the signal processing module, automatically classifying and identifying the current signals by utilizing a pre-trained deep neural network model;
Topology construction updating module: constructing and updating a topological graph of the low-voltage area according to the identification result of the feature extraction and classification module;
And a state synchronization module: the topology map of the low-voltage transformer area is used for keeping the topology map of the low-voltage transformer area generated by the topology construction and updating module synchronous with the actual physical equipment and network state of the power distribution network.
The invention has the beneficial effects that:
according to the invention, the topological structure of the power distribution network can be automatically and accurately identified by collecting the current signal data of the key nodes of the power distribution network in real time and combining the fast Fourier transform and the deep learning technology, the automatic identification process not only greatly improves the working efficiency and reduces the input of manpower resources, but also can reflect the latest state of the power distribution network in real time, and provides powerful support for the stable operation of a power system.
According to the invention, through intelligent analysis of the current signal characteristics of the power distribution network, abnormal conditions and potential faults in the power distribution network can be found and identified in time, so that technical guarantee is provided for fault prevention and quick response of a power system, and the early warning mechanism can obviously reduce the power interruption risk caused by the faults of the power distribution network and ensure the continuity and reliability of power supply.
According to the invention, not only are the management and operation and maintenance strategies of the power distribution network optimized, but also the intelligent level of the power system is improved, and through automatic topology identification and real-time state monitoring, operation and maintenance personnel of the power distribution network can more efficiently carry out power grid management and decision, and meanwhile, reliable data support and technical foundation are provided for intelligent upgrading and optimization of the power distribution network in the future.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only of the invention and that other drawings can be obtained from them without inventive effort for a person skilled in the art.
Fig. 1 is a schematic flow chart of a method for automatically identifying a topology of a power distribution network according to an embodiment of the present invention;
fig. 2 is a schematic diagram of an automatic topology identification system for a power distribution network according to an embodiment of the present invention.
Detailed Description
The present invention will be further described in detail with reference to specific embodiments in order to make the objects, technical solutions and advantages of the present invention more apparent.
It is to be noted that unless otherwise defined, technical or scientific terms used herein should be taken in a general sense as understood by one of ordinary skill in the art to which the present invention belongs. The terms "first," "second," and the like, as used herein, do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. The word "comprising" or "comprises", and the like, means that elements or items preceding the word are included in the element or item listed after the word and equivalents thereof, but does not exclude other elements or items.
As shown in fig. 1, a method for automatically identifying a topology of a power distribution network based on data analysis includes the following steps:
S1: installing intelligent sensors at key nodes of the power distribution network, and collecting current signal data in real time, wherein the data comprise amplitude and phase information of current, so that necessary original input is provided for topology identification;
S2: performing Fast Fourier Transform (FFT) on the current signals acquired in S1, and converting each current signal from a time domain to a frequency domain, so as to acquire spectral features of the signals, thereby ensuring that the characteristics of the current signals can be accurately analyzed in subsequent steps;
S3: selecting proper sampling frequency and sampling point number to process the signal according to the fast Fourier transform result of S2, thereby ensuring the complete capture of the signal characteristics and high-quality spectrum analysis;
s4: analyzing the frequency domain information obtained in the step S3, and identifying and recording key frequency spectrum characteristics of the current signal, including frequency components and amplitude, wherein the step provides scientific basis for accurate identification of a signal source;
S5: utilizing the spectral characteristics obtained by analysis in the step S4, automatically classifying and identifying the current signals through a pre-trained machine learning model so as to identify current signal characteristics of different types, and associating the identified current signal characteristics with corresponding electric equipment or network paths;
S6: based on the identification result in the step S5, a low-voltage transformer area topological graph of the power distribution network is constructed, and the topological graph is used for reflecting the electric connection relation in the power distribution transformer area and provides a basis for accurately managing and controlling the power distribution network;
S7: synchronizing the low-voltage station topological graph generated in the step S6 with actual physical equipment and network states, and ensuring that the positions of each equipment and connection in the topological graph are consistent with the actual states.
S1 specifically comprises:
S11: firstly, determining key nodes in a power distribution network, wherein the key nodes comprise sensitive areas of transformer outlets, branch line starting points and important load access points, and the key nodes are the positions with the most obvious current signal characteristic change and can provide key data for accurately monitoring the state of the whole power distribution network;
S12: installing an intelligent sensor at the key node determined in the step S11, wherein the intelligent sensor comprises a current transformer and a voltage transformer, can measure the amplitude and the phase of current and voltage at the same time, and supports broadband operation to capture transient and harmonic components of a current signal;
S13: the intelligent sensor is connected with the data acquisition unit and is used for converting analog signals captured by the sensor into digital signals, and the data acquisition unit is provided with a high-speed analog-to-digital converter and is used for ensuring high-precision and high-speed acquisition of current and voltage signals;
s14: transmitting the current signal data acquired by the data acquisition unit in the step S13 to a central processing system through a preset communication module, wherein the communication module supports wired (such as Ethernet) and wireless (such as LTE and 5G) communication technologies so as to adapt to different power distribution network environments and installation conditions;
S15: after the central processing system receives the current signal data, preliminary processing including filtering and denoising of the signal and time synchronization processing are performed to ensure the data quality and consistency of subsequent fast Fourier transformation and signal analysis.
S2 specifically comprises
S21: receiving the current signal data processed in the step S15, and storing the received current signal data in a time sequence form, wherein the data stored in the time sequence form comprises the information of the amplitude and the phase of the current and the voltage, and the original input is prepared for Fourier transformation;
S22: preprocessing each current signal data by using a window function, specifically selecting a preset window function to window the data so as to reduce spectrum leakage caused by signal interception, wherein the preset window function comprises a hanning window or a hamming window;
s23: performing a fast fourier transform algorithm on the preprocessed current signal, and converting the signal from a time domain to a frequency domain, wherein the fast fourier transform algorithm has the formula:
Where x (N) represents a current signal sample in the time domain, n=0, 1,..; x (k) represents the result in the frequency domain, k=0, 1,..n-1, corresponding to different frequency components; n is the number of sampling points, which determines the resolution of the frequency spectrum and the transformation range; w (n) is a window function, applied to each sampling point, for reducing the edge effect of the non-periodic signal fast fourier transform, improving the smoothness of the spectrum and the detection capability of the signal transient characteristic, e -j2πnk/N is a complex form twiddle factor, responsible for converting the time domain signal into the frequency domain;
s24: after the fast fourier transformation is completed, the calculation result of the frequency domain signal is reserved, wherein the result comprises frequency components and corresponding amplitude and phase information, and the frequency spectrum characteristics of the current signal are accurately captured and recorded as a basis for further analysis.
S3 specifically comprises:
S31: analyzing the fast fourier transform result of step S2, in particular analyzing the frequency content in the frequency domain result X (k) to determine a highest frequency f max of the current signal, the analysis being used to ensure that the selected sampling frequency can meet the requirements of signal acquisition, the sampling frequency f s being at least twice the highest frequency, i.e. f s≥2fmax, according to the nyquist sampling theorem;
S32: the sampling frequency f s is chosen to ensure that the sampling frequency f s meets the nyquist sampling theorem and leaves a suitable margin to prevent the presence of unexpected high frequency components in the signal, e.g., if f max from analysis is 50Hz, the sampling frequency chosen may be 120Hz or higher to ensure complete capture of the signal and allow for frequency fluctuations over a range;
S33: determining a sampling point number N, the selection of the sampling point number being based on the requirements of the fast fourier transform and the time length T of the signal to obtain a sufficient spectral resolution, the sampling point number N being selected according to a required minimum frequency resolution Δf, where Δf = 1/T, whereby for a given signal length and required spectral resolution the sampling point number N = f s x T;
S34: adjusting parameter settings of a fast fourier transform algorithm, adjusting sampling frequency fs and sampling point number N to optimize execution of the fast fourier transform, wherein the step ensures that the FFT algorithm can effectively process collected current signals, and simultaneously maximally reduces information loss and spectrum leakage;
S35: the fast Fourier transform of the step S2 is re-executed, and the adjusted sampling frequency and the sampling point number are used to obtain an improved frequency domain signal analysis result, so that the spectrum characteristics of the current signal are ensured to be accurately captured and analyzed, and high-quality data is provided for automatic topology identification of the power distribution network;
Through the steps S31-S35, the most suitable sampling frequency and sampling point number can be selected according to the specific characteristics of the signals, so that the signal processing process is optimized, the accuracy and the efficiency of signal analysis are ensured, the method has important significance for monitoring and managing the state of the power distribution network, and a reliable basis is provided for subsequent analysis and diagnosis of the power distribution network.
S4 specifically comprises the following steps:
s41: extracting spectral data from the frequency domain information obtained in the step S3, wherein the spectral data is expressed in the form of a frequency component k and a corresponding amplitude X (k), and the spectral data reflects the energy distribution condition of the current signal under different frequencies;
s42: the extracted spectrum data is processed by applying a spectrum analysis algorithm to identify main frequency components and harmonics of the current signal, the specific spectrum analysis algorithm is a peak detection algorithm, and the peak detection algorithm is used for identifying frequency components with amplitude significantly higher than background noise, and has the formula:
P (k) = { k|x (k) > X (k-1) and X (k) > X (k+1) and X (k) > θ } where P (k) is a set of frequency components identified as peaks and θ is a threshold value for distinguishing effective signals from noise;
S43: the amplitude analysis is carried out on each identified main frequency component and the harmonic wave thereof, the amplitude ratio of the main frequency component to the reference frequency (such as 50Hz or 60 Hz) is determined, the step is helpful for understanding the relative intensity of each main frequency component in the current signal, and clues are provided for the nature and the source of the current signal;
s44: recording the identified frequency components and their amplitude information into a database, each record including frequency values, amplitude values, and time-stamped information, which allows subsequent analysis to track spectral changes in the current signal at specific points in time;
s45: the recorded spectral features are analyzed to identify typical patterns or anomalies of the current signal, for example, transient events in the current signal or abnormal operating conditions in the system can be identified by comparing the spectral data at different points in time.
The spectral features of the analysis record in S45 specifically include:
S451: the spectrum characteristic set of the current signal is defined as follows: f= { (F i,Ai) |i=1, 2, …, n }, where F i is the ith frequency component, a i is the corresponding amplitude value, n is the total number of frequency components, and by comparing the amplitude of each frequency component with its historical average amplitude or expected amplitude, to identify abnormal spectral features, the threshold for anomaly detection is set to τ, and the specific formula is: wherein/> The historical average amplitude value of the frequency component f i is the frequency component set identified as abnormal, and the threshold tau is determined according to actual conditions and empirical data and is used for controlling the sensitivity of abnormal detection;
S452: analyzing the relation between the abnormal frequency component E and the known typical mode, specifically by matching the known fault characteristic frequency or the harmonic mode, and when f i is matched with the characteristic frequency in the known fault mode, marking the frequency component as the abnormal of the corresponding type, wherein the specific matching formula is as follows:
M= { (f i,T)|fi E and f i match the characteristic frequency of a specific failure mode, T is the failure type }, where M represents the identified abnormal frequency component matching the specific failure mode and its failure type;
s453: finally, the identified abnormal conditions and typical modes are recorded in a database, a basis is provided for subsequent maintenance and fault diagnosis, and through the analysis, the abnormal conditions and typical fault modes in the current signals can be effectively identified, and scientific data support is provided for reliable operation and fault prevention of the power distribution network.
S5 specifically comprises the following steps:
S51: constructing an input data set from the spectral features analyzed in step S4, each data point including a frequency component f i and its corresponding amplitude a i as a feature vector x= [ f 1,A1,f2,A2,…,fn,An ], where n is the total number of frequency components;
S52: designing a deep neural network DNN model, wherein the model structure comprises an input layer, a plurality of hidden layers and an output layer, the hidden layers use a ReLU activation function, the output layer uses a softmax function to carry out multi-classification, and the specific formula of the model is as follows:
Y＝softmax(Wm·ReLU(Wm-1·(…ReLU(W1·X+b1)…)+bm-1)+bm)，
Wherein, W 1,...,Wm and b 1,...,bm are respectively the weight and bias of the network layer, m is the number of hidden layers, Y is the prediction result of the output layer, and represents the probability distribution of different types of current signals;
s53: pre-training a DNN model by using historical current signal data and known categories, and adjusting model parameters (weights and biases) through supervised learning so as to maximize the classification accuracy of the model on a training set;
s54: inputting the feature vector X constructed in the step S51 by using the trained DNN model, outputting a prediction result Y, wherein the result Y is the probability of each category to which the current signal belongs, and determining the category of the current signal according to the category with the largest probability;
S55: according to the classification result of step S54, each current signal feature is associated with a corresponding powered device or network path, e.g. if the prediction result indicates that a certain current signal belongs to a specific fault type, the signal is associated with a device or network path that may cause such a fault.
S6 specifically comprises the following steps:
S61: collecting DNN model identification results in the step S5, wherein the DNN model identification results comprise the category of the current signal and the associated information of the corresponding electric equipment or network path, and the associated information is used as basic data for constructing a topological graph;
S62: determining all key equipment and network nodes in a low-voltage transformer area, including transformers, power distribution cabinets, branch lines and important loads, and allocating unique identifiers to each equipment and node so as to be represented in a topological graph;
S63: determining the electrical connection relation between each device and the node according to the association of the current signal characteristic and the device or the path identified in the step S61, and particularly when a certain current signal characteristic is matched with the fault type on a specific branch line, indicating that the branch line is electrically connected with the device related to the fault;
S64: a topology graph is constructed using a graphical representation method, wherein the topology graph comprises nodes (representing equipment and network nodes) and edges (representing electrical connection relationships between the nodes), and specifically, the edges are drawn in the graph to connect the relevant nodes by using the connection relationships determined in step S63, so as to form a complete low-voltage transformer area electrical topology graph.
S7 specifically comprises the following steps:
S71: step S6, a generated topological graph of the low-voltage transformer area is obtained and is used for indicating all electric equipment and connection relations in the power distribution transformer area;
S72: the method comprises the steps of collecting real-time data, including current, voltage, power and equipment state information, specifically through an intelligent sensor and a data collection system, wherein the data reflect the actual physical state and the running condition of the power distribution network;
S73: analyzing the collected real-time data, and determining the running state (such as opening, closing, fault and the like) of each electrical device and the actual load condition of the network, wherein the step is realized by comparing the real-time data with a preset threshold value or a pattern recognition algorithm;
s74: updating the device state and the network load condition obtained by analysis into a low-voltage station area topological graph in the step S71, wherein the topological graph reflects the current physical state of the power distribution network;
s75: displaying the updated low-voltage transformer area topological graph through a visualization tool, and providing visual network structure and equipment state information for operation and maintenance personnel;
Through the S71-S75, synchronization between the topological graph of the low-voltage transformer area and actual physical equipment and network states of the power distribution network is ensured, and efficiency and accuracy of power distribution network management and operation are improved.
As shown in fig. 2, an automatic identification system for a power distribution network topology based on data analysis is configured to implement the above automatic identification method for a power distribution network topology based on data analysis, and includes:
And a data acquisition module: the method comprises the steps of acquiring real-time current signal data from key nodes of a power distribution network, and providing original input data for topology identification;
And a signal processing module: receiving the current signal data acquired by the data acquisition module, performing fast Fourier transform and window function processing on the current signal data, and converting the current signal from a time domain to a frequency domain;
and the feature extraction and classification module is used for: based on the frequency domain information obtained by the signal processing module, automatically classifying and identifying the current signals by utilizing a pre-trained deep neural network model;
Topology construction updating module: according to the identification result of the feature extraction and classification module, constructing and updating a topology map of the low-voltage transformer area, wherein the topology construction and updating module matches the identified current signal features with actual physical equipment and connection relations in the power distribution network, and automatically generates the topology map of the low-voltage transformer area reflecting the electrical connection relations in the power distribution transformer area;
And a state synchronization module: the topology map of the low-voltage transformer area is used for keeping the topology map of the low-voltage transformer area generated by the topology construction and updating module synchronous with the actual physical equipment and network state of the power distribution network.
The present invention is intended to embrace all such alternatives, modifications and variances which fall within the broad scope of the appended claims. Therefore, any omission, modification, equivalent replacement, improvement, etc. of the present invention should be included in the scope of the present invention.
Claims (10)
1. The automatic identification method for the power distribution network topology based on the data analysis is characterized by comprising the following steps of:
s1: installing intelligent sensors at key nodes of the power distribution network, collecting current signal data in real time, and providing necessary original input for topology identification;
S2: performing fast fourier transform on the current signals acquired in S1, and converting each current signal from a time domain to a frequency domain;
s3: selecting proper sampling frequency and sampling point number to process signal according to the fast Fourier transform result of S2;
s4: analyzing the frequency domain information obtained in the step S3, and identifying and recording key frequency spectrum characteristics of the current signal, wherein the key frequency spectrum characteristics comprise frequency components and amplitude;
S5: utilizing the spectral characteristics obtained by analysis in the step S4, automatically classifying and identifying the current signals through a pre-trained machine learning model so as to identify current signal characteristics of different types, and associating the identified current signal characteristics with corresponding electric equipment or network paths;
S6: constructing a low-voltage transformer area topological graph of the power distribution network based on the identification result in the step S5, wherein the topological graph is used for reflecting the electric connection relation in the power distribution transformer area;
S7: synchronizing the low-voltage station topological graph generated in the step S6 with actual physical equipment and network states, and ensuring that the positions of each equipment and connection in the topological graph are consistent with the actual states.
2. The automatic identification method for the topology of the power distribution network based on the data analysis according to claim 1, wherein the step S1 specifically comprises:
S11: firstly, determining key nodes in a power distribution network, wherein the key nodes comprise sensitive areas of a transformer outlet, a branch line starting point and an important load access point;
S12: installing an intelligent sensor at the key node determined in the step S11, wherein the intelligent sensor comprises a current transformer and a voltage transformer, can measure the amplitude and the phase of current and voltage at the same time, and supports broadband operation to capture transient and harmonic components of a current signal;
S13: the intelligent sensor is connected with the data acquisition unit and is used for converting analog signals captured by the sensor into digital signals, and the data acquisition unit is provided with a high-speed analog-to-digital converter and is used for ensuring high-precision and high-speed acquisition of current and voltage signals;
s14: transmitting the current signal data acquired by the data acquisition unit in the step S13 to a central processing system through a preset communication module, wherein the communication module supports wired and wireless communication technologies so as to adapt to different power distribution network environments and installation conditions;
S15: after the central processing system receives the current signal data, preliminary processing including filtering, denoising and time synchronization processing of the signals is performed.
3. The method for automatically identifying a topology of a power distribution network based on data analysis according to claim 2, wherein S2 specifically comprises
S21: receiving the current signal data processed in the step S15, and storing the received current signal data in a time sequence form, wherein the data stored in the time sequence form comprises the information of the amplitude and the phase of the current and the voltage, and the original input is prepared for Fourier transformation;
s22: preprocessing each piece of current signal data by using a window function, and specifically selecting a preset window function to window the data so as to reduce spectrum leakage caused by signal interception, wherein the preset window function comprises a hanning window or a hamming window;
S23: performing a fast fourier transform algorithm on the preprocessed current signal, and converting the signal from a time domain to a frequency domain, wherein the fast fourier transform algorithm has a formula as follows:
where x (N) represents a current signal sample in the time domain, n=0, 1, …, N-1; x (k) represents the result in the frequency domain, k=0, 1, …, N-1, corresponding to different frequency components; n is the number of sampling points, which determines the resolution of the frequency spectrum and the transformation range; w (n) is a window function, applied to each sampling point, for reducing the edge effect of the non-periodic signal fast fourier transform, improving the smoothness of the spectrum and the detection capability of the signal transient characteristic, e -j2πnk/N is a complex form twiddle factor, responsible for converting the time domain signal into the frequency domain;
s24: after the fast fourier transform is completed, the calculation result of the frequency domain signal is reserved, and the result comprises frequency components and corresponding amplitude and phase information.
4. The automatic identification method for the topology of the power distribution network based on data analysis according to claim 3, wherein the step S3 specifically comprises:
S31: analyzing the fast fourier transform result of step S2, in particular analyzing the frequency content in the frequency domain result X (k) to determine a highest frequency f max of the current signal, the analysis being used to ensure that the selected sampling frequency can meet the requirements of signal acquisition, the sampling frequency f s being at least twice the highest frequency, i.e. f s≥2fmax, according to the nyquist sampling theorem;
S32: selecting a sampling frequency f s to ensure that the sampling frequency f s meets the nyquist sampling theorem and leaves a proper margin to prevent the presence of unexpected high frequency components in the signal to ensure complete capture of the signal and allow for frequency fluctuations within a certain range;
S33: determining a sampling point number N, the selection of the sampling point number being based on the requirements of the fast fourier transform and the time length T of the signal to obtain a sufficient spectral resolution, the sampling point number N being selected according to a required minimum frequency resolution Δf, where Δf = 1/T, whereby for a given signal length and required spectral resolution the sampling point number N = f s x T;
s34: adjusting parameter settings of a fast fourier transform algorithm, wherein the parameter settings comprise a sampling frequency f s and a sampling point number N to optimize the execution of the fast fourier transform;
s35: the fast fourier transform of step S2 is re-performed using the adjusted sampling frequency and the number of sampling points to obtain an improved frequency domain signal analysis result.
5. The automatic identification method for the topology of the power distribution network based on data analysis according to claim 4, wherein the step S4 specifically comprises:
s41: extracting spectral data from the frequency domain information obtained in the step S3, wherein the spectral data is expressed in the form of a frequency component k and a corresponding amplitude X (k), and the spectral data reflects the energy distribution condition of the current signal under different frequencies;
s42: the extracted spectrum data is processed by applying a spectrum analysis algorithm to identify main frequency components and harmonics of the current signal, the specific spectrum analysis algorithm is a peak detection algorithm, and the peak detection algorithm is used for identifying frequency components with amplitude significantly higher than background noise, and has the formula:
p (k) = { k|x (k) > X (k-1) and X (k) > X (k+1) and X (k) > θ } where P (k) is a set of frequency components identified as peaks and θ is a threshold for distinguishing effective signals from noise;
S43: carrying out amplitude analysis on each identified main frequency component and harmonic wave thereof, and determining the amplitude ratio of each identified main frequency component to the reference frequency;
S44: recording the identified frequency components and amplitude information thereof into a database, wherein each record comprises frequency values, amplitude values and time stamp information;
S45: the recorded spectral features are analyzed to identify typical patterns or anomalies of the current signal.
6. The automatic identification method for power distribution network topology based on data analysis according to claim 5, wherein the analyzing and recording the spectrum features in S45 specifically includes:
S451: the spectrum characteristic set of the current signal is defined as follows: f= { (F i,Ai) |i=1, 2, …, n }, where F i is the ith frequency component, a i is the corresponding amplitude value, n is the total number of frequency components, and by comparing the amplitude of each frequency component with its historical average amplitude or expected amplitude, to identify abnormal spectral features, the threshold for anomaly detection is set to τ, and the specific formula is: wherein/> The historical average amplitude value of the frequency component f i is the frequency component set identified as abnormal, and the threshold tau is determined according to actual conditions and empirical data and is used for controlling the sensitivity of abnormal detection;
S452: analyzing the relation between the abnormal frequency component E and the known typical mode, particularly by matching the known fault characteristic frequency or the harmonic mode, and marking the frequency component as abnormal of a corresponding type when f i is matched with the characteristic frequency in the known fault mode;
S453: finally, the identified abnormal situation and the typical pattern are recorded in a database.
7. The automatic identification method for the topology of the power distribution network based on data analysis according to claim 6, wherein the step S5 specifically comprises:
S51: constructing an input data set from the spectral features analyzed in step S4, each data point including a frequency component f i and its corresponding amplitude a i as a feature vector x= [ f 1,A1,f2,A2,…,fn,An ], where n is the total number of frequency components;
S52: designing a deep neural network DNN model, wherein the model structure comprises an input layer, a plurality of hidden layers and an output layer, the hidden layers use a ReLU activation function, the output layer uses a softmax function to carry out multi-classification, and the specific formula of the model is as follows:
Y＝softmax(Wm·ReLU(Wm-1·(…ReLU(W1·X+b1)…)+bm-1)+bm)，
Wherein, W 1,…,Wm and b 1,…,bm are respectively the weight and bias of the network layer, m is the number of hidden layers, Y is the prediction result of the output layer, and represents the probability distribution of different types of current signals;
s53: pre-training a DNN model by using historical current signal data and known categories, and adjusting model parameters through supervised learning to maximize the classification accuracy of the model on a training set;
s54: inputting the feature vector X constructed in the step S51 by using the trained DNN model, outputting a prediction result Y, wherein the result Y is the probability of each category to which the current signal belongs, and determining the category of the current signal according to the category with the largest probability;
s55: according to the classification result of step S54, each current signal feature is associated with a corresponding powered device or network path.
8. The automatic identification method for the topology of the power distribution network based on the data analysis of claim 7, wherein the step S6 specifically comprises:
S61: collecting DNN model identification results in the step S5, wherein the DNN model identification results comprise the category of the current signal and the associated information of the corresponding electric equipment or network path, and the associated information is used as basic data for constructing a topological graph;
S62: determining all key equipment and network nodes in a low-voltage transformer area, including transformers, power distribution cabinets, branch lines and important loads, and allocating unique identifiers to each equipment and node so as to be represented in a topological graph;
S63: determining the electrical connection relation between each device and the node according to the association of the current signal characteristic and the device or the path identified in the step S61, and particularly when a certain current signal characteristic is matched with the fault type on a specific branch line, indicating that the branch line is electrically connected with the device related to the fault;
s64: and constructing a topological graph by using a graphical representation method, wherein the topological graph comprises nodes and edges, and particularly, drawing the edges in the graph to connect the related nodes by utilizing the connection relation determined in the step S63 so as to form a complete low-voltage area electric topological graph.
9. The automatic identification method for the topology of the power distribution network based on the data analysis of claim 8, wherein the step S7 specifically comprises:
S71: step S6, a generated topological graph of the low-voltage transformer area is obtained and is used for indicating all electric equipment and connection relations in the power distribution transformer area;
s72: collecting real-time data, including current, voltage, power and equipment state information, specifically through an intelligent sensor and a data collection system;
S73: analyzing the collected real-time data, and determining the running state of each electrical device and the actual load condition of the network;
s74: updating the device state and the network load condition obtained by analysis into a low-voltage station area topological graph in the step S71, wherein the topological graph reflects the current physical state of the power distribution network;
s75: and displaying the updated low-voltage area topological graph through a visualization tool, and providing visual network structure and equipment state information for operation and maintenance personnel.
10. An automatic identification system for power distribution network topology based on data analysis, for implementing the automatic identification method for power distribution network topology based on data analysis as claimed in any one of claims 1 to 9, comprising:
And a data acquisition module: the method comprises the steps of acquiring real-time current signal data from key nodes of a power distribution network, and providing original input data for topology identification;
And a signal processing module: receiving the current signal data acquired by the data acquisition module, performing fast Fourier transform and window function processing on the current signal data, and converting the current signal from a time domain to a frequency domain;
and the feature extraction and classification module is used for: based on the frequency domain information obtained by the signal processing module, automatically classifying and identifying the current signals by utilizing a pre-trained deep neural network model;
Topology construction updating module: constructing and updating a topological graph of the low-voltage area according to the identification result of the feature extraction and classification module;
And a state synchronization module: the topology map of the low-voltage transformer area is used for keeping the topology map of the low-voltage transformer area generated by the topology construction and updating module synchronous with the actual physical equipment and network state of the power distribution network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410224655.0A CN118114019A (en) | 2024-02-29 | 2024-02-29 | Automatic identification method and system for power distribution network topology based on data analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410224655.0A CN118114019A (en) | 2024-02-29 | 2024-02-29 | Automatic identification method and system for power distribution network topology based on data analysis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118114019A true CN118114019A (en) | 2024-05-31 |
Family
ID=91217578
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410224655.0A Pending CN118114019A (en) | 2024-02-29 | 2024-02-29 | Automatic identification method and system for power distribution network topology based on data analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118114019A (en) |
-
2024
- 2024-02-29 CN CN202410224655.0A patent/CN118114019A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110956220B (en) | Non-invasive household appliance load identification method | |
| CN105629136B (en) | Cable insulation status online auto monitoring and diagnostic system | |
| CN111382789B (en) | Power load identification method and system based on machine learning | |
| CN111242391B (en) | Machine learning model training method and system for power load identification | |
| CN111861211A (en) | System with double-layer anti-electricity-stealing model | |
| CN111209934A (en) | Fan fault prediction and alarm method and system | |
| CN114580666A (en) | Multi-mode fusion motor intelligent maintenance system | |
| CN116800199A (en) | Distributed photovoltaic operation electric energy quality monitoring and analyzing method | |
| CN117668751B (en) | High-low voltage power system fault diagnosis method and device | |
| CN117169657B (en) | High-voltage cable state monitoring method and system based on artificial intelligence | |
| CN117110798B (en) | Fault detection method and system for intelligent power distribution network | |
| CN117277566B (en) | Power grid data analysis power dispatching system and method based on big data | |
| CN109635430A (en) | Grid power transmission route transient signal monitoring method and system | |
| Liu et al. | Hierarchical context-aware anomaly diagnosis in large-scale PV systems using SCADA data | |
| CN116881661A (en) | Performance automatic analysis method and system based on low-voltage power capacitor | |
| CN118114019A (en) | Automatic identification method and system for power distribution network topology based on data analysis | |
| US20230054387A1 (en) | System for monitoring and analyzing electric parameters | |
| CN114760194A (en) | Fault studying and judging method for Internet of things equipment | |
| Thiyagarajan et al. | Wavelet approach and support vector networks based power quality events recognition and categorisation | |
| CN117388594B (en) | Intelligent ammeter power signal decomposition method based on EMD signal processing technology | |
| US20230153389A1 (en) | Systems and methods for automatically identifying, analyzing and reducing extraneous waveform captures | |
| Saucedo-Dorantes et al. | Novelty Detection Methodology Based on Self-Organizing Maps for Power Quality Monitoring | |
| Majumdar et al. | Assessment of transient disturbance using discrete fourier transform and feed forward neural network based hybrid classifier | |
| LU503644B1 (en) | A voltage monitoring and early warning method and system for an offshore wind farm | |
| CN116298881B (en) | Electrical signal motor health early warning method based on channel attention multi-module LMMD |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination |