Low-voltage electric energy quality problem monitoring terminal and detection processing method

Abstract

The embodiment of the invention provides a low-voltage electric energy quality problem monitoring terminal and a detection processing method, and belongs to the technical field of electric energy quality monitoring. The monitoring terminal comprises a central processor, a memory, a data acquisition module, a data preprocessing module, a feature extraction module, a problem identification module and an equipment positioning module. The data preprocessing module is used for preprocessing the collected data; the feature extraction module is used for extracting multi-dimensional features from the preprocessed data; the problem identification module is used for classifying and identifying the power quality problem by using a deep learning model and outputting an identification result; according to the monitoring terminal and the detection processing method, efficient, accurate and automatic monitoring and identification of the electric energy quality problem are realized, and powerful technical support is provided for detection of the electric energy quality problem of an electric power system.

Description

Technical Field

[0001] The present invention relates to the technical field of power quality monitoring, and in particular to a low-voltage power quality problem monitoring terminal and a detection and processing method. Background Art

[0002] The power quality-related electrical equipment in the power system mainly includes: Flexible Transmission System (FACTS), High Voltage Direct Current (HVDC) and filter branches, electrified railways / rail transit locomotives, high voltage / low voltage power quality management equipment, energy storage, electric vehicles, wind power generation equipment, photovoltaic power generation equipment, nonlinear load equipment, linear load equipment and sensitive load equipment. These power quality-related electrical equipment require circuit breakers to access different levels and different types of busbars in the power system.

[0003] Among them, the electrical equipment related to low-voltage power quality problems mainly include nonlinear load equipment, energy storage equipment, electric vehicles, wind power generation equipment, photovoltaic power generation equipment, etc. The low-voltage power quality problem user monitoring terminal (hereinafter referred to as the monitoring terminal) mainly uses the monitoring terminal to bind with the corresponding low-voltage bus outgoing line circuit breaker, collects voltage from the grid-connected bus voltage, and collects current from the current transformer of the circuit breaker for analysis and judgment; currently, the monitoring of low-voltage power quality problems is mainly monitoring and recording, which does not have intelligent historical dimension analysis, so the existing analysis methods cannot meet the requirements of intelligent analysis.

[0004] The inventors of the present application discovered during the process of implementing the present invention that the above-mentioned solutions in the prior art have the defects of poor convenience and intelligence in monitoring low-voltage power quality problems. Summary of the invention

[0005] The purpose of the embodiments of the present invention is to provide a low-voltage power quality problem monitoring terminal and a detection and processing method, which have the function of improving the convenience and intelligence of low-voltage power quality problem monitoring.

[0006] In order to achieve the above-mentioned purpose, an embodiment of the present invention provides a low-voltage power quality problem monitoring terminal on the one hand, including a central processing unit, a memory, a data acquisition module, a data preprocessing module, a feature extraction module, a problem identification module, and a device positioning module; The data acquisition module is used to collect voltage and current data of the low-voltage bus outlet circuit breaker; The data preprocessing module is used to preprocess the collected data; The feature extraction module is used to extract multi-dimensional features from the preprocessed data; The problem identification module is used to classify and identify power quality problems using a deep learning model and output identification results; The device positioning module is used to locate the electrical equipment causing the power quality problem according to the identification result; The central processing unit is used to coordinate the work of each module and perform computing tasks; The memory is used to store collected data, processing results and model parameters of the deep learning model.

[0007] On the other hand, the present invention also provides a low-voltage power quality problem monitoring terminal detection and processing method, comprising: collecting voltage and current data of a low-voltage bus outlet circuit breaker; Preprocess the collected voltage and current data, including filtering, normalization, and time series alignment; Using multi-dimensional feature extraction processing methods, time domain, frequency domain and time-frequency domain features are extracted from the preprocessed data; Based on the extracted features, a deep learning model is used to classify and identify power quality problems and output the identification results.

[0008] Optionally, preprocessing the collected voltage and current data further includes: Use wavelet transform to denoise the collected data; The data were normalized using amplitude normalization and time normalization methods; Perform time series alignment on voltage and current data.

[0009] Optionally, the multi-dimensional feature extraction processing method includes: Extracting statistical characteristics of the mean, variance, peak factor and waveform distortion factor of the voltage data to be detected or the current data to be detected of the time domain signal in the time domain; The time domain signal is converted into the frequency domain using fast Fourier transform, and then the harmonic content, total harmonic distortion rate and fundamental power factor of the voltage data to be detected or the current data to be detected are extracted; The transient features are extracted using HHT transformation in the time-frequency domain; the extracted transient features are formed into a data set and sent to a central server.

[0010] Optionally, a deep learning model is used to classify and identify power quality issues, and the output identification results include: A convolutional neural network is used to extract the transient features of energy, peak value and duration of each IMF; Constructing a convolutional neural network layer, a long short-term memory network layer, and a multi-task output layer; the convolutional neural network layer includes two convolutional layers and a maximum pooling layer; the convolutional layers are respectively convolutional layer 1 and convolutional layer 2; The output of the convolutional neural network layer is input into the long short-term memory network layer, and then the output of the long short-term memory network layer uses the self-attention mechanism and outputs the features of different scales; The multi-task output layer sets multiple output branches to simultaneously identify three power quality problems: voltage sag, harmonic pollution and three-phase imbalance.

[0011] Optionally, constructing a convolutional neural network layer, a long short-term memory network layer, and a multi-task output layer includes: The convolutional neural network layer is used to receive the preprocessed voltage and current data, and the data shape is (batch_size, 1000, 6), where 6 indicates that each time point contains the transient characteristics of the energy, peak value and duration of the IMF of the voltage and the transient characteristics of the energy, peak value and duration of the IMF of the current; Convolutional layer 1, with 32 3x1 convolution kernels and a stride of 1; Convolutional layer 2, with 64 3x1 convolution kernels and a stride of 1; Max pooling layer with a 2x1 pooling window; A long short-term memory network layer, used to receive the output of the convolutional neural network layer, wherein the long short-term memory network layer includes 128 LSTM units and the number of time steps is 250; A self-attention mechanism applied to the output of the LSTM layer, with 8 attention heads, each of dimension 16; Multi-task output layer for identifying different power quality issues including: The multi-task output layer is used to perform voltage sag identification in Task 1, and outputs two categories through the fully connected layer: sag / no sag. The multi-task output layer is used to perform task 2, i.e., identification of harmonic pollution levels. It outputs three categories, namely, mild / moderate / severe, through the fully connected layer. The multi-task output layer is used to perform the three-phase imbalance degree identification of Task 3, and outputs three categories through the fully connected layer: normal / mild / severe.

[0012] Optionally, after classifying and identifying the power quality problems using a deep learning model based on the extracted features and outputting the identification results, it also includes determining the equipment causing the power quality problems through association analysis.

[0013] Optionally, determining the device causing the power quality problem through correlation analysis includes the following steps: Establish a correlation model between power quality issues and the operating status of power equipment; Use correlation models to locate devices that cause power quality issues.

[0014] Optionally, the correlation model uses the following formula group to calculate the comprehensive power quality index EQI: EQI=w1*THD+w2*VUF+w3*PF+w4*VF; Among them, THD is the total harmonic distortion, and the calculation formula is: , h is the 2nd to 40th harmonic; VUF is the voltage unbalance, and the calculation formula is: VUF=|V-| / |V+|*100%; Among them, V- is the negative sequence voltage component, and V+ is the positive sequence voltage component; PF is the power factor, and the calculation formula is: ; Where P is active power and Q is reactive power; VF is the voltage fluctuation coefficient, and the calculation formula is: ; Among them, Vmax is the maximum voltage, Vmin is the minimum voltage, Vn is the rated voltage, w1, w2, w3, w4 are weight coefficients, and w1+w2+w3+w4=1 is satisfied.

[0015] Through the above technical solution, the present invention provides a low-voltage power quality problem monitoring terminal and detection and processing method. By timely discovering and identifying power quality problems, the system can effectively protect power equipment and avoid equipment damage or failure caused by power quality problems. At the same time, through the feedback of monitoring and identification results, the power system can be optimized to improve the overall power quality.

[0016] The above-mentioned low-voltage power quality problem monitoring terminal and its detection and processing method, through data acquisition, preprocessing, feature extraction and deep learning technology, realize efficient, accurate and automatic monitoring and identification of power quality problems, providing strong technical support for the stable operation of the power system and the optimization of power quality.

[0017] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings: Figure 1 is a schematic diagram of a low voltage power quality problem monitoring terminal according to an embodiment of the present invention; Figure 2 is a flow chart of a low voltage power quality problem monitoring terminal detection and processing method according to an embodiment of the present invention; Figure 3is a flow chart of data preprocessing in a low-voltage power quality problem monitoring terminal detection and processing method according to an embodiment of the present invention; Figure 4 It is a flowchart of multi-dimensional feature extraction in a low-voltage power quality problem monitoring terminal detection and processing method according to an embodiment of the present invention.

[0019] Description of Reference Numerals DETAILED DESCRIPTION

[0020] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0021] Figure 1 is a schematic diagram of a low voltage power quality problem monitoring terminal according to an embodiment of the present invention, such as Figure 1 As shown, the present invention provides a low-voltage power quality problem monitoring terminal, including a central processing unit 10, a memory 20, a data acquisition module 30, a data preprocessing module 40, a feature extraction module 50, a problem identification module 60, and a device positioning module 70; Data acquisition module, used to collect voltage and current data of low-voltage busbar outgoing circuit breaker; A data preprocessing module, used for preprocessing the collected data; Feature extraction module, used to extract multi-dimensional features from preprocessed data; The problem identification module is used to classify and identify power quality problems using a deep learning model and output identification results; An equipment positioning module is used to locate the electrical equipment that causes the power quality problem based on the identification results; The central processing unit is used to coordinate the work of each module and perform computing tasks; Memory, used to store collected data, processing results, and model parameters of deep learning models.

[0022] The system includes a central processing unit 10, a memory 20, a data acquisition module 30, a data preprocessing module 40, a feature extraction module 50, a problem identification module 60, and an equipment positioning module 70; wherein the data acquisition module is used to collect voltage and current data of the low-voltage busbar outgoing circuit breaker; the data preprocessing module preprocesses the collected voltage and current data, including filtering, normalization and time series alignment; the feature extraction module uses a multi-dimensional feature extraction processing method to extract time domain, frequency domain and time-frequency domain features from the preprocessed data; the problem identification module uses a deep learning model based on the extracted features to classify and identify power quality problems and output the identification results.

[0023] The low-voltage power quality problem monitoring terminal and detection and processing method provided by the present invention can effectively protect power equipment and avoid equipment damage or failure caused by power quality problems by timely discovering and identifying power quality problems. At the same time, through the feedback of monitoring and identification results, the power system can be optimized to improve the overall power quality.

[0024] The above-mentioned low-voltage power quality problem monitoring terminal and its detection and processing method, through data acquisition, preprocessing, feature extraction and deep learning technology, realize efficient, accurate and automatic monitoring and identification of power quality problems, providing strong technical support for the stable operation of the power system and the optimization of power quality.

[0025] like Figure 2 As shown, the present invention provides a low-voltage power quality problem monitoring terminal detection and processing method, which uses the low-voltage power quality problem monitoring terminal provided above to implement detection and processing, including the following steps: S1. Collect the voltage and current data of the low-voltage busbar outgoing circuit breaker: By collecting the voltage and current data of the low-voltage busbar outgoing circuit breaker, the comprehensiveness of monitoring is ensured. Voltage and current are the most basic physical quantities of the power system, and their monitoring is the basis for identifying power quality problems.

[0026] S2. Preprocess the collected voltage and current data, including filtering, normalization and time series alignment; Preprocess the collected data, including filtering, normalization and time series alignment, which can effectively remove noise and interference and improve the accuracy and reliability of the data. Filtering can reduce high-frequency noise, normalization helps standardize data, and time series alignment ensures the consistency of data in time.

[0027] S3. Use multi-dimensional feature extraction and processing methods to extract time domain, frequency domain and time-frequency domain features from pre-processed data; Use multi-dimensional feature extraction and processing methods to extract time domain, frequency domain and time-frequency domain features from pre-processed data. This multi-dimensional feature extraction can more comprehensively reflect the state of power quality and provide rich information for subsequent classification and identification.

[0028] S4. Based on the extracted features, a deep learning model is used to classify and identify power quality problems, and the identification results are output. Based on the extracted features, a deep learning model is used to classify and identify power quality problems. Due to its powerful nonlinear fitting ability, the deep learning model can effectively handle complex data relationships and improve the accuracy and efficiency of identification. Analysis shows that the output identification results directly reflect the current state of power quality and can provide a reliable reference for the maintenance and management of the power system. This real-time monitoring and identification helps to promptly discover and deal with power quality problems and ensure the stable operation of the power system.

[0029] like Figure 3 As shown, the preprocessing in step S2 also includes: S21, denoising the collected data using wavelet transform; S22, using amplitude normalization and time normalization methods to normalize the data; S23. Perform time series alignment on the voltage and current data.

[0030] like Figure 4 As shown, the multi-dimensional feature extraction processing method in step S3 includes: S31, extracting statistical characteristics of the mean, variance, peak factor and waveform distortion factor of the voltage data to be detected or the current data to be detected of the time domain signal in the time domain; S32, using fast Fourier transform to convert the time domain signal into the frequency domain, and then extracting the harmonic content, total harmonic distortion rate and fundamental power factor of the voltage data to be detected or the current data to be detected; S33. Extract transient features using HHT transformation in the time-frequency domain; form a data set of the extracted transient features and send it to a central server.

[0031] The multi-dimensional feature extraction processing method of step S3 is described in detail, wherein the time domain feature extraction: Time domain features mainly reflect the statistical characteristics of the signal in the time dimension. Specifically include: (a) Mean: reflects the average level of the signal; μ= (1 / N)*Σ(xi), i=1toN; Among them, μ is the mean, N is the number of sampling points, and xi is the value of the i-th sampling point; (b) Variance: reflects the degree of signal fluctuation σ²= (1 / N)*Σ((xi-μ)²), i=1toN; Here, σ² is the variance.

[0032] (c) Crest factor: reflects the peak and trough characteristics of the signal; Cf = |xpeak| / xrms; Where Cf is the crest factor, xpeak is the signal peak value, and xrms is the root mean square value; (d) Waveform distortion factor: reflects the degree of deviation of the signal from the standard sine wave; Fd = xrms / |x|avg; Where, Fd is the waveform distortion factor, |x|avg is the average value of the absolute value of the signal; Specific examples: Assume that the collected voltage data is [220, 225, 218, 230, 222, 219, 228, 221]V; Mean μ = 222.875 V; Variance σ² ≈ 16.984V²; Crest factor Cf ≈ 1.414; Waveform distortion factor Fd ≈ 1.108; Frequency domain feature extraction: The time domain signal is converted to the frequency domain using Fast Fourier Transform (FFT), and the following features are extracted: (a) Harmonic content: the amplitude of each harmonic; Vh = |FFT(x)[h]|, h = 1, 2, 3, ...; Where h is the harmonic order; (b) Total harmonic distortion (THD): reflects the degree of harmonic pollution of the signal; THD = √(Σ(Vh²) / V1²), h = 2 to 40; Where V1 is the fundamental component; (c) Fundamental power factor: reflects the ratio of fundamental active power to apparent power; PF1 = P1 / S1; P1 is the fundamental wave active power, S1 is the fundamental wave apparent power; Specific examples: Assume that after FFT, the fundamental amplitude V1 = 220V, the 3rd harmonic V3 = 6.6V, and the 5th harmonic V5 = 4.4V; THD ≈ 3.56%; If the fundamental wave active power P1 = 1000W, the fundamental wave apparent power S1 = 1100VA, then PF1 ≈ 0.909; Time-frequency domain feature extraction: The signal is decomposed and analyzed using the Hilbert-Huang transform (HHT) to extract the following features: (a) Transient characteristics: Capturing the transient changes of the signal through intrinsic mode function (IMF) decomposition; IMFi = HHT(x)[i], i = 1, 2, 3, ...; The features of each IMF such as energy, peak value and duration are extracted.

[0033] Specific examples: Assume that the voltage signal is decomposed by HHT to obtain three IMFs: IMF1: high frequency component, energy E1 = 10, peak value P1 = 5V, duration T1 = 0.01s; IMF2: intermediate frequency component, energy E2=30, peak value P2= 10V, duration T2 = 0.05s; IMF3: low frequency component, energy E3 = 60, peak value P3 = 15V, duration T3 = 0.1s; These multi-dimensional feature extraction and processing methods can fully capture the various characteristics of power quality signals and provide rich input information for subsequent problem identification and classification. Time domain features reflect the overall statistical characteristics of the signal, frequency domain features reveal the harmonic composition of the signal, and time-frequency domain features can capture the transient changes and non-stationary characteristics of the signal. By comprehensively utilizing these features, various power quality problems can be more accurately identified and classified.

[0034] Preferably, as an implementable solution, the deep learning model in step S4 is a multi-task learning network, including: A convolutional neural network is used to extract the transient features of energy, peak value and duration of each IMF; Construct a convolutional neural network (CNN) layer, a long short-term memory network (LSTM) layer, and a multi-task output layer; the convolutional neural network (CNN) layer includes two convolutional layers and one maximum pooling layer; the convolutional layers are convolutional layer 1 and convolutional layer 2; The output of the convolutional neural network (CNN) layer is input into the long short-term memory network (LSTM) layer, and then the output of the long short-term memory network (LSTM) layer uses the self-attention mechanism and outputs the fusion of features of different scales; The multi-task output layer sets multiple output branches to simultaneously identify three power quality problems: voltage sag, harmonic pollution and three-phase imbalance.

[0035] Preferably, as an implementable solution, a convolutional neural network (CNN) layer, a long short-term memory network (LSTM) layer, and a multi-task output layer are constructed, including: A convolutional neural network (CNN) layer is used to receive the preprocessed voltage and current data, and the data shape is (batch_size, 1000, 6), where 6 means that each time point contains the transient characteristics of the energy, peak value and duration of the IMF of the voltage and the transient characteristics of the energy, peak value and duration of the IMF of the current; Convolutional layer 1, with 32 3x1 convolution kernels and a stride of 1; Convolutional layer 2, with 64 3x1 convolution kernels and a stride of 1; Max pooling layer with a 2x1 pooling window; The LSTM layer receives the output of the CNN layer. The LSTM layer contains 128 LSTM units and has a time step of 250. A self-attention mechanism applied to the output of the LSTM layer with 8 attention heads, each of dimension 16; Multi-task output layer for identifying different power quality issues including: The multi-task output layer is used to perform voltage sag identification in Task 1 and outputs two categories (with / without voltage sag) through the fully connected layer. The multi-task output layer is used to perform task 2, i.e., identification of harmonic pollution levels, and outputs three categories (mild / moderate / severe) through the fully connected layer; The multi-task output layer is used to perform the three-phase imbalance degree identification of Task 3, and outputs three categories (normal / mild / severe) through the fully connected layer.

[0036] Construct a convolutional neural network (CNN) layer: Input: preprocessed voltage and current data, shape is (batch_size, 1000, 6), representing 1000 time points, each time point contains the transient features of the energy, peak value and duration of the IMF of the voltage and the transient features of the energy, peak value and duration of the IMF of the current.

[0037] Convolutional layer 1: 32 3x1 convolution kernels with a stride of 1; Convolutional layer 2: 64 3x1 convolution kernels with a stride of 1; Max pooling layer: 2x1 pooling window; (b) Constructing the Long Short-Term Memory (LSTM) layer: Input: output of CNN layer; number of LSTM units: 128; Number of time steps: 250 (sequence length after pooling); (c) Attention mechanism: self-attention mechanism is applied to the output of the long short-term memory network (LSTM) layer; number of attention heads: 8; dimension of each head: 16; (d) Multi-task output layer: Task 1 (voltage sag identification): fully connected layer, output 2 categories (with / without sag); Task 2 (harmonic pollution degree): fully connected layer, output 3 categories (mild / moderate / severe); Task 3 (three-phase imbalance degree): fully connected layer, output 3 categories (normal / mild / severe); Case results: Analyze the data for a certain period of time, and the model output is: Voltage sag: A voltage sag event was detected with a confidence level of 0.92 Harmonic pollution: moderate pollution, confidence level 0.85 Three-phase imbalance: slight imbalance, confidence level 0.78.

[0038] Preferably, as an implementable embodiment, the method further comprises the following steps: (a) Establish a correlation model between power quality issues and the operating status of power equipment; (b) Use the correlation model to locate the equipment that causes power quality problems.

[0039] Preferably, as an implementable solution, the correlation model uses the following formula group to calculate the comprehensive power quality index EQI: EQI=w1*THD+w2*VUF+w3*PF+w4*VF; Among them, THD is the total harmonic distortion, and the calculation formula is: , h is the 2nd to 40th harmonic; VUF is the voltage unbalance, and the calculation formula is: VUF=|V-| / |V+|*100%; Among them, V- is the negative sequence voltage component, and V+ is the positive sequence voltage component; PF is the power factor, and the calculation formula is: ; Where P is active power and Q is reactive power; VF is the voltage fluctuation coefficient, and the calculation formula is: ; Among them, Vmax is the maximum voltage, Vmin is the minimum voltage, Vn is the rated voltage, w1, w2, w3, w4 are weight coefficients, and w1+w2+w3+w4=1 is satisfied.

[0040] In summary, the low-voltage power quality problem monitoring terminal and detection and processing method adopted in the embodiment of the present invention realizes efficient, accurate and automatic monitoring and identification of power quality problems through data acquisition, preprocessing, feature extraction and deep learning technology, and provides strong technical support for the stable operation of the power system and the optimization of power quality.

[0041] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0042] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0043] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0044] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0045] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0046] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0047] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0048] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0049] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A low voltage power quality problem monitoring terminal, characterized in that: It includes a central processing unit, a memory, a data acquisition module, a data preprocessing module, a feature extraction module, a problem identification module, and a device positioning module; The data acquisition module is used to collect voltage and current data of the low-voltage bus outlet circuit breaker; The data preprocessing module is used to preprocess the collected data; The feature extraction module is used to extract multi-dimensional features from the preprocessed data; The problem identification module is used to classify and identify power quality problems using a deep learning model and output identification results; The device positioning module is used to locate the electrical equipment causing the power quality problem according to the identification result; The central processing unit is used to coordinate the work of each module and perform computing tasks; The memory is used to store collected data, processing results and model parameters of the deep learning model.

2. A low-voltage power quality problem monitoring terminal detection and processing method applied to a low-voltage power quality problem monitoring terminal as claimed in claim 1, characterized in that: include: Collect voltage and current data of low-voltage busbar outgoing circuit breaker; Preprocess the collected voltage and current data, including filtering, normalization, and time series alignment; Using multi-dimensional feature extraction processing methods, time domain, frequency domain and time-frequency domain features are extracted from the preprocessed data; Based on the extracted features, a deep learning model is used to classify and identify power quality problems and output the identification results.

3. The low voltage power quality problem monitoring terminal detection and processing method according to claim 2 is characterized in that: Preprocessing the collected voltage and current data also includes: Use wavelet transform to denoise the collected data; The data were normalized using amplitude normalization and time normalization methods; Perform time series alignment on voltage and current data.

4. The low-voltage power quality problem monitoring terminal detection and processing method according to claim 3 is characterized in that: The multi-dimensional feature extraction processing method includes: Extracting statistical characteristics of the mean, variance, peak factor and waveform distortion factor of the voltage data to be detected or the current data to be detected of the time domain signal in the time domain; The time domain signal is converted into the frequency domain using fast Fourier transform, and then the harmonic content, total harmonic distortion rate and fundamental power factor of the voltage data to be detected or the current data to be detected are extracted; The transient features are extracted using HHT transformation in the time-frequency domain; the extracted transient features are formed into a data set and sent to a central server.

5. The low-voltage power quality problem monitoring terminal detection and processing method according to claim 4 is characterized in that: Use deep learning models to classify and identify power quality issues. The output identification results include: A convolutional neural network is used to extract the transient features of energy, peak value and duration of each IMF; Constructing a convolutional neural network layer, a long short-term memory network layer, and a multi-task output layer; the convolutional neural network layer includes two convolutional layers and a maximum pooling layer; the convolutional layers are respectively convolutional layer 1 and convolutional layer 2; The output of the convolutional neural network layer is input into the long short-term memory network layer, and then the output of the long short-term memory network layer uses the self-attention mechanism and outputs the features of different scales; The multi-task output layer sets multiple output branches to simultaneously identify three power quality problems: voltage sag, harmonic pollution and three-phase imbalance.

6. The low-voltage power quality problem monitoring terminal detection and processing method according to claim 5 is characterized in that: Constructing the convolutional neural network layer, the long short-term memory network layer, and the multi-task output layer includes: The convolutional neural network layer is used to receive the preprocessed voltage and current data, and the data shape is (batch_size, 1000, 6), where 6 indicates that each time point contains the transient characteristics of the energy, peak value and duration of the IMF of the voltage and the transient characteristics of the energy, peak value and duration of the IMF of the current; Convolutional layer 1, with 32 3x1 convolution kernels and a stride of 1; Convolutional layer 2, with 64 3x1 convolution kernels and a stride of 1; Max pooling layer with a 2x1 pooling window; A long short-term memory network layer, used to receive the output of the convolutional neural network layer, wherein the long short-term memory network layer includes 128 LSTM units and the number of time steps is 250; A self-attention mechanism applied to the output of the LSTM layer, with 8 attention heads, each of dimension 16; Multi-task output layer for identifying different power quality issues including: The multi-task output layer is used to perform voltage sag identification in Task 1, and outputs two categories through the fully connected layer: sag / no sag. The multi-task output layer is used to perform task 2, i.e., identification of harmonic pollution levels. It outputs three categories, namely, mild / moderate / severe, through the fully connected layer. The multi-task output layer is used to perform the three-phase imbalance degree identification of Task 3, and outputs three categories through the fully connected layer: normal / mild / severe.

7. The low voltage power quality problem monitoring terminal detection and processing method according to claim 6 is characterized in that: Based on the extracted features, a deep learning model is used to classify and identify power quality problems. After the identification results are output, the device that causes the power quality problem is determined through association analysis.

8. The low-voltage power quality problem monitoring terminal detection and processing method according to claim 7 is characterized in that: Determine the device causing the power quality problem through correlation analysis, including the following steps: Establish a correlation model between power quality issues and the operating status of power equipment; Use correlation models to locate devices that cause power quality issues.

9. The low-voltage power quality problem monitoring terminal detection and processing method according to claim 8 is characterized in that: The correlation model uses the following formula group to calculate the comprehensive power quality index EQI: EQI=w1*THD+w2*VUF+w3*PF+w4*VF; Among them, THD is the total harmonic distortion, and the calculation formula is: , h is the 2nd to 40th harmonic; VUF is the voltage unbalance, and the calculation formula is: VUF=|V-| / |V+|*100%; Among them, V- is the negative sequence voltage component, and V+ is the positive sequence voltage component; PF is the power factor, and the calculation formula is: ; Where P is active power and Q is reactive power; VF is the voltage fluctuation coefficient, and the calculation formula is: ; Among them, Vmax is the maximum voltage, Vmin is the minimum voltage, Vn is the rated voltage, w1, w2, w3, w4 are weight coefficients, and w1+w2+w3+w4=1 is satisfied.

Images