A photovoltaic power grid fault identification method and system based on circuit analysis

Abstract

The application discloses a photovoltaic power grid fault identification method and system based on circuit analysis, relates to the technical field of fault identification, and comprises the following steps: obtaining operation parameters of a photovoltaic power grid, and constructing a circuit analysis model; based on the circuit analysis model, equivalent response curves under different fault scenarios are extracted, and a difference residual sequence is generated by comparing a benchmark operation state; the difference residual sequence is subjected to time-frequency joint decomposition, photovoltaic output fluctuation and load disturbance components are stripped, and pure circuit characteristic components are obtained; based on the pure circuit characteristic components, a multi-dimensional feature coordinate space is formed, a dynamic bending rate of a fault response track is used to distinguish fault types; and the fault type distinguishing result is mapped back to the circuit analysis model, and a fault branch position is located in combination with local disturbance distribution of a node impedance matrix. The application realizes accurate discrimination of complex fault types and fault branch positioning by combining pure circuit characteristic component extraction and multi-dimensional feature space dynamic analysis.

Description

Technical Field

[0001] This invention relates to the field of fault identification technology, and more specifically, to a method and system for identifying photovoltaic power grid faults based on circuit analysis. Background Technology

[0002] With the large-scale integration of distributed photovoltaic (PV) power into distribution networks, rapid fault identification and location in PV grids has become a critical issue for ensuring the safe and stable operation of the power grid. Existing PV grid fault identification methods are mostly based on circuit analysis techniques, using indicators such as node impedance, equivalent current surges, or voltage anomalies to determine the fault type and location. However, due to the characteristics of PV grids—highly random power output, numerous load disturbances, and widely distributed grid-connected nodes—the differences in circuit indicator characteristics between complex faults such as single-phase grounding, phase-to-phase short circuits, and high-resistance grounding are not significant, easily leading to misjudgments or omissions using traditional methods. Furthermore, existing methods typically cannot effectively isolate the interference of PV power output fluctuations and load disturbances on fault signals, limiting the accuracy of fault feature extraction and making it difficult to meet the requirements for rapid and accurate fault identification.

[0003] The above-disclosed technical solutions have at least the following technical problems: circuit analysis methods rely heavily on indicators such as impedance and equivalent current change, but the characteristic differences between complex faults such as single-phase grounding, phase-to-phase short circuit and high-resistance grounding are not significant, which can easily lead to misjudgment or omission.

[0004] To address the above problems, this invention proposes a solution. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a photovoltaic grid fault identification method and system based on circuit analysis. By introducing a photovoltaic output disturbance factor correction circuit analytical model and combining pure circuit feature component extraction and multi-dimensional feature space dynamic analysis, the method achieves accurate identification of complex fault types and fault branch location, thereby solving the problem of misjudgment or missed judgment in the prior art due to photovoltaic power output fluctuations, load disturbances and insignificant differences in fault characteristics.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] On the one hand, a photovoltaic power grid fault identification method based on circuit analysis includes the following steps: obtaining the operating parameters of the photovoltaic power grid and constructing a circuit analytical model with photovoltaic output disturbance factors; based on the circuit analytical model, extracting equivalent response curves under different fault scenarios and generating differentiated residual sequences by comparing with the baseline operating state; performing time-frequency joint decomposition on the differentiated residual sequences to separate photovoltaic output fluctuations and load disturbance components, obtaining pure circuit feature components; based on the pure circuit feature components, forming a multi-dimensional feature coordinate space, and using the dynamic curvature of the fault response trajectory to determine the fault type; mapping the fault type determination result back to the circuit analytical model, and locating the fault branch location by combining the local disturbance distribution of the node impedance matrix.

[0008] In a preferred embodiment, the step of obtaining the operating parameters of the photovoltaic power grid and constructing a circuit analytical model with photovoltaic output disturbance factors specifically involves: obtaining the voltage, current, and line impedance parameters of each node in the photovoltaic power grid; establishing a corresponding initial equivalent circuit analytical model based on the electrical topology as a static reference model; introducing time series data of photovoltaic power into the static reference model; analyzing its power fluctuation amplitude, rate of change, and randomness index through a sliding time window to obtain a set of disturbance features used to characterize the uncertainty of photovoltaic output; generating dynamic correction parameters that characterize the influence of photovoltaic output randomness on node voltage, current, and line impedance based on the set of disturbance features; and embedding the correction parameters into the static reference model to obtain a circuit analytical model with photovoltaic output disturbance characteristics.

[0009] In a preferred embodiment, the step of extracting equivalent response curves under different fault scenarios based on the circuit analytical model and generating a differentiated residual sequence by comparing with the baseline operating state specifically involves: applying disturbance conditions of single-phase grounding, phase-to-phase short circuit, and high-resistance grounding under different fault conditions to the circuit analytical model to obtain dynamic change data of voltage and current at each node; solving the nonlinear coupling relationship between the node equivalent impedance and the injected current based on the dynamic change data to generate node response functions under different fault scenarios; plotting the curves of the evolution of node voltage and current residuals over time based on the node response functions and comparing them with the baseline curve under fault-free conditions to obtain a set of differentiated equivalent response curves; and calculating the differential residual sequence between the baseline curve and the fault curve based on the set of equivalent response curves and marking the residual characteristics under different fault conditions.

[0010] In a preferred embodiment, the step of solving the nonlinear coupling relationship between the node equivalent impedance and the injected current based on dynamically changing data to generate node response functions under different fault scenarios specifically involves: constructing a node equivalent impedance matrix under fault disturbance based on the dynamically changing data, combined with line impedance parameters and node voltage and current measurements; establishing a nonlinear coupling expression between voltage response and current input based on the node equivalent impedance matrix and node injected current; and numerically solving the nonlinear coupling expression using a recursive iterative method to obtain the node response functions under different fault scenarios.

[0011] In a preferred embodiment, the step of performing time-frequency joint decomposition on the differentiated residual sequence to separate photovoltaic power output fluctuations and load disturbance components and obtain pure circuit characteristic components specifically involves: constructing dual constraints for time-frequency decomposition based on the photovoltaic power output prediction curve and load inertia characteristics; performing wavelet packet decomposition on the differentiated residual sequence and selecting the optimal sub-band for reconstruction based on the residual energy concentration to obtain reconstructed components; mapping the reconstructed components to a time-frequency energy map and using a sparse curvature tracking method to identify high curvature abrupt change points as fault signals; correcting the decomposition boundary through an adversarial reconstruction mechanism, and finally separating photovoltaic power output fluctuations and load disturbance components to obtain pure circuit characteristic components.

[0012] In a preferred embodiment, the step of correcting the decomposition boundary through the adversarial reconstruction mechanism specifically involves: iteratively training a discriminative model, comparing the differences between the stripped components and the real fault samples, and adjusting the wavelet packet decomposition level and sub-band selection if the deviation is too large, until the separation effect meets the accuracy threshold.

[0013] In a preferred embodiment, the step of performing wavelet packet decomposition on the differentiated residual sequence and selecting the optimal sub-band for reconstruction based on the residual energy concentration specifically involves: performing wavelet packet decomposition on the differentiated residual sequence at several levels to obtain sub-signals in different frequency bands; calculating the energy of each sub-signal to form an energy distribution set, and calculating the energy proportion of each sub-band based on the energy distribution set to obtain candidate sub-bands; calculating the residual mutation rate index and frequency band coherence index of each candidate sub-band, and performing weighted fusion after normalization of the two to obtain a sub-band comprehensive weight coefficient; applying the sub-band comprehensive weight coefficient to the candidate sub-band signal and applying an attenuation factor to the non-target sub-band signal during inverse wavelet packet reconstruction to obtain the reconstructed signal; comparing the correlation between the reconstructed signal and the original residual sequence, and if the correlation coefficient is lower than a preset threshold, adjusting the comprehensive weight coefficient and repeating the reconstruction until the correlation reaches the threshold to obtain the final reconstructed component.

[0014] In a preferred embodiment, the step of forming a multi-dimensional feature coordinate space based on the pure circuit feature components and using the dynamic curvature of the fault response trajectory to determine the fault type specifically involves: combining the amplitude, phase, and energy characteristics of the pure circuit feature components at different time points or frequency points to form a multi-dimensional feature vector for each node; mapping the multi-dimensional feature vector to the multi-dimensional feature coordinate space and connecting the feature vectors under the same fault scenario in chronological order to form a fault response trajectory; calculating the local curvature of continuous points on each fault response trajectory and performing statistical analysis to form classification rules; and matching the curvature of the fault response trajectory to be tested with the classification rules to determine the fault type.

[0015] In a preferred embodiment, mapping the fault type discrimination result back to the circuit analytical model and locating the fault branch location by combining the local perturbation distribution of the node impedance matrix specifically involves: determining the set of nodes that may be faulty based on the fault type discrimination result and marking them in the circuit analytical model; calculating the local impedance perturbation index of each marked node under the fault scenario based on the connection relationship of the marked nodes in the circuit analytical model and the node impedance matrix; combining the node local impedance perturbation index with the fault type information of the corresponding node to generate a node fault sensitivity score; and statistically analyzing the fault sensitivity scores of the nodes at both ends of each branch in the circuit, identifying the branch with the highest score as the faulty branch.

[0016] On the other hand, a photovoltaic grid fault identification system based on circuit analysis includes the following modules:

[0017] The circuit analysis model construction module is used to obtain the operating parameters of the photovoltaic power grid and construct a circuit analysis model with photovoltaic output disturbance factors. The differential residual sequence generation module is used to extract equivalent response curves under different fault scenarios based on the circuit analysis model and generate differential residual sequences by comparing them with the baseline operating state. The feature decomposition module is used to perform time-frequency joint decomposition on the differential residual sequence, separating the photovoltaic output fluctuation and load disturbance components to obtain pure circuit feature components. The fault discrimination module is used to form a multi-dimensional feature coordinate space based on the pure circuit feature components and use the dynamic curvature of the fault response trajectory to distinguish the fault type. The fault location module is used to map the fault type discrimination result back to the circuit analysis model and locate the fault branch location by combining the local disturbance distribution of the node impedance matrix.

[0018] The technical effects and advantages of the photovoltaic power grid fault identification method and system based on circuit analysis of this invention are as follows:

[0019] 1. This invention constructs a circuit analytical model with photovoltaic output disturbance factors, and combines time-frequency joint decomposition and inverse wavelet packet reconstruction to remove photovoltaic power output fluctuations and load disturbance components, thereby extracting pure circuit characteristic components. This enables accurate reflection of the dynamic response characteristics of photovoltaic grid faults and improves the identification accuracy of fault signals.

[0020] 2. This invention uses the dynamic curvature of the fault response trajectory in a multi-dimensional feature coordinate space to determine the fault type, and combines the node impedance matrix to calculate local disturbances to generate node fault sensitivity scores, thereby achieving accurate location of fault branches and effectively distinguishing different fault types such as single-phase grounding, phase-to-phase short circuit, and high-resistance grounding, significantly improving the reliability and real-time performance of fault identification and location. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating a photovoltaic grid fault identification method based on circuit analysis according to the present invention.

[0022] Figure 2 This is a schematic diagram of the structure of a photovoltaic power grid fault identification system based on circuit analysis according to the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0024] Example 1, Figure 1 The present invention provides a photovoltaic power grid fault identification method based on circuit analysis, comprising the following steps:

[0025] S1. Obtain the operating parameters of the photovoltaic power grid and construct a circuit analytical model with photovoltaic output disturbance factors;

[0026] The operating parameters include the voltage, current, and line impedance parameters of each node in the photovoltaic power grid.

[0027] The photovoltaic output disturbance factor characterizes the dynamic correction effect of the randomness of photovoltaic output on circuit parameters. It refers to a dynamic parameter that quantifies the photovoltaic power output in real time based on the fluctuations of light and temperature, and is used to correct the circuit model.

[0028] In this embodiment, the step of obtaining the operating parameters of the photovoltaic power grid and constructing a circuit analytical model with photovoltaic output disturbance factors specifically involves:

[0029] The voltage, current and line impedance parameters of each node in the photovoltaic power grid are obtained, and the corresponding initial equivalent circuit analytical model is established based on the electrical topology relationship as a static reference model.

[0030] By introducing time series data of photovoltaic power into the static benchmark model, and analyzing its power fluctuation amplitude, rate of change and randomness index through a sliding time window, a set of disturbance characteristics for characterizing the uncertainty of photovoltaic output is obtained.

[0031] Based on the perturbation feature set, dynamic correction parameters are generated to characterize the impact of photovoltaic output randomness on node voltage, current and line impedance. The correction parameters are then embedded into the static reference model to obtain a circuit analytical model with photovoltaic output perturbation characteristics.

[0032] S2, based on the circuit analytical model, extracts the equivalent response curves under different fault scenarios and generates differentiated residual sequences by comparing them with the baseline operating state;

[0033] In this embodiment, the step of extracting equivalent response curves under different fault scenarios based on the circuit analytical model and generating differentiated residual sequences by comparing them with the baseline operating state is specifically as follows:

[0034] In the circuit analytical model, disturbance conditions of different fault conditions such as single-phase grounding, phase-to-phase short circuit and high-resistance grounding are applied to obtain dynamic change data of voltage and current at each node.

[0035] Based on dynamically changing data, the nonlinear coupling relationship between the node equivalent impedance and the injected current is solved to generate node response functions under different fault scenarios;

[0036] Based on the node response function, plot the curves of the evolution of node voltage and current residuals over time, and compare them with the baseline curves under fault-free conditions to obtain a set of differentiated equivalent response curves.

[0037] Based on the equivalent response curve set, the difference residual sequence between the baseline curve and the fault curve is calculated, and the residual characteristics under different fault conditions are labeled.

[0038] The method involves solving the nonlinear coupling relationship between the node's equivalent impedance and the injected current based on dynamically changing data, and generating node response functions under different fault scenarios. Specifically:

[0039] Based on the dynamic data, combined with the line impedance parameters and the measured values ​​of node voltage and current, the equivalent impedance matrix of nodes under fault disturbance is constructed.

[0040] Based on the node equivalent impedance matrix and node injected current, a nonlinear coupling expression between voltage response and current input is established.

[0041] A recursive iterative method is used to numerically solve the nonlinear coupling expression to obtain the node response function under different fault scenarios.

[0042] S3, perform time-frequency joint decomposition on the differential residual sequence to separate the photovoltaic power output fluctuation and load disturbance components, and obtain pure circuit characteristic components;

[0043] In this embodiment, the step of performing time-frequency joint decomposition on the differential residual sequence to separate photovoltaic power output fluctuations from load disturbance components and obtain pure circuit characteristic components specifically involves:

[0044] Based on the photovoltaic output prediction curve and load inertia characteristics, a dual constraint condition for time-frequency decomposition is constructed. The photovoltaic output prediction curve is used to limit the components in the residual sequence that exhibit periodic fluctuations with changes in illumination, while the load inertia characteristics are used to describe the low-frequency disturbance trend caused by large-scale load switching.

[0045] Wavelet packet decomposition is performed on the differential residual sequence, and the optimal sub-band is selected for reconstruction based on the residual energy concentration to obtain the reconstructed components;

[0046] The reconstructed components are mapped to a time-frequency energy map, and high curvature abrupt changes are identified as fault signals using a sparse curvature tracking method. These curvature abrupt changes correspond to rapid nonlinear changes in circuit parameters and can be considered as response characteristics of fault occurrence. Photovoltaic power fluctuations and load disturbances, on the other hand, appear as components with low curvature and smooth changes on the time-frequency map. This allows for the preliminary identification and separation of fault signals.

[0047] By correcting the decomposition boundary through an adversarial reconstruction mechanism, the photovoltaic power output fluctuation and load disturbance components are finally removed, resulting in pure circuit characteristic components.

[0048] The method of correcting the decomposition boundary through the adversarial reconstruction mechanism specifically includes:

[0049] A discriminative model is trained iteratively, and the differences between the stripped components and real fault samples are compared. If the deviation is too large, the level and sub-band selection of wavelet packet decomposition are adjusted until the separation effect meets the accuracy threshold. This mechanism can further eliminate misclassification and omission, and improve the purity of the stripped features.

[0050] Through the above steps, this embodiment can effectively remove the photovoltaic output fluctuation and load disturbance components in complex photovoltaic grid-connected environments, and obtain pure circuit characteristic components that only characterize the fault mechanism, providing reliable input for subsequent fault type identification and location.

[0051] In this embodiment, wavelet packet decomposition is performed on the differentiated residual sequence, and the optimal sub-band is selected for reconstruction based on the residual energy concentration. Specifically:

[0052] Differential residual sequences J-level wavelet packet decomposition was performed to obtain sub-signals in different frequency bands. , where j represents the decomposition layer number and k represents the corresponding sub-band number;

[0053] Calculate the energy of each sub-signal to form an energy distribution set, and calculate the energy proportion of each sub-band based on the energy distribution set to obtain candidate sub-bands;

[0054] The residual mutation rate index and the frequency band coherence index of each candidate sub-band are calculated, and after normalization, they are weighted and fused to obtain the sub-band comprehensive weight coefficient.

[0055] When performing inverse wavelet packet reconstruction, subband comprehensive weighting coefficients are applied to candidate subband signals, and attenuation factors are applied to non-target subband signals to obtain the reconstructed signal.

[0056] The reconstructed signal is compared with the original residual sequence for correlation. If the correlation coefficient is lower than the preset threshold, it indicates that there are still many non-fault components. At this time, the comprehensive weight coefficient is automatically adjusted and the reconstruction is repeated until the correlation reaches the threshold to obtain the final reconstructed component.

[0057] The calculation of the energy of each sub-signal is specifically as follows:

[0058]

[0059] The calculation of the energy percentage of each sub-band is specifically as follows:

[0060]

[0061] in, The energy of each sub-signal, These are sub-signals in different frequency bands. Let be the total energy of the wavelet packet decomposition at the same level, and m be the sub-band number of all waveslet packets in the same level of wavelet packet decomposition. This represents the energy percentage of each subband.

[0062] The residual mutation rate index is used to measure the strength of transient mutations in a photovoltaic grid signal in a certain subband, so as to reflect whether the subband contains typical grid fault impact characteristics.

[0063] The frequency band coherence index is used to measure the degree of correlation between a sub-band signal and the theoretical fault response under the circuit analytical model, in order to determine whether the sub-band is a fault-sensitive frequency band.

[0064] The residual mutation rate index is obtained through the following steps: constructing a first-order difference sequence for the sub-band signal and identifying the number of mutation points whose difference amplitude exceeds a preset threshold. The ratio of the number of mutation points to the total number of sampling points is defined as the residual mutation rate of the subband.

[0065] The frequency band coherence index is obtained by the following steps: calculating the cross spectrum between the sub-band signal and the theoretical fault response, and combining their respective power spectra to obtain the coherence coefficient; averaging the coherence coefficient within the target frequency band to obtain the frequency band coherence index of the sub-band.

[0066] The residual mutation rate index is specifically calculated using the following formula:

[0067]

[0068] The first-order difference sequence is specifically:

[0069] The specific formula for calculating the frequency band coherence index is as follows:

[0070]

[0071]

[0072]

[0073] The sub-band comprehensive weighting coefficient is specifically as follows:

[0074]

[0075] in, The residual mutation rate is an indicator. This represents the number of abrupt changes where the differential amplitude exceeds a preset threshold. The total number of sampling points. It is a first-order difference sequence. As a frequency band coherence index, For the candidate subband set, The coherence coefficient, , These are the power spectra of the subband signal and the fault response, respectively. The preset adjustment parameter takes values ​​in the range [0,1]. Let be the frequency domain representation of the k-th sub-band signal in the j-th layer after Fourier transform. This is the theoretical fault response signal. The cross spectrum of subband signal and fault response. , These are the normalized residual mutation rate and the frequency band coherence index, respectively. This is the sub-band comprehensive weighting coefficient.

[0076] The reconstructed signal is specifically:

[0077]

[0078] in, To reconstruct the signal, As the attenuation factor, This is the set of indices for the non-target subband.

[0079] S4, based on the feature components of the pure circuit, forms a multi-dimensional feature coordinate space, and uses the dynamic curvature of the fault response trajectory to determine the fault type;

[0080] The fault response trajectory refers to the dynamic curve formed by the evolution of the residual sequence over time in the multidimensional circuit feature space, which is used to reflect the response differences under different fault types.

[0081] In this embodiment, the step of forming a multi-dimensional feature coordinate space based on the feature components of the pure circuit, and using the dynamic curvature of the fault response trajectory to determine the fault type, specifically involves:

[0082] The amplitude, phase, and energy characteristics of the pure circuit's feature components at different time points or frequency points are combined to form a multidimensional feature vector for each node at each time / frequency point. ;

[0083] Mapping multidimensional feature vectors to multidimensional feature coordinate space Furthermore, feature vectors from the same fault scenario are concatenated in chronological order to form a fault response trajectory. The multidimensional feature coordinate space is used to describe the overall dynamic behavior of the circuit fault response;

[0084] The local curvature of each fault response trajectory at consecutive points is calculated and statistically analyzed to form classification rules. The curvature value reflects the transient change intensity and dynamic characteristics of the response trajectory.

[0085] The curvature of the fault response trajectory under test is matched with classification rules to determine the fault type. The fault types include single-phase grounding, phase-to-phase short circuit, and high-resistance grounding.

[0086] The specific formula for calculating the local curvature is as follows:

[0087]

[0088] in, For local curvature, The first derivative of the fault response trajectory. It is the second derivative of the fault response trajectory.

[0089] S5 maps the fault type determination result back to the circuit analytical model, and combines the local disturbance distribution of the node impedance matrix to locate the fault branch.

[0090] In this embodiment, mapping the fault type determination result back to the circuit analytical model and locating the fault branch location by combining the local perturbation distribution of the node impedance matrix specifically involves:

[0091] The set of nodes that may fail is determined based on the fault type discrimination results and marked in the circuit analytical model;

[0092] Based on the connection relationship of the marked nodes in the circuit analytical model, and combined with the node impedance matrix, the local impedance perturbation index of each marked node under the fault scenario is calculated. The local impedance perturbation index reflects the degree of impact of the fault on the node voltage and current.

[0093] By combining the node local impedance disturbance index with the corresponding node fault type information, a node fault sensitivity score is generated.

[0094] The fault sensitivity scores of the nodes at both ends of each branch in the statistical circuit are used to determine the branch with the highest score as the faulty branch.

[0095] The node local impedance disturbance index is specifically as follows:

[0096]

[0097] The node fault sensitivity score is specifically as follows:

[0098]

[0099] in, This is an index of local impedance disturbance at the node. , Let be the voltage and current of node i under fault conditions, respectively. , Let be the voltage and current of node i under the baseline operating conditions, respectively. , Weighting coefficients (set based on historical data). This represents the maximum value of the local perturbation at all nodes. The correlation coefficient between node i and the fault type (obtained from the dynamic curvature discrimination result, in the range [0,1]). Node fault sensitivity score.

[0100] Example 2, Figure 2 This invention presents a photovoltaic power grid fault identification system based on circuit analysis, comprising the following modules:

[0101] Circuit analytical model construction module: used to obtain the operating parameters of the photovoltaic power grid and construct a circuit analytical model with photovoltaic output disturbance factors;

[0102] Differentiated residual sequence generation module: used to extract equivalent response curves under different fault scenarios based on circuit analytical models, and generate differentiated residual sequences by comparing them with the baseline operating state;

[0103] Feature decomposition module: used to perform time-frequency joint decomposition on differential residual sequences, separating photovoltaic power output fluctuations from load disturbance components to obtain pure circuit feature components;

[0104] Fault discrimination module: It is used to form a multi-dimensional feature coordinate space based on the feature components of pure circuit, and to judge the fault type by utilizing the dynamic curvature of the fault response trajectory;

[0105] Fault location module: used to map the fault type identification result back to the circuit analytical model, and locate the fault branch by combining the local disturbance distribution of the node impedance matrix.

[0106] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0107] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0108] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0109] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0110] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0111] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A photovoltaic grid fault identification method based on circuit analysis, characterized in that, The method comprises the following steps: obtaining operation parameters of a photovoltaic power grid, and constructing a circuit analysis model with a photovoltaic output disturbance factor; based on the circuit analysis model, extracting equivalent response curves under different fault scenarios, and generating a differential residual sequence by comparing a reference operation state; performing time-frequency joint decomposition on the differential residual sequence to separate photovoltaic output fluctuation and load disturbance components, and obtaining a pure circuit characteristic component, specifically: based on a photovoltaic output prediction curve and load inertia characteristics, constructing double constraint conditions for time-frequency decomposition; performing wavelet packet decomposition on the differential residual sequence for several layers to obtain sub-signals in different frequency bands; calculating the energy of each sub-signal to form an energy distribution set, and calculating the energy proportion of each sub-band based on the energy distribution set to obtain a candidate sub-band; calculating the residual mutation rate index and the frequency band coherence index of each candidate sub-band, and performing normalization processing and weighted fusion on the two indexes to obtain a sub-band comprehensive weight coefficient; when performing inverse wavelet packet reconstruction, the candidate sub-band signal is subjected to the sub-band comprehensive weight coefficient, and a non-target sub-band signal is subjected to an attenuation factor to obtain a reconstructed signal; comparing the correlation of the reconstructed signal and the original residual sequence, if the correlation coefficient is lower than a preset threshold, adjusting the comprehensive weight coefficient and repeating the reconstruction until the correlation reaches the threshold to obtain a final reconstructed component; mapping the reconstructed component to a time-frequency energy diagram, and identifying a high curvature mutation point as a fault signal by using a sparse curvature tracking method; correcting the decomposition boundary through an adversarial reconstruction mechanism to separate the photovoltaic output fluctuation and the load disturbance component, and obtaining a pure circuit characteristic component; based on the pure circuit characteristic component, forming a multi-dimensional feature coordinate space, and using the dynamic bending rate of the fault response trajectory to distinguish the fault type; mapping the fault type discrimination result back to the circuit analysis model, and positioning the fault branch location combined with the local disturbance distribution of the node impedance matrix.

2. The circuit analysis based photovoltaic grid fault identification method of claim 1, wherein, The operation parameters of the photovoltaic power grid are obtained, and a circuit analysis model with a photovoltaic output disturbance factor is constructed, specifically: obtaining the voltage, current and line impedance parameters of each node of the photovoltaic power grid, and establishing an initial equivalent circuit analysis model based on the electrical topological relationship as a static reference model; introducing the time series data of photovoltaic power in the static reference model, analyzing the power fluctuation amplitude, change rate and randomness index through a sliding time window to obtain a disturbance feature set for characterizing the uncertainty of photovoltaic output; based on the disturbance feature set, generating dynamic correction parameters representing the influence of photovoltaic output randomness on node voltage, current and line impedance, and embedding the correction parameters into the static reference model to obtain a circuit analysis model with photovoltaic output disturbance characteristics.

3. The circuit analysis based photovoltaic grid fault identification method of claim 2, wherein, Based on the circuit analysis model, the equivalent response curves under different fault scenarios are extracted, and the differential residual sequence is generated by comparing the reference operation state, specifically: in the circuit analysis model, the disturbance conditions of single-phase grounding, phase-to-phase short circuit and high-resistance grounding are applied to obtain the dynamic change data of the node voltage and current; based on the dynamic change data, the nonlinear coupling relationship between the node equivalent impedance and the injected current is solved to generate the node response function under different fault scenarios; According to the node response function, the curves of the node voltage and the current residual are drawn with time evolution, and the benchmark curve under the fault-free working condition is compared to obtain a set of differential equivalent response curves; Based on the set of equivalent response curves, the difference residual sequence between the benchmark curve and the fault curve is calculated, and the residual characteristics under different fault working conditions are labeled.

4. The circuit analysis based photovoltaic grid fault identification method of claim 3, wherein, The nonlinear coupling relationship between the node equivalent impedance and the injected current is solved based on the dynamic change data, and the node response function under different fault scenarios is generated, specifically: According to the dynamic change data, the node equivalent impedance matrix under fault disturbance is constructed by combining the line impedance parameters and the node voltage and current measurement values; Based on the node equivalent impedance matrix and the node injected current, a nonlinear coupling expression between the voltage response and the current input is established; The nonlinear coupling expression is numerically solved by using the recursive iteration method to obtain the node response function under different fault scenarios.

5. The circuit analysis based photovoltaic grid fault identification method of claim 4, wherein, The decomposition boundary is corrected through the adversarial reconstruction mechanism, specifically: By iteratively training a discrimination model, the difference between the stripped component and the real fault sample is compared, and if the deviation is too large, the level and sub-band selection of wavelet packet decomposition are adjusted until the separation effect meets the accuracy threshold.

6. The circuit analysis based photovoltaic grid fault identification method of claim 5, wherein, Based on the pure circuit feature components, a multi-dimensional feature coordinate space is formed, and the dynamic bending rate of the fault response trajectory is used to distinguish the fault type, specifically: The amplitude, phase, and energy features of the pure circuit feature components at different time points or frequency points are combined to form a multi-dimensional feature vector for each node; The multi-dimensional feature vector is mapped to a multi-dimensional feature coordinate space, and the feature vectors under the same fault scenario are connected in time sequence to form a fault response trajectory; The local bending rate of each fault response trajectory is calculated and statistically analyzed to form a classification rule; The bending rate of the fault response trajectory to be tested is matched with the classification rule to distinguish the fault type.

7. The circuit analysis based photovoltaic grid fault identification method of claim 6, wherein, The fault type discrimination result is mapped back to the circuit analysis model, and the local disturbance distribution of the node impedance matrix is combined to locate the fault branch position, specifically: Based on the fault type discrimination result, the node set that may occur fault is determined, and is marked in the circuit analysis model; Based on the connection relationship of the marked nodes in the circuit analysis model, the local impedance disturbance index of each marked node under the fault scenario is calculated by combining the node impedance matrix; The node local impedance disturbance index is combined with the fault type information of the corresponding node to generate a node fault sensitivity score; The fault sensitivity scores of the nodes at both ends of each branch in the circuit are counted, and the branch with the highest score is determined as the fault branch.

8. A system for using a circuit analysis based photovoltaic grid fault identification method as claimed in any one of claims 1 to 7, characterized by, It comprises the following modules: A circuit analysis model construction module is used to obtain the operating parameters of the photovoltaic power grid and construct a circuit analysis model with photovoltaic output disturbance factors; A differential residual sequence generation module is used to extract equivalent response curves under different fault scenarios based on the circuit analysis model, and generate differential residual sequences by comparing the benchmark operating state; A feature decomposition module is used to perform time-frequency joint decomposition on the differential residual sequence, separate the photovoltaic output fluctuation and load disturbance components, and obtain pure circuit feature components. The fault discrimination module is used for forming a multi-dimensional feature coordinate space based on the pure circuit feature components, and discriminating the fault type by using the dynamic bending rate of the fault response trajectory. The fault positioning module is used for mapping the fault type discrimination result back to the circuit analysis model, and positioning the fault branch position in combination with the local disturbance distribution of the node impedance matrix.

Images