Series arc fault and reignition detection method based on time-frequency characteristics

Abstract

The invention relates to the field of electrical equipment fault diagnosis and safety protection calculation, in particular to a series arc fault and reignition detection method based on time-frequency characteristics. The method comprises the following steps: collecting and preprocessing a line current signal; performing high-resolution time-frequency analysis on the signal by adopting generalized S-transform, and extracting time-frequency characteristics in a frequency band; constructing a two-dimensional feature vector; and global optimization is carried out on the support vector machine hyper-parameters by using a grey wolf optimization-particle swarm hybrid algorithm, a classification model is established, and accurate identification of three working conditions of a normal state, a fault arc and contact arc reignition is realized. According to the method, the classification model is established, the defects of a traditional method in the aspect of distinguishing instantaneous reignition and continuous fault arcs are overcome, the method has the advantages of being high in recognition accuracy, high in anti-interference capacity, clear in physical significance and the like, and a reliable technical scheme is provided for early fault early warning and intelligent protection of an electrical system.

Description

Technical Field

[0001] This invention relates to the field of electrical equipment fault diagnosis and safety protection technology, and in particular to a method for detecting series arc faults and reignition based on time-frequency characteristics. Background Technology

[0002] Currently, research on arc detection in series faults mainly relies on the characteristic analysis of line current signals, leading to three mainstream technical paths based on different analysis domains. Methods based on time-domain characteristics extract statistical properties of the current waveform in the time domain for judgment, such as identifying the "flat shoulder" phenomenon in the current zero-crossing region, calculating the current difference between adjacent cycles, and statistically analyzing the current "zero-crossing" time. These methods are intuitive in principle, have low computational complexity, and are easy to implement on low-cost hardware. However, the extracted features are extremely sensitive to normal load fluctuations, startup shocks, and complex electromagnetic noise interference in the field, resulting in poor robustness. In actual power environments containing nonlinear loads, their false alarm rate increases significantly, and their stability fails to meet the requirements of high-reliability protection. Methods based on frequency-domain characteristics mainly rely on the Fast Fourier Transform (FFT), detecting faults by analyzing the distortion of harmonic components in the current signal before and after the fault, and the increase in the amplitude of specific high-frequency components. While the Fourier Transform can effectively reveal the global spectral structure of a signal, it is essentially a global transformation and cannot provide local information on the evolution of spectral components over time. For electric arcs, a typical non-stationary, transient, and abrupt signal, frequency domain analysis methods lose their crucial time-frequency localization characterization capabilities, making it difficult to capture the precise moment of fault occurrence and its dynamic evolution, resulting in insufficient timeliness of detection. Summary of the Invention

[0003] This invention aims to at least solve one of the technical problems existing in related technologies. To this end, this invention provides a method for detecting series arc faults and reignition based on time-frequency characteristics, which enables accurate and rapid identification of three operating conditions: normal state, series fault arc, and contact arc reignition, thereby providing a reliable technical means for early fault warning and active safety protection of electrical systems.

[0004] This invention provides a method for detecting series arc faults and reignition based on time-frequency characteristics, comprising: S1: Real-time acquisition of the current signal of the circuit under test, and filtering, denoising and standardization preprocessing of the current signal to obtain the current signal sequence to be analyzed; S2: Perform time-frequency analysis on the current signal sequence to be analyzed. Use the generalized S-transform to process the current signal sequence to be analyzed and obtain the time-frequency distribution matrix of the signal in the 0-10kHz frequency band. S3: Based on the time-frequency distribution matrix, calculate and extract the 2kHz root mean square amplitude ratio and singular value entropy; S4: Combine the extracted 2kHz root mean square amplitude ratio with the singular value entropy to construct a two-dimensional feature vector; S5: Input the two-dimensional feature vector into the pre-trained support vector machine classification model and output the classification result. The classification result is used to indicate whether the current line status is normal, fault arc, or contact arc reignition.

[0005] According to the present invention, a method for detecting series arc faults and reignition based on time-frequency characteristics is provided, wherein the window function of the generalized S-transform is an adjustable Gaussian window. Its expression is: in, For frequency, For time, Parameters for controlling the position of the Gaussian window on the time axis, As an adjustment factor, λ takes values ​​ranging from 0 to 1.

[0006] According to the present invention, a method for detecting series arc faults and reignition based on time-frequency characteristics is provided, wherein the adjustment factor is 0.7.

[0007] According to the present invention, a method for detecting series arc faults and reignition based on time-frequency characteristics is provided. The method for calculating the 2kHz root mean square amplitude ratio is as follows: extract the amplitude sequence near the 2kHz frequency component from the time-frequency distribution matrix, calculate its root mean square value, and then compare the root mean square value with the overall root mean square value of the entire 0-10kHz analysis frequency band to obtain the 2kHz root mean square amplitude ratio. The method for calculating the singular value entropy is as follows: perform singular value decomposition on the time-frequency distribution matrix to obtain a singular value sequence, calculate the information entropy of the singular value sequence, and obtain the singular value entropy.

[0008] According to the time-frequency characteristic-based series arc fault and reignition detection method provided by the present invention, step S5 involves global optimization using a gray wolf optimizer-particle swarm optimization hybrid algorithm. The optimization process of the gray wolf optimizer-particle swarm optimization hybrid algorithm includes: S51: Set up a mixed population and randomly initialize the parameter combination of the support vector machine as the individual position; S52: In each iteration, each individual updates its position simultaneously based on the social hierarchy guidance mechanism of the gray wolf optimizer and the individual experience and group experience guidance mechanism of the particle swarm optimization. S53: Calculate the fitness of each individual's current position, where the fitness is the classification accuracy of the support vector machine model constructed with the parameter combination on the validation set; S54: Repeat steps S52-S53 until the preset termination condition is reached, and output the parameter combination with the highest fitness as the optimal hyperparameter as the output classification result.

[0009] According to the time-frequency characteristic-based series arc fault and reignition detection method provided by the present invention, in step S52, the position update formula of the gray wolf optimizer is: in, For time, for First update location in time for The location of the most fit individual in the current population at time. This is the first coefficient vector of the individual with the best fitness in the current population. This represents the second coefficient vector of the individual with the best fitness in the current population. for Time-based optimization is needed for individual location; for Second update position in time. for The position of the second-best fittest individual in the current population at time. This is the first coefficient vector of the second-best fittest individuals in the current population. The second coefficient vector of the second fittest individual in the current population. for The third update position in time. for The position of the third fittest individual in the current population at that time. This is the first coefficient vector of the third fittest individual in the current population. This is the second coefficient vector of the third fittest individual in the current population. This indicates a modulo operation. Update the position of the Grey Wolf optimizer; The velocity and position update formulas for particle swarm optimization are: in, The velocity results for particle swarm optimization For inertial weights, For the individual speed to be optimized, The first acceleration constant, Let be the first random vector. For individuals, based on their own historical best position The second acceleration constant, For the second random vector, This represents the best historical position for the entire population. Position updates optimized for particle swarm optimization; The hybrid algorithm fuses data through a weighted or selection mechanism. and Generate the next generation of individual locations : in, The fusion weights are dynamically adjusted.

[0010] According to the present invention, a method for detecting series arc faults and reignition based on time-frequency characteristics further includes, after step S5: S6: Perform the corresponding operation according to the classification result; if the identification result is a normal state, the system continues to monitor; if the identification result is a fault arc state or a contact arc reignition state, the system immediately issues an alarm signal and outputs a trip command to the circuit breaker to disconnect the fault circuit.

[0011] The present invention also provides a series arc fault and reignition detection system based on time-frequency characteristics, comprising: Signal acquisition and preprocessing module: used to acquire AC current signals of the circuit under test in real time, and to perform filtering, noise reduction and standardization preprocessing on the current signals to obtain the current signal sequence to be analyzed; Time-frequency analysis module: used to perform time-frequency analysis on the current signal sequence to be analyzed. It uses generalized S-transform to process the current signal sequence to be analyzed and obtains the time-frequency distribution matrix of the signal in the 0-10kHz frequency band. Feature extraction module: used to calculate and extract the 2kHz root mean square amplitude ratio and singular value entropy based on the time-frequency distribution matrix; Feature fusion module: used to combine the extracted 2kHz root mean square amplitude ratio with the singular value entropy to construct a two-dimensional feature vector; Model recognition and state determination module: used to input the two-dimensional feature vector into a pre-trained support vector machine classification model and output the classification result. The classification result is used to indicate whether the current line state is normal, fault arc state or contact arc reignition state.

[0012] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the series arc fault and reignition detection method based on time-frequency characteristics as described above.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the series arc fault and reignition detection method based on time-frequency characteristics as described above.

[0014] The above-described one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects: This invention provides a method for detecting series arc faults and reignition based on time-frequency characteristics. Its beneficial effects include: 1. This invention proposes a new framework for fault arc detection that integrates high-resolution time-frequency analysis and hybrid intelligent optimization. It accurately extracts the transient and steady-state time-frequency features of the arc signal through generalized S-transform, effectively overcoming the problems of difficult basis function selection, spectral aliasing, and inaccurate feature representation in traditional wavelet transform and other methods.

[0015] 2. This invention innovatively extracts and integrates two complementary features with clear physical meaning and strong discriminative power, namely "2kHz root mean square amplitude ratio" and "singular value entropy", to construct a low-dimensional and efficient two-dimensional feature vector, which lays a solid foundation for high-precision classification and significantly reduces the computational complexity of subsequent models.

[0016] 3. This invention develops an SVM hyperparameter optimization method based on the GWO-PSO hybrid algorithm, which effectively combines the advantages of the two algorithms and avoids the shortcomings of standard PSO being prone to getting trapped in local optima and GWO having slow convergence in the later stages, so that the SVM classifier can obtain better generalization performance and higher recognition accuracy.

[0017] 4. This invention is the first to clearly define the independent status of "contact arc reignition" as a key predictive fault in the detection framework, and realizes the three-dimensional accurate identification of normal state, continuous fault arc and transient contact arc reignition. This not only improves the dimensions of fault detection, but also provides the possibility for condition-based predictive maintenance of switchgear and ultra-early prevention of electrical fires.

[0018] 5. The feature vector and classification model constructed in this invention have high computational efficiency and are easy to implement on embedded hardware platforms, meeting the engineering requirements of online monitoring and real-time protection. This provides a complete and reliable technical solution for developing a new generation of intelligent arc fault protection devices. Existing technologies have not yet reported on the deep integration of the high-resolution time-frequency analysis capability of the Generalized S-Transform (GST) with a Support Vector Machine (SVM) model globally optimized using a hybrid Gray Wolf Optimization-Particle Swarm Optimization (GWO-PSO) algorithm, and the construction of a ternary identification framework capable of accurately distinguishing between "persistent fault arcs" and "transient contact arc reignition." This is precisely the core technical challenge that this invention aims to solve.

[0019] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating a method for detecting series arc faults and reignition based on time-frequency characteristics, provided in an embodiment of the present invention.

[0022] Figure 2 A three-dimensional schematic diagram of time-frequency analysis of three typical current signals based on the generalized S-transform (GST) provided for embodiments of the present invention.

[0023] Figure 3 This is a schematic diagram of the optimization process of the Grey Wolf Optimizer-Particle Swarm Optimization (GWO-PSO) hybrid algorithm used in an embodiment of the present invention.

[0024] Figure 4 This is a comparison of the fitness convergence curves of different optimization algorithms in the process of optimizing the hyperparameters of the support vector machine in the embodiments of the present invention.

[0025] Figure 5 This is a structural block diagram of a series arc fault and reignition detection system based on time-frequency characteristics, provided in an embodiment of the present invention.

[0026] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention.

[0027] Figure label: 101. Signal Acquisition and Preprocessing Module; 102. Time-Frequency Analysis Module; 103. Feature Extraction Module; 104. Feature Fusion Module; 105. Model Recognition and State Determination Module; 810. Processor; 820. Communication Interface; 830. Memory; 840. Communication Bus. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention. The following embodiments are used to illustrate this invention but cannot be used to limit the scope of this invention.

[0029] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0030] The following is combined Figures 1 to 6 This invention is described.

[0031] Example like Figure 1 As shown, this invention provides a method for detecting series arc faults and reignition based on time-frequency characteristics, comprising the following steps: S1: Real-time acquisition of the current signal of the circuit under test, and filtering, denoising and standardization preprocessing of the current signal to obtain the current signal sequence to be analyzed; S2: Perform time-frequency analysis on the current signal sequence to be analyzed. Use the generalized S-transform to process the current signal sequence to be analyzed and obtain the time-frequency distribution matrix of the signal in the 0-10kHz frequency band. S3: Based on the time-frequency distribution matrix, calculate and extract the 2kHz root mean square amplitude ratio and singular value entropy; S4: Combine the extracted 2kHz root mean square amplitude ratio with the singular value entropy to construct a two-dimensional feature vector; S5: Input the two-dimensional feature vector into the pre-trained support vector machine classification model and output the classification result. The classification result is used to indicate whether the current line status is normal, fault arc, or contact arc reignition.

[0032] Specifically, step S1 includes: The AC current signal of the circuit under test is acquired in real time through a standardized AC arc test platform. Based on equipment operation events or preset cycles, the continuous current waveform is segmented into signal segments for independent analysis periods. Digital filtering technology is used to achieve frequency band separation and noise suppression, effectively filtering out power frequency interference and random high-frequency noise. The denoised signal is then Z-score standardized to eliminate amplitude differences caused by load fluctuations, forming a standardized data sequence suitable for subsequent time-frequency analysis.

[0033] Specifically, the window function of the generalized S-transform in step S2 is an adjustable Gaussian window. Its expression is: in, For frequency, For time, Parameters for controlling the position of the Gaussian window on the time axis, The adjustment factor λ ranges from 0 to 1. In this embodiment of the invention, the adjustment factor is 0.7. Comparative experiments show that when the adjustment factor λ is 0.7, the generalized S-transform achieves the best balance between time resolution and frequency resolution, and can most clearly present the transient high-frequency characteristics of contact arc reignition. Therefore, it is the preferred embodiment.

[0034] The generalized S-transform employs a frequency-dependent window function Fourier transform for time-frequency domain analysis, with an adjustable Gaussian window as its key element. This technique transforms one-dimensional time-series data into a two-dimensional time-spectrum surface, effectively capturing the local time-domain characteristics and spectral composition of the signal. The mathematical expression of the generalized S-transform is as follows: Defined as: in, This is the preprocessed, standardized current signal. It is the imaginary unit.

[0035] To suppress boundary effects, a symmetric mirror extension preprocessing strategy is employed in the actual calculations: 5% of the sampling points at each end of the signal are reverse-mirrored to enhance boundary continuity. After transformation, the time-frequency distribution of the corresponding segment of the original signal is extracted as the final result. This processing significantly reduces edge distortion during the transformation process and improves the reliability of feature extraction. The resulting time-frequency distribution matrix clearly characterizes the energy distribution differences in the time-frequency plane among the three states: normal current, fault arc, and contact arc reignition, providing a high-resolution time-frequency representation foundation for subsequent essential feature extraction.

[0036] Specifically, based on the high-resolution time-frequency distribution matrix obtained from the generalized S-transform, to establish an accurate state recognition model, it is necessary to extract sensitive features that can essentially distinguish between normal states, series fault arcs, and contact arc reignition. This study selects and defines two core features with clear physical meaning and complementary discriminative dimensions: "2kHz root mean square amplitude ratio" and "singular value entropy". By fusing the two, a low-dimensional, high-discriminative two-dimensional feature vector is constructed, providing high-quality input for subsequent intelligent classification.

[0037] Specifically, the calculation method for the 2kHz root mean square amplitude ratio is as follows: extract the amplitude sequence near the 2kHz frequency component from the time-frequency distribution matrix, calculate its root mean square value, and then compare the root mean square value with the overall root mean square value of the entire 0-10kHz analysis frequency band. in, The root mean square amplitude ratio is 2kHz. The root mean square value of the amplitude sequence for all time-frequency points within a narrow band centered at 2kHz (width not exceeding 10Hz). This is the root mean square value of the total amplitude sequence.

[0038] The physical significance lies in the fact that the phenomenon of contact arc reignition manifests as repetitive transient high-frequency breakdown, with its energy more easily concentrated in a specific high-frequency band; while the high-frequency energy of a persistent fault arc also increases, but its distribution is relatively broad and dispersed. Therefore, an abnormally high 2kHz root mean square amplitude ratio is a key indicator for identifying contact arc reignition.

[0039] The method for calculating the singular value entropy is as follows: perform singular value decomposition on the time-frequency distribution matrix to obtain a singular value sequence, and calculate the information entropy of the singular value sequence.

[0040] This feature aims to characterize the complexity and randomness of the time-frequency distribution of the current signal. Singular value decomposition is performed on the time-frequency distribution matrix to obtain a set of non-negative singular values. This sequence reflects the distribution of time-frequency energy on the principal components. The information entropy of this singular value sequence is calculated: The physical significance is that the time-frequency structure of a normal current signal is simple and orderly, with energy concentrated in a few principal components and a low singular entropy value; the discharge process of a persistent fault arc is random and chaotic, resulting in an extremely complex time-frequency structure, with energy dispersed among numerous components and a significantly increased singular entropy value; although the reignition of a contact arc is a transient process, its pattern has certain regularity, and its singular entropy value is usually between that of a normal state and a fault arc.

[0041] To verify the discriminative power of the above features, a large amount of sample data covering various load types, including resistive and inductive loads, was acquired on a standardized experimental platform, and their eigenvalues ​​were calculated. Taking resistive loads as an example, a generalized S-transform analysis was performed to extract five most relevant feature parameters from different aspects: 2kHz root mean square amplitude ratio, singular value entropy, 2kHz energy amplitude ratio, instantaneous frequency variance, and ridge steepness. The results calculated for three cases—normal current, fault arc, and contact arc reignition—are shown in Table 1.

[0042] Table 1 Calculation results of characteristic parameters

[0043] Under normal conditions, both the 2kHz root mean square amplitude ratio and the singular entropy value remain at a low level. Under fault arc conditions, the 2kHz root mean square amplitude ratio increases significantly, and the singular entropy value increases dramatically, indicating a surge in signal complexity. However, under the contact arc reignition condition, the 2kHz root mean square amplitude ratio is abnormally high (even exceeding that of some fault arc samples). This pattern remains stable under different load conditions, confirming that the selected feature has a stable and significant ability to distinguish between the three states.

[0044] By fusing the 2kHz root mean square amplitude ratio and the singular value entropy—two features extracted from two orthogonal physical dimensions, namely "high-frequency energy concentration" and "time-frequency structural complexity"—this low-dimensional vector integrates complementary discriminative information, enabling the three classes of samples to form good clusters and separations in the feature space. A visualization can be found here. Figure 2 This effectively avoids the "curse of dimensionality" and computational redundancy that high-dimensional features may bring, laying a solid foundation for subsequent fast and accurate recognition based on classifiers such as support vector machines. Figure 2 (a) is the time-frequency diagram of the arc current interrupted by a normal switch. Figure 2 (b) is the time-frequency diagram of the contact arc reignition current. Figure 2 (c) is the time-frequency diagram of the arc current in a series fault.

[0045] Specifically, such as Figure 3 As shown, in step S5, global optimization is performed using a hybrid algorithm of gray wolf optimizer and particle swarm optimization. The optimization process of the gray wolf optimizer-particle swarm optimization hybrid algorithm includes: S51: Set up a mixed population and randomly initialize the parameter combination of the support vector machine as the individual position; S52: In each iteration, each individual updates its position simultaneously based on the social hierarchy guidance mechanism of the gray wolf optimizer and the individual experience and group experience guidance mechanism of the particle swarm optimization. S53: Calculate the fitness of each individual's current position, where the fitness is the classification accuracy of the support vector machine model constructed with the parameter combination on the validation set; S54: Repeat steps S52-S53 until the preset termination condition is reached, and output the parameter combination with the highest fitness as the optimal hyperparameter.

[0046] Specifically, in step S52, the position update formula for the gray wolf optimizer is: in, For time, for First update location in time for The location of the most fit individual in the current population at time. This is the first coefficient vector of the individual with the best fitness in the current population. This represents the second coefficient vector of the individual with the best fitness in the current population. for Time-based optimization is needed for individual location; for Second update position in time. for The position of the second-best fittest individual in the current population at time. This is the first coefficient vector of the second-best fittest individuals in the current population. The second coefficient vector of the second fittest individual in the current population. for The third update position in time. for The position of the third fittest individual in the current population at that time. This is the first coefficient vector of the third fittest individual in the current population. This is the second coefficient vector of the third fittest individual in the current population. This indicates a modulo operation. Update the position of the Grey Wolf optimizer; The velocity and position update formulas for particle swarm optimization are: in, The velocity results for particle swarm optimization For inertial weights, For the individual speed to be optimized, The first acceleration constant, Let be the first random vector. For individuals, based on their own historical best position The second acceleration constant, For the second random vector, This represents the best historical position for the entire population. Position updates optimized for particle swarm optimization; The hybrid algorithm fuses data through a weighted or selection mechanism. and Generate the next generation of individual locations in, The fusion weights are dynamically adjusted.

[0047] To verify the superiority of the proposed method, a comparative experiment was conducted, and the experimental results are as follows: Figure 4 As shown in the figure. The results show that the performance of the two-dimensional feature vector extracted based on the generalized S-transform and the feature vector extracted based on the wavelet transform is compared under the same SVM classifier. To construct and evaluate the classification model, this study collected a large amount of operating condition data on a standardized experimental platform covering resistive, resistive-inductive, common switching, and nonlinear loads. From this data, 150 sets of sample data were randomly selected proportionally, representing normal current, fault arc, and contact arc reignition states under various load types, to form the training set of the model. To objectively evaluate the generalization performance of the model, another 150 sets of sample data were prepared as an independent test set using the same data selection method. Finally, the experimental results show that the recognition accuracy of the features proposed by this method on the test set reaches 96.67%, which is significantly higher than the recognition rate of the wavelet transform features.

[0048] The GWO-PSO optimization algorithm used in this invention is compared with the standard optimization algorithms PSO, LS, and SSA-PSO on the same task. The GWO-PSO algorithm shows a significant advantage in classification accuracy optimization, maintaining stable convergence during iteration and achieving the highest accuracy, surpassing the benchmark algorithms PSO and SSA-PSO. This algorithm achieves 90% accuracy within 50 iterations, combining fast convergence with high precision. Its advantage stems from the effective integration of the global search capability of the Grey Wolf optimization algorithm and the local exploitation capability of particle swarm optimization. While maintaining computational efficiency, it avoids the premature convergence problem of traditional PSO, demonstrating its advantage in balancing global search and local exploitation.

[0049] The final trained GWO-PSO-SVM model can receive real-time extracted two-dimensional feature vectors as input, perform decision function calculations, and output corresponding state labels, providing accurate judgment basis for subsequent protection actions.

[0050] Specifically, step S6 includes: The final step of this detection method is to execute corresponding closed-loop control and protection actions based on the state judgment results output by the intelligent classification model, thereby transforming the online diagnostic conclusions into actual equipment responses and achieving proactive safety protection for the electrical system. The core of this step lies in building a stable, reliable, and rapidly responsive execution mechanism to ensure that alarms are issued promptly and faulty circuits are disconnected when dangerous conditions are identified.

[0051] After computation by the GWO-PSO-SVM model, the output is a discrete state label. The system executes a preset control strategy based on this label: like The system determines that the line is operating normally and does not trigger any protection actions. The data processing and identification module continues to analyze and monitor the subsequently acquired current signals in real time, forming a continuous online monitoring cycle.

[0052] like That is, the system is in a fault arc state; or If the system is in a state of contact arc reignition, the system will immediately determine that there is a safety hazard in the line and simultaneously activate a two-level protection response mechanism.

[0053] The system immediately activates the audible and visual alarm device via the I / O interface of the output and execution module. The alarm signals are designed to be clear and highly alert; the audible alarm uses an intermittent buzzer, and the visual alarm uses a flashing red LED to effectively alert on-site personnel to the risk of an arc fault. The alarm signal will continue until the system is reset or the fault condition is cleared.

[0054] Simultaneously with issuing the alarm signal, the system sends a tripping command to the intelligent circuit breaker or the tripping coil controlled by a relay on the target line. This command is a standard switching signal. To ensure selectivity and avoid false tripping, the command can be set with an extremely short adjustable delay, typically 20ms. This delay is sufficient to avoid transient interference from normal line operation or transient processes caused by switching operations, while ensuring rapid disconnection even during a persistent fault. After the tripping command is issued, the system monitors the status of the circuit breaker's auxiliary contacts or whether the line current has dropped to zero to confirm that the faulty circuit has been reliably isolated.

[0055] To achieve the above functions, the algorithm and logic of this method can be integrated and deployed on an embedded hardware platform. Depending on the implementation requirements, this platform can use a high-performance embedded microcontroller, digital signal processor, or field-programmable gate array as the core processing unit. At the software level, the entire signal processing flow and control logic are implemented with efficient code, ensuring that real-time requirements are met with limited hardware resources. Experimental verification shows that at a sampling rate of 100kHz, the entire process time from current signal acquisition to completion of state identification and preparation for control output can be controlled within 100ms, while the delay in issuing the tripping command can be controlled within 20ms, fully meeting the rapid action time requirements of relevant standards for arc fault protection.

[0056] Once a fault triggers the trip, the system will enter a protection lockout state, continuously outputting an alarm and maintaining the trip signal to prevent the faulty line from being accidentally reclosed. The lockout state must be released by maintenance personnel through on-site inspection and confirmation that the fault has been eliminated, followed by manual reset via the system reset button or remote command. The system will then return to normal monitoring status.

[0057] like Figure 5 As shown, this invention provides a series arc fault and reignition detection system based on time-frequency characteristics. The series arc fault and reignition detection system based on time-frequency characteristics described below can be referred to in correspondence with the series arc fault and reignition detection method based on time-frequency characteristics described above. It includes: Signal acquisition and preprocessing module 101: used to acquire AC current signals of the circuit under test in real time, and to perform filtering, noise reduction and standardization preprocessing on the current signals to obtain the current signal sequence to be analyzed; Time-frequency analysis module 102: used to perform time-frequency analysis on the current signal sequence to be analyzed, and to process the current signal sequence to be analyzed using generalized S-transform to obtain the time-frequency distribution matrix of the signal in the 0-10kHz frequency band; Feature extraction module 103: used to calculate and extract the 2kHz root mean square amplitude ratio and singular value entropy based on the time-frequency distribution matrix; Feature fusion module 104: used to combine the extracted 2kHz root mean square amplitude ratio with the singular value entropy to construct a two-dimensional feature vector; Model recognition and state determination module 105: used to input the two-dimensional feature vector into a pre-trained support vector machine classification model and output the classification result, which is used to indicate whether the current line state is normal, fault arc state or contact arc reignition state.

[0058] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other via the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute a method for detecting series arc faults and reignition based on time-frequency characteristics. This method includes: S1: Real-time acquisition of AC current signal of the circuit under test, and filtering, denoising and standardization preprocessing of the current signal to obtain the current signal sequence to be analyzed; S2: Perform time-frequency analysis on the current signal sequence to be analyzed. Use the generalized S-transform to process the current signal sequence to be analyzed and obtain the time-frequency distribution matrix of the signal in the 0-10kHz frequency band. S3: Based on the time-frequency distribution matrix, calculate and extract the 2kHz root mean square amplitude ratio and singular value entropy; S4: Combine the extracted 2kHz root mean square amplitude ratio with the singular value entropy to construct a two-dimensional feature vector; S5: Input the two-dimensional feature vector into a pre-trained support vector machine classification model and output the classification result. The classification result is used to indicate whether the current line status is normal, fault arc, or contact arc reignition.

[0059] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0060] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the aforementioned methods for detecting series arc faults and reignition based on time-frequency characteristics, the method comprising: S1: Real-time acquisition of AC current signal of the circuit under test, and filtering, denoising and standardization preprocessing of the current signal to obtain the current signal sequence to be analyzed; S2: Perform time-frequency analysis on the current signal sequence to be analyzed. Use the generalized S-transform to process the current signal sequence to be analyzed and obtain the time-frequency distribution matrix of the signal in the 0-10kHz frequency band. S3: Based on the time-frequency distribution matrix, calculate and extract the 2kHz root mean square amplitude ratio and singular value entropy; S4: Combine the extracted 2kHz root mean square amplitude ratio with the singular value entropy to construct a two-dimensional feature vector; S5: Input the two-dimensional feature vector into a pre-trained support vector machine classification model and output the classification result. The classification result is used to indicate whether the current line status is normal, fault arc, or contact arc reignition.

[0061] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0062] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0064] It should be noted that the embodiments of this disclosure can be implemented using hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a programmable memory or a data carrier such as an optical or electronic signal carrier.

[0065] Furthermore, although the operation of the methods of this disclosure is described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Rather, the steps depicted in the flowcharts may be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps. It should also be noted that the features and functions of two or more devices according to this disclosure may be embodied in one device. Conversely, the features and functions of one device described above may be further divided and embodied by multiple devices.

[0066] While this disclosure has been described with reference to several specific embodiments, it should be understood that this disclosure is not limited to the specific embodiments disclosed. This disclosure is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims

1. A method for detecting series arc faults and reignition based on time-frequency characteristics, characterized in that, Includes the following steps: S1: Real-time acquisition of the current signal of the circuit under test, and filtering, denoising and standardization preprocessing of the current signal to obtain the current signal sequence to be analyzed; S2: Perform time-frequency analysis on the current signal sequence to be analyzed. Use the generalized S-transform to process the current signal sequence to be analyzed and obtain the time-frequency distribution matrix of the signal in the 0-10kHz frequency band. S3: Based on the time-frequency distribution matrix, calculate and extract the 2kHz root mean square amplitude ratio and singular value entropy; S4: Combine the extracted 2kHz root mean square amplitude ratio with the singular value entropy to construct a two-dimensional feature vector; S5: Input the two-dimensional feature vector into the pre-trained support vector machine classification model and output the classification result. The classification result is used to indicate whether the current line status is normal, fault arc, or contact arc reignition.

2. The method for detecting series arc faults and reignition based on time-frequency characteristics according to claim 1, characterized in that, The window function of the generalized S-transform is an adjustable Gaussian window. Its expression is: in, For frequency, For time, To control the parameters of the Gaussian window's position on the time axis, As a regulating factor, The range of values ​​for is 0 < λ < 1.

3. The method for detecting series arc faults and reignition based on time-frequency characteristics according to claim 2, characterized in that, The adjustment factor is set to 0.

7.

4. The method for detecting series arc faults and reignition based on time-frequency characteristics according to claim 1, characterized in that, The method for calculating the 2kHz root mean square amplitude ratio is as follows: extract the amplitude sequence near the 2kHz frequency component from the time-frequency distribution matrix, calculate its root mean square value, and then compare the root mean square value with the overall root mean square value of the entire 0-10kHz analysis frequency band to obtain the 2kHz root mean square amplitude ratio. The method for calculating the singular value entropy is as follows: perform singular value decomposition on the time-frequency distribution matrix to obtain a singular value sequence, calculate the information entropy of the singular value sequence, and obtain the singular value entropy.

5. The method for detecting series arc faults and reignition based on time-frequency characteristics according to claim 1, characterized in that, In step S5, global optimization is performed using a hybrid algorithm of gray wolf optimizer and particle swarm optimization. The optimization process of the gray wolf optimizer-particle swarm optimization hybrid algorithm includes: S51: Set up a mixed population and randomly initialize the parameter combination of the support vector machine as the individual position; S52: In each iteration, each individual updates its position simultaneously based on the social hierarchy guidance mechanism of the gray wolf optimizer and the individual experience and group experience guidance mechanism of the particle swarm optimization. S53: Calculate the fitness of each individual's current position, where the fitness is the classification accuracy of the support vector machine model constructed with the parameter combination on the validation set; S54: Repeat steps S52-S53 until the preset termination condition is reached, and output the parameter combination with the highest fitness as the optimal hyperparameter as the output classification result.

6. The method for detecting series arc faults and reignition based on time-frequency characteristics according to claim 5, characterized in that, In step S52, the position update formula for the gray wolf optimizer is: in, For time, for First update location in time for The location of the most fit individual in the current population at time. This is the first coefficient vector of the individual with the best fitness in the current population. This represents the second coefficient vector of the individual with the best fitness in the current population. for Time-based optimization is needed for individual location; for Second update position in time. for The position of the second-best fittest individual in the current population at time. This is the first coefficient vector of the second-best fittest individuals in the current population. The second coefficient vector of the second fittest individual in the current population. for The third update position in time. for The position of the third fittest individual in the current population at that time. This is the first coefficient vector of the third fittest individual in the current population. This is the second coefficient vector of the third fittest individual in the current population. This indicates a modulo operation. Update the position of the Grey Wolf optimizer; The velocity and position update formulas for particle swarm optimization are: in, The velocity results for particle swarm optimization For inertial weights, For the individual speed to be optimized, The first acceleration constant, Let be the first random vector. For individuals, based on their own historical best position. The second acceleration constant, For the second random vector, This represents the best historical position for the entire population. Position updates optimized for particle swarm optimization; The hybrid algorithm fuses data through a weighted or selection mechanism. and Generate the next generation of individual locations : in, The fusion weights are dynamically adjusted.

7. The method for detecting series arc faults and reignition based on time-frequency characteristics according to claim 1, characterized in that, The process after step S5 also includes: S6: Perform the corresponding operation according to the classification result; if the identification result is a normal state, the system continues to monitor; if the identification result is a fault arc state or a contact arc reignition state, the system immediately issues an alarm signal and outputs a trip command to the circuit breaker to disconnect the fault circuit.

8. A series arc fault and reignition detection system based on time-frequency characteristics, used to execute the series arc fault and reignition detection method based on time-frequency characteristics as described in any one of claims 1 to 7, characterized in that, Signal acquisition and preprocessing module: used to acquire AC current signals of the circuit under test in real time, and to perform filtering, noise reduction and standardization preprocessing on the current signals to obtain the current signal sequence to be analyzed; Time-frequency analysis module: used to perform time-frequency analysis on the current signal sequence to be analyzed. It uses generalized S-transform to process the current signal sequence to be analyzed and obtains the time-frequency distribution matrix of the signal in the 0-10kHz frequency band. Feature extraction module: used to calculate and extract the 2kHz root mean square amplitude ratio and singular value entropy based on the time-frequency distribution matrix; Feature fusion module: used to combine the extracted 2kHz root mean square amplitude ratio with the singular value entropy to construct a two-dimensional feature vector; Model recognition and state determination module: used to input the two-dimensional feature vector into a pre-trained support vector machine classification model and output the classification result. The classification result is used to indicate whether the current line state is normal, fault arc state or contact arc reignition state.

9. An electronic device, comprising a processor, a communication interface, a memory, and a communication bus, characterized in that, When the processor executes the computer program, it implements the steps of the series arc fault and reignition detection method based on time-frequency characteristics as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for detecting series arc faults and reignition based on time-frequency characteristics as described in any one of claims 1 to 7.

Images