Electric arc detection method and system based on wavelet transformation

Abstract

The invention discloses an electric arc detection method and system based on wavelet transformation, and aims to solve the problem that fault electric arcs in a low-voltage power distribution system are difficult to detect. The method comprises the following steps: firstly, selecting a Cassie arc model, building a simulation model by utilizing Matlab / Simulink, and simulating arc conditions under various loads to obtain data; then db4 is screened out from multiple wavelet bases to serve as an optimal wavelet base, and discrete wavelet transformation is conducted on the collected current signals; the method is characterized in that an anti-spectrum aliasing improved wavelet transform algorithm based on odd extraction before node addition is provided, spectrum aliasing caused by traditional even extraction is eliminated, and fault features are accurately extracted. And determining a fault arc by taking the mean value, variance, energy, peak-to-average ratio and the like of the detail coefficient of the cD1 as criteria and combining a preset threshold value. The matching system is composed of a data acquisition unit, a processing unit, a judgment alarm unit and a storage unit. The method and the system can accurately detect the electric arc, reduce the misjudgment rate, have strong adaptability and guarantee the safety of an electric power system.

Description

Technical Field

[0001] The present invention belongs to the technical field related to power system fault detection, and particularly relates to an arc detection method and system based on wavelet transform. Background Art

[0002] With the modern development of low-voltage distribution systems, the electricity consumption and industrial load have climbed, and fault arcs have become one of the factors seriously threatening power safety. As a complex gas discharge phenomenon, fault arcs often occur in insulating media, accompanied by strong light, high heat, and the volatilization of electrode materials, and their physical processes are complex. During the operation of the power system, arcs can be triggered either by normal operations or by system faults. Among them, series fault arcs caused by insulation aging, loose connections, etc. are extremely difficult to detect and locate due to factors such as the insensitivity of traditional protection devices to fault current identification, the easy confusion of the normal operating current characteristics of household appliances, the masking of arcs by mixed loads, and electromagnetic distortion noise.

[0003] Currently, there are mainly three types of current fault arc detection methods: one is to identify by combining mathematical models. However, due to the complex arc mechanism and imperfect models, its development is slow and the algorithms are complex; the second is to identify based on physical characteristics. Although abnormal behaviors such as heat, light, and electromagnetism can be used for judgment, it is restricted by diverse load properties, fixed sensor positions, and the uncertainty of fault points; the third is to identify based on the change of loop electrical signal characteristics, mostly by measuring current. However, existing methods based on time domain, frequency domain, or time-frequency combination often require manual adjustment of preset thresholds.

[0004] Therefore, it is extremely urgent to develop a precise, efficient, and adaptable fault arc detection technology. Summary of the Invention

[0005] The purpose of the present invention is to provide an arc detection method and system based on wavelet transform to solve the problems raised in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] An arc detection method based on wavelet transform, comprising the following steps:

[0008] Arc fault model construction step:

[0009] Select the Cassie arc model, use Matlab / Simulink to build an arc simulation model, and set the power supply parameters, where the peak value of the power supply voltage is 220V, the frequency is 50Hz, and the initial phase angle is 90°. The model parameters are set as Uc = 130V, τ = 2.25e-4 seconds = 225 us, g(0) = 1.17e-4 S = 0.000117 S;

[0010] By changing the load type in the model, simulate the occurrence of arcs under various different load combinations including at least inductive, capacitive, and resistive loads, and obtain the corresponding arc current and voltage waveform data. The load change includes, but is not limited to, switching the right - hand load among R, L, and C components. For example, when R = 0Ω and L = 3.6mH, test by respectively matching with L = 0.6256H, R = 0.5Ω, and C = 0.5mF on the right - hand side;

[0011] Wavelet basis optimization steps:

[0012] Select wavelet bases that are effective for extracting fault arc characteristics from five wavelet basis functions: Sym4, Db4, Coif2, Bior3.3, and Rbio3.1, and process the arc simulation data output by the Cassie arc model under the above - mentioned multiple loads;

[0013] Based on the peak - to - mean ratio feature of the cD1 detail coefficient, calculate and compare the half - wave peak - to - mean ratios of the arc fault waveforms generated by each load under different wavelet bases, and select the wavelet basis that maximizes the half - wave peak - to - mean ratio in all load cases as the optimal wavelet basis, and determine that the optimal wavelet basis is db4;

[0014] Anti - spectral aliasing wavelet transform steps:

[0015] For the collected current signal to be detected, use discrete wavelet transform for analysis. In the two - downsampling link of discrete wavelet transform, abandon the traditional even - downsampling method, and adopt an anti - spectral aliasing improved wavelet transform algorithm based on odd - downsampling before adding nodes;

[0016] Preferably, the odd - downsampling algorithm includes: setting the odd - downsampling sequence relationship as y(n)=x(2n - 1), and performing Z - transform on it to obtain Combined with the Z - transform displacement - invariant property, through transformation Pre - process the sequence so that the spectral aliasing part can be cancelled before the subsequent two - interpolation, and finally obtain a wavelet decomposition and reconstruction signal that meets the condition of eliminating spectral aliasing;

[0017] Fault arc criterion determination steps:

[0018] Extract the cD1 detail coefficient after the above anti - spectral aliasing wavelet transform, and calculate its mean value according to the following formula Variance Energy value where N is the number of samples, is the mean value of the cD1 detail coefficient;

[0019] Calculate the ratio K2 / K1 of the mean value, variance, and energy value corresponding to the cD1 wavelet coefficients after and before the occurrence of a faulty arc, as well as the half-wave peak-to-average ratio based on the cD1 detail coefficients. Compare the above indicators with the preset thresholds. When the half-wave peak-to-average ratio or the ratio of other indicators exceeds the corresponding preset threshold, it is determined that a faulty arc has occurred.

[0020] Preferably, the Cassie arc model is constructed based on the following assumptions: It is assumed that the temperature is uniformly distributed across the arc cross-section, and the boundary of the channel, i.e., its diameter, is relatively clear. The impedance outside the diameter is extremely large. The temperature of the arc is relatively stable in time and space. The boundary of the gas channel varies with the magnitude of the current, and the energy dissipation rate is proportional to the change in the arc column cross-section. Its In the formula, g is the arc conductance; τ is the arc time constant; u h is the instantaneous value of the arc voltage gradient; u c is the arc voltage constant.

[0021] Preferably, during the derivation of the anti-spectrum aliasing improved wavelet transform algorithm, the decimation and interpolation processes in the traditional Mallat algorithm are deeply analyzed: In the decimation process, the Z-transform relationship of traditional even decimation is In the present invention, odd decimation and corresponding transformations are used to partially cancel the spectrum aliasing, so that the low-frequency output result of the final wavelet reconstructed signal is The output result of the detail part is The reconstructed signal is The spectrum aliasing part containing -z in the traditional algorithm is eliminated; in the interpolation process, based on the Z-transform relationship Y(Z) = X(Z 2 ), combined with the improved decimation result above, it is ensured that the signal can be accurately reconstructed after interpolation and filtering, avoiding signal distortion caused by spectrum aliasing.

[0022] The precision current sensor is installed at the key monitoring nodes of the low-voltage distribution line. Its sampling frequency can meet the requirements of capturing the transient current changes during the occurrence of a faulty arc and is used to collect the line current signal in real time;

[0023] The data processing unit: integrates a software module that executes the arc detection method based on wavelet transform described in claim 1, including building an arc simulation model using the Matlab / Simulink environment, performing discrete wavelet transform using the selected db4 wavelet basis, executing the anti-spectrum aliasing improved wavelet transform algorithm, and calculating the faulty arc criterion indicators;

[0024] Judgment and alarm unit: It stores preset thresholds determined through statistical analysis of a large amount of experimental data and actual working conditions under different load types. It compares the fault feature vector output by the data processing unit with the preset thresholds, and judges whether a fault arc occurs based on the half-wave peak-to-average ratio threshold crossing criterion and other indicators. Once a fault is determined, it immediately triggers an alarm device to send an alarm signal. The alarm device includes but is not limited to an audible and visual alarm and a communication module that sends fault information to the monitoring terminal.

[0025] Preferably, the sampling frequency of the current sensor of the data acquisition unit is not lower than 5 kHz, ensuring that it can accurately capture the high-frequency transient current changes generated by the fault arc and provide a high-quality data basis for subsequent wavelet analysis.

[0026] Preferably, the preset thresholds of the judgment and alarm unit are stored in an updatable storage medium, and can be dynamically adjusted according to subsequent new experimental data, changes in different power consumption scenarios, or system upgrade requirements, so as to continuously improve the accuracy and adaptability of fault arc recognition.

[0027] Preferably, when the data processing unit performs wavelet transform, it has the function of preprocessing the acquired signal. The preprocessing includes but is not limited to operations such as noise removal and filtering, so as to improve the signal quality and further enhance the ability of wavelet transform to extract fault arc features.

[0028] Preferably, the system further includes a data storage unit for storing the acquired original current data, processed intermediate data, simulation model parameters, wavelet transform results, and calculation results of fault arc criterion indicators, etc., for subsequent data analysis, system optimization, and fault traceability query.

[0029] Compared with the prior art, the present invention provides an arc detection method and system based on wavelet transform, having the following beneficial effects:

[0030] The anti-spectrum aliasing improved wavelet transform algorithm proposed by the present invention fundamentally overcomes the problem of spectrum aliasing in traditional wavelet transform, greatly improving the processing accuracy of wavelet analysis for fault arc signals. By accurately eliminating the spectrum aliasing phenomenon, the fault features can be extracted more clearly, effectively reducing the misjudgment rate and laying a solid foundation for the accurate detection of fault arcs;

[0031] With the help of the wavelet basis optimization strategy, db4 is determined as the optimal wavelet basis, which is significantly innovative in the field of fault arc detection. The db4 wavelet basis shows excellent feature extraction ability for fault arc signals. Especially when dealing with different load conditions, it can highlight key features such as the half-wave peak-to-average ratio, providing a more reliable and distinguishable basis for subsequent fault arc judgment and greatly improving the detection accuracy;

[0032] The multi-index fault arc criterion constructed based on wavelet detail coefficients comprehensively considers information from multiple dimensions such as mean, variance, energy, and peak-to-average ratio, breaking the limitations of traditional single-threshold judgment. Combining with the criterion of the half-wave peak-to-average ratio exceeding the limit enables the entire detection method to adapt to different electricity usage scenarios. Whether it is a complex industrial electricity environment or a changing household electricity occasion, it can accurately distinguish normal current from fault arc current, greatly improving the correct rate of fault arc recognition and effectively ensuring the safe and stable operation of the low-voltage distribution system;

[0033] From the perspective of the overall system design, the present invention fully takes into account the actual application requirements. In terms of hardware, the selected high-precision current sensor has a reasonable cost, is easy to install and deploy, and will not bring too high a transformation cost to the existing power system; the software algorithm is efficient and stable, can monitor the line current in real time online, and can detect potential fault arc hazards in a timely manner. The entire detection system has extremely high practical value and broad promotion prospects, and is expected to become the mainstream technical solution in the future field of power fault detection. Brief Description of the Drawings

[0034] Figure 1 An arc simulation model is built for the present invention using Matlab / Simulink.

[0035] Figure 2 For Figure 1 inside the Cassie / Mayr mathematical model module in [], the differential equation solving algorithm for implementing the two mathematical models is realized. Detailed Embodiment

[0036] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0037] The present invention provides an arc detection method based on wavelet transform as shown in Figure 1-2 and includes the following steps:

[0038] Arc fault model construction step:

[0039] Select the Cassie arc model, use Matlab / Simulink to build an arc simulation model, and set the power supply parameters. Among them, the peak value of the power supply voltage is 220V, the frequency is 50Hz, and the initial phase angle is 90°. The model parameters are set as Uc = 130V, τ = 2.25e-4 seconds = 225 us, g(0) = 1.17e-4 S = 0.000117 S;

[0040] By changing the load type in the model, simulate the arc generation under various different load combinations including at least inductive, capacitive, and resistive loads, and obtain the corresponding arc current and voltage waveform data. The load change includes, but is not limited to, switching the right - hand load among R, L, and C components. For example, when R = 0Ω and L = 3.6mH, test with L = 0.6256H, R = 0.5Ω, and C = 0.5mF on the right - hand side respectively;

[0041] Wavelet basis optimization steps:

[0042] Select wavelet bases effective for fault arc feature extraction from five wavelet basis functions: Sym4, Db4, Coif2, Bior3.3, and Rbio3.1, and process the arc simulation data output by the Cassie arc model under the above - mentioned various loads;

[0043] Based on the peak - to - mean ratio feature of the cD1 detail coefficient, calculate and compare the half - wave peak - to - mean ratios of the arc fault waveforms generated by each load under different wavelet bases, and select the wavelet basis that maximizes the half - wave peak - to - mean ratio in all load cases as the optimal wavelet basis, and determine that the optimal wavelet basis is db4;

[0044] Anti - spectral aliasing wavelet transform steps:

[0045] For the collected current signal to be detected, perform analysis using discrete wavelet transform. In the two - downsampling link of the discrete wavelet transform, abandon the traditional even - downsampling method and adopt an anti - spectral aliasing improved wavelet transform algorithm based on odd - downsampling before adding nodes;

[0046] The odd - downsampling algorithm includes: setting the odd - downsampling sequence relationship as y(n)=x(2n - 1), and performing Z - transform on it to obtain Combined with the Z - transform shift - invariant property, through transformation Pre - process the sequence so that the spectral aliasing part can be canceled before the subsequent two - interpolation, and finally obtain a wavelet decomposition and reconstruction signal that meets the condition of eliminating spectral aliasing;

[0047] Fault arc criterion determination steps:

[0048] Extract the cD1 detail coefficient after the above anti - spectral aliasing wavelet transform, and calculate its mean value according to the following formula Variance Energy value where N is the number of samples, is the mean value of the cD1 detail coefficient;

[0049] Calculate the ratios K2 / K1 of the means, variances, and energy values corresponding to the cD1 wavelet coefficients after and before the occurrence of a faulty arc, as well as the half-wave peak-to-average ratio based on the cD1 detail coefficients. Compare the above indicators with the preset thresholds. When the half-wave peak-to-average ratio or the ratio of other indicators exceeds the corresponding preset threshold, it is determined that a faulty arc has occurred.

[0050] The Cassie arc model is constructed based on the following assumptions: It is assumed that the temperature is uniformly distributed across the arc cross-section, and the boundary of the channel, i.e., its diameter, is relatively well-defined. The impedance outside the diameter is extremely large. The temperature of the arc is relatively stable in both time and space. The boundary of the gas channel varies with the magnitude of the current, and the energy dissipation rate is proportional to the change in the arc column cross-section. Its where g is the arc conductance; τ is the arc time constant; u h is the instantaneous value of the arc voltage gradient; u c is the arc voltage constant.

[0051] During the derivation of the anti-spectrum aliasing improved wavelet transform algorithm, the decimation and interpolation steps in the traditional Mallat algorithm are deeply analyzed: In the decimation step, the Z-transform relationship of traditional even decimation is The present invention realizes partial cancellation of spectrum aliasing through odd decimation and corresponding transformations, such that the low-frequency output result of the final wavelet reconstructed signal is The output result of the detail part is The reconstructed signal is The spectrum aliasing part containing -z in the traditional algorithm is eliminated; in the interpolation step, based on the Z-transform relationship Y(Z) = X(Z 2 ), combined with the improved decimation result above, it is ensured that the signal can be accurately reconstructed after interpolation and filtering, avoiding signal distortion caused by spectrum aliasing.

[0052] The precision current sensor is installed at the key monitoring nodes of the low-voltage distribution line, and its sampling frequency can meet the requirement of capturing the transient current change during the occurrence of a faulty arc, and is used to collect the line current signal in real time;

[0053] The data processing unit: integrated with a software module for executing the arc detection method based on wavelet transform described in claim 1, including building an arc simulation model using the Matlab / Simulink environment, performing discrete wavelet transform using the selected db4 wavelet basis, executing the anti-spectrum aliasing improved wavelet transform algorithm, and calculating the faulty arc criterion indicators;

[0054] Judgment and alarm unit: It stores preset thresholds determined by statistical analysis of a large amount of experimental data and actual working conditions under different load types, compares the fault feature vector output by the data processing unit with the preset thresholds, and judges whether a fault arc occurs based on the half-wave peak-to-average ratio threshold exceeding criterion and other indicators. Once a fault is determined, it immediately triggers an alarm device to send an alarm signal. The alarm device includes, but is not limited to, an audible and visual alarm and a communication module that sends fault information to the monitoring terminal.

[0055] The sampling frequency of the current sensor of the data acquisition unit is not less than 5 kHz, ensuring that it can accurately capture the high-frequency transient current changes generated by the fault arc and provide a high-quality data basis for subsequent wavelet analysis.

[0056] The preset thresholds of the judgment and alarm unit are stored in a replaceable storage medium and can be dynamically adjusted according to subsequent new experimental data, changes in different power usage scenarios, or system upgrade requirements to continuously improve the accuracy and adaptability of fault arc recognition.

[0057] When performing wavelet transform, the data processing unit has the function of preprocessing the acquired signal. The preprocessing includes, but is not limited to, operations such as noise removal and filtering to improve the signal quality and further enhance the ability of wavelet transform to extract fault arc features.

[0058] The system also includes a data storage unit for storing the acquired original current data, processed intermediate data, simulation model parameters, wavelet transform results, and calculation results of fault arc criterion indicators, etc., for subsequent data analysis, system optimization, and fault traceability query.

[0059] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An arc detection method based on wavelet transform, characterized in that It includes the following steps: Arc fault model construction step: Select the Cassie arc model, use Matlab / Simulink to build an arc simulation model, and set the power supply parameters. The peak value of the power supply voltage is 220V, the frequency is 50Hz, and the initial phase angle is 90°. The model parameters are set as Uc = 130V, τ = 2.25e-4 s = 225 us, g(0) = 1.17e-4 S = 0.000117 S; By changing the load type in the model, simulate the arc occurrence under at least various different load combinations including inductive, capacitive, and resistive, and obtain the corresponding arc current and voltage waveform data. The load change includes but is not limited to switching the right-side load among R, L, and C components. For example, when R = 0Ω and L = 3.6mH, test by respectively matching with L = 0.6256H, R = 0.5Ω, and C = 0.5mF on the right side; Wavelet basis optimization step: Select the wavelet basis effective for fault arc feature extraction from five wavelet basis functions of Sym4, Db4, Coif2, Bior3.3, and Rbio3.1, and process the arc simulation data output by the Cassie arc model under the above various loads; Based on the peak-to-average ratio feature of the cD1 detail coefficient, calculate and compare the half-wave peak-to-average ratio of the arc fault waveforms generated by each load under different wavelet bases, and select the wavelet basis with the largest half-wave peak-to-average ratio in various load cases as the optimal wavelet basis, and determine that the optimal wavelet basis is db4; Anti-spectrum aliasing wavelet transform step: For the collected current signal to be detected, use discrete wavelet transform for analysis. In the two-downsampling link of discrete wavelet transform, abandon the traditional even downsampling method, and adopt an anti-spectrum aliasing improved wavelet transform algorithm based on odd downsampling before adding nodes; The odd extraction algorithm includes: setting the odd extraction sequence relationship as y(n) = x(2n - 1), and performing Z-transform on it to obtain Combined with the shift-invariant property of the Z-transform, through transformation preprocess the sequence so that the aliased part of the spectrum can be cancelled before the subsequent two interpolations, and finally obtain the wavelet decomposition and reconstruction signal that meets the condition of eliminating spectral aliasing; Fault arc criterion determination step: Extract the cD1 detail coefficients after the above anti-aliasing wavelet transform and calculate their mean according to the following formula Variance Energy value where N is the number of samples, is the mean of the cD1 detail coefficients; Calculate the ratios K2 / K1 of the mean value, variance, and energy value corresponding to the cD1 wavelet coefficients after and before the occurrence of the fault arc, and the half-wave peak-to-average ratio based on the cD1 detail coefficient, and compare the above indicators with the preset thresholds. When the half-wave peak-to-average ratio or the ratio of other indicators exceeds the corresponding preset thresholds, it is determined that a fault arc has occurred.

2. The arc detection method based on wavelet transform according to claim 1, characterized in that: The Cassie arc model is constructed based on the following assumptions: it is assumed that the temperature is uniformly distributed across the arc cross-section, and the boundary of the channel, i.e., its diameter, is relatively well-defined, the impedance outside the diameter is very large, the temperature of the arc is relatively stable in time and space, the boundary of the gas channel varies with the magnitude of the current, and the energy dissipation rate is proportional to the change in the arc column cross-section. Its where g is the arc conductance; τ is the arc time constant; y h is the instantaneous value of the arc voltage gradient; u c is the arc voltage constant.

3. The arc detection method based on wavelet transform according to claim 1, characterized in that: In the derivation process of the anti-spectrum aliasing improved wavelet transform algorithm, the decimation and interpolation links in the traditional Mallat algorithm are deeply analyzed: in the decimation link, the Z-transform relationship of traditional even decimation is Y The present invention realizes partial cancellation of spectrum aliasing through odd decimation and corresponding transformation, so that the low-frequency output result of the final wavelet reconstruction signal is The output result of the detail part is The reconstructed signal is The spectrum aliasing part containing -z in the traditional algorithm is eliminated; in the interpolation link, according to the Z-transform relationship Y(Z)=X(Z 2 ), combined with the above improved decimation result, it is ensured that the signal can be accurately reconstructed after interpolation and filtering, avoiding signal distortion caused by spectrum aliasing.

4. An arc detection system based on wavelet transform, characterized in that: Data acquisition unit: Configure a high-precision current sensor, which is installed at the key monitoring nodes of the low-voltage distribution line, and its sampling frequency can meet the requirement of capturing the transient current change when a fault arc occurs, and is used to collect the line current signal in real time; Data processing unit: Integrate a software module that executes the arc detection method based on wavelet transform described in claim 1, including building an arc simulation model using the Matlab / Simulink environment, performing discrete wavelet transform using the selected db4 wavelet basis, executing the anti-spectrum aliasing improved wavelet transform algorithm, and calculating the fault arc criterion indicators; Judgment and alarm unit: It stores preset thresholds determined through statistical analysis of a large amount of experimental data and actual working conditions under different load types, compares the fault feature vector output by the data processing unit with the preset thresholds, and judges whether a faulty arc occurs based on the half-wave peak-to-average ratio threshold exceeding criterion and other indicators. Once a fault is determined, it immediately triggers an alarm device to send an alarm signal. The alarm device includes but is not limited to an audible and visual alarm and a communication module that sends fault information to the monitoring terminal.

5. The arc detection method and system based on wavelet transform according to claim 4, characterized in that: The sampling frequency of the current sensor of the data acquisition unit is not less than 5 kHz, ensuring that it can accurately capture the high-frequency transient current changes generated by the faulty arc and providing a high-quality data basis for subsequent wavelet analysis.

6. The arc detection method and system based on wavelet transform according to claim 4, characterized in that: The preset thresholds of the judgment and alarm unit are stored in an updatable storage medium and can be dynamically adjusted according to subsequent new experimental data, changes in different power consumption scenarios, or system upgrade requirements to continuously improve the accuracy and adaptability of faulty arc recognition.

7. The arc detection method and system based on wavelet transform according to claim 4, characterized in that: When performing wavelet transform, the data processing unit has the function of preprocessing the collected signal. The preprocessing includes but is not limited to operations such as noise removal and filtering to improve the signal quality and further enhance the ability of wavelet transform to extract faulty arc features.

8. The arc detection method and system based on wavelet transform according to claim 4, characterized in that: The system also includes a data storage unit for storing the collected original current data, processed intermediate data, simulation model parameters, wavelet transform results, and calculation results of faulty arc criterion indicators, etc., for subsequent data analysis, system optimization, and fault traceability query.

Images