A load series arc detection method, system and device
By calculating the volt-ampere characteristic curve and matching comparison, combining discrete wavelet transformation and LSTM-Attention regression model, the problems of difficult and high misjudgment rate in series arc fault detection in the prior art are solved, and efficient and accurate fault detection and positioning are achieved.
Patent Information
- Application Number
- CN202411803446.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-10
- Publication Date
- 2025-03-07
- Estimated Expiration
- 2044-12-10
AI Technical Summary
The existing fault arc detection method is difficult to detect when a series arc fault occurs in resistive load, and the response is slow, and the calculation accuracy becomes worse when the grid frequency fluctuates, which can easily lead to misjudgment.
By collecting the circuit signal, the volt-ampere characteristic standard curve and the target volt-ampere characteristic curve are calculated, and the matching comparison is performed to determine whether it is a fault arc, and fault location is performed through discrete wavelet transformation, singular value decomposition and LSTM-Attention regression model.
Accurate detection and positioning of series arc faults is achieved, detection efficiency and accuracy are improved, and misjudgment is reduced.
Smart Images
Figure CN119270007B_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the technical field of fault arc detection, and specifically, relates to a load series arc detection method, system and device. Background Art
[0002] An electric arc is a gas ion discharge phenomenon, also a plasma. It is generally a high-temperature discharge phenomenon caused by aging lines, loose connections, poor contact, etc. When an electric arc occurs, a large amount of heat is released, which may ignite surrounding objects, causing fires or even explosions. Traditional electrical protection devices, such as fuses and circuit breakers, can only protect against short-circuit faults, and residual current protectors can only protect against grounding faults. The effective value of the current of a series arc fault is much smaller than that of a metallic short circuit. For a series fault arc occurring on the line, the amplitude of the fault current is much smaller than the action value of the protection device, allowing the series arc to burn hidden for a long time, posing a serious safety hazard.
[0003] Existing arc fault detection methods generally obtain current waveform data through A / D sampling, and then use wavelet transform, Fourier transform, filtering and other methods to detect whether an arc occurs. However, when a series arc fault occurs in a resistive load, the harmonic content is small, the current waveform characteristics are not obvious, and the flat shoulder decreases with the increase of current. Fault detection is difficult and the response is slow. When the grid frequency fluctuates, the calculation accuracy will deteriorate, resulting in misjudgment, false operation and refusal to operate. Summary of the invention
[0004] In order to solve the above problems and technical defects, the embodiment of the present application adopts the following technical solution, a load series arc detection method, comprising the following steps:
[0005] Step 1, collecting target data to be detected and circuit signals of normal operation of the line under various loads, including load current waveform, load voltage, current, and fault arc voltage signal;
[0006] Step 2: Calculate and obtain a standard volt-ampere characteristic curve according to the circuit signal, wherein the volt-ampere characteristic curve includes a standard volt-ampere characteristic curve and a target volt-ampere characteristic curve of the target data to be detected;
[0007] Step 3, matching and comparing the target volt-ampere characteristic curve of the target data to be detected with the volt-ampere characteristic standard curve to determine whether the target data to be detected is a fault arc;
[0008] Step 4: If it is determined that the target data to be detected is a fault arc, the fault arc is located on the target data to be detected.
[0009] Preferably, the volt-ampere characteristic curve calculation formula is as follows:
[0010] U a =Ui +bK r / I q a
[0011] U a is the voltage across the arc, U i is the voltage drop, I q a is the arc current, K is the distance between the two ends of the electrode, and b, q, and r are constant coefficients.
[0012] Preferably, the process of matching and comparing the target volt-ampere characteristic curve with the volt-ampere characteristic standard curve is as follows:
[0013] Align the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve on the timeline according to the timestamp, and evenly set multiple detection time points on the timeline;
[0014] A time window is preset, and the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve are passed through the time window along with the time line;
[0015] Whenever a detection time point enters the time window, the comparison between the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve entering the time window begins, and when the detection time point leaves the time window, the comparison ends;
[0016] A judgment is made based on all the comparison results to determine whether the target volt-ampere characteristic curve is a fault arc.
[0017] Furthermore, the judgment process based on all the comparison results is as follows:
[0018] First, a first ratio threshold θ and a second ratio threshold δ are preset;
[0019] Each time the comparison result is abnormal, the first assignment γ is obtained;
[0020] Each time the comparison result is normal, the second assignment λ is obtained;
[0021] When the proportion of the first assigned value γ in all assigned values is greater than or equal to the first proportion threshold θ, it is determined that the target volt-ampere characteristic curve is a fault arc;
[0022] When the proportion of the second assigned value λ in all assigned values is greater than or equal to the second proportion threshold δ, it is determined that the target volt-ampere characteristic curve is a normal arc;
[0023] Otherwise, the target volt-ampere characteristic curve is judged to be a suspected arc, and whether the suspected arc is a fault arc is manually judged.
[0024] Furthermore, the time length of the time window is greater than a time interval between detecting adjacent detection time points and less than two time intervals between detecting adjacent detection time points.
[0025] Furthermore, the arc fault location process is as follows:
[0026] S01, performing discrete wavelet transform on the curve data corresponding to the first assignment γ in the target volt-ampere characteristic curve, calculating and obtaining a discrete signal, and generating a discrete wavelet coefficient sequence according to the discrete signal,
[0027] S02, performing singular value decomposition on the discrete wavelet coefficient sequence, obtaining and determining characteristic parameters of the current signal, and constructing a characteristic matrix;
[0028] S03. Use the LSTM-Attention regression model to predict faults on the feature matrix and obtain the fault location results.
[0029] Furthermore, before generating the discrete wavelet coefficient sequence, the discrete signal needs to be smoothed, and the smoothing formula is as follows:
[0030] s(n)=[g(n+1)+g(n-1)-2g(n)] / [[g(n)] 2 -g(n+1)·g(n-1)]
[0031] Among them, g(n) is the nth discrete signal, and s(n) is the signal after smoothing the nth discrete signal.
[0032] Furthermore, before constructing the feature matrix, the feature parameters need to be subjected to triple exponential smoothing to achieve filtering enhancement. The calculation formula for triple exponential smoothing is as follows:
[0033] t1(n)=β[g(n)-t1(n-1)]+t1(n-1)
[0034] t2(n)=β[t(n)-t2(n-1)]+t2(n-1)
[0035] t3(n)=β[t2(n)-t3(n-1)]+t3(n-1)
[0036] Among them, t i (n) is the result of the i-th exponential smoothing, i=1, 2, 3, β is the smoothing coefficient, 0<β<1.
[0037] A load series arc detection system, comprising:
[0038] A signal acquisition module, used to collect target data to be detected and circuit signals of normal operation of the line under various loads;
[0039] A curve generating module, used for calculating a standard volt-ampere characteristic curve according to a circuit signal, wherein the volt-ampere characteristic curve includes a standard volt-ampere characteristic curve and a target volt-ampere characteristic curve of target data to be detected;
[0040] A fault judgment module is used to match and compare the target volt-ampere characteristic curve of the target data to be detected with the volt-ampere characteristic standard curve to determine whether the target data to be detected is a fault arc;
[0041] The fault location module is used to locate the fault arc on the target data to be detected which is the fault arc.
[0042] A load series arc detection device, the device includes a service processor and a distributed memory, the service processor is connected to the memory, the distributed memory stores a service self-management program, and is configured to store machine-readable instructions. The service processor executes the service self-management program, and the instructions, when executed by the processor, implement the method as described above.
[0043] Compared with the prior art, the beneficial effects of the embodiments of the present application are:
[0044] The present application obtains a standard volt-ampere characteristic curve through circuit signal calculation, matches and compares the target volt-ampere characteristic curve of the target data to be detected with the standard volt-ampere characteristic curve, performs segmented comparison of the time window, and accurately determines whether the target data to be detected is a fault arc through multiple comparisons and judgments. At the same time, a discrete wavelet transform is performed on the curve data of the fault arc to obtain a discrete signal, generate a discrete wavelet coefficient sequence, perform singular value decomposition, construct a feature matrix, perform fault prediction, realize fault location, determine which type of fault it is, and finally complete fault detection with high detection efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In the attached picture:
[0046] Figure 1 A schematic diagram of the method steps of an embodiment of the present application;
[0047] Figure 2 A schematic diagram of the system structure of an embodiment of the present application;
[0048] Figure 3 Schematic diagram of the device structure of an embodiment of the present application. DETAILED DESCRIPTION
[0049] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Generally, the components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.
[0050] Example 1: Figure 1 As shown, a load series arc detection method comprises the following steps:
[0051] Collect the target data to be detected and the circuit signals of normal operation of the line under various loads, including load current waveform, load voltage, current, and fault arc voltage signals;
[0052] A volt-ampere characteristic standard curve is obtained by calculation according to the circuit signal, wherein the volt-ampere characteristic curve includes a volt-ampere characteristic standard curve and a target volt-ampere characteristic curve of the target data to be detected;
[0053] The calculation formula of the volt-ampere characteristic curve is as follows:
[0054] U a =U i +bK r / I q a
[0055] U a is the voltage across the arc, U i is the voltage drop, I q a is the arc current, K is the distance between the two ends of the electrode, and b, q, and r are constant coefficients.
[0056] Match and compare the target volt-ampere characteristic curve of the target data to be detected with the volt-ampere characteristic standard curve to determine whether the target data to be detected is a fault arc;
[0057] The process of matching and comparing the target volt-ampere characteristic curve with the volt-ampere characteristic standard curve is as follows:
[0058] Align the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve on the timeline according to the timestamp, and evenly set multiple detection time points on the timeline;
[0059] A time window is preset, and the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve are passed through the time window along with the time line;
[0060] Whenever a detection time point enters the time window, the comparison between the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve entering the time window begins, and when the detection time point leaves the time window, the comparison ends;
[0061] A judgment is made based on all the comparison results to determine whether the target volt-ampere characteristic curve is a fault arc.
[0062] The judgment process based on all the comparison results is as follows:
[0063] First, a first ratio threshold θ and a second ratio threshold δ are preset;
[0064] Each time the comparison result is abnormal, the first assignment γ is obtained;
[0065] Each time the comparison result is normal, the second assignment λ is obtained;
[0066] When the proportion of the first assigned value γ in all assigned values is greater than or equal to the first proportion threshold θ, it is determined that the target volt-ampere characteristic curve is a fault arc;
[0067] When the proportion of the second assigned value λ in all assigned values is greater than or equal to the second proportion threshold δ, it is determined that the target volt-ampere characteristic curve is a normal arc;
[0068] Otherwise, the target volt-ampere characteristic curve is judged to be a suspected arc, and whether the suspected arc is a fault arc is manually judged.
[0069] For example, the first ratio threshold θ is 20%, and the second ratio threshold δ is 90%;
[0070] When the proportion of the first assigned value γ is between 0% and 10% and the proportion of the second assigned value λ is between 90% and 100%, the target volt-ampere characteristic curve is a normal arc;
[0071] When the proportion of the first assigned value γ is between 10% and 20% and the proportion of the second assigned value λ is between 80% and 90%, the target volt-ampere characteristic curve is a suspected arc;
[0072] When the proportion of the first assigned value γ is between 20% and 100% and the proportion of the second assigned value λ is between 0% and 80%, the target volt-ampere characteristic curve is a fault arc.
[0073] The time length of the time window is greater than a time interval between detecting adjacent detection time points and less than two time intervals between detecting adjacent detection time points.
[0074] t <T<2t
[0075] Wherein, t is the time interval between detecting adjacent time points, and T is the preset time window length.
[0076] For example, the time length of the time window is 3s, and the time interval between adjacent detection time points is 2s, namely the 1st second, the 3rd second, the 5th second, the 7th second, ... When the 1st second enters the time window, the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve are compared, and the first comparison takes 1 to 4s. When the 3rd second enters the time window, the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve are compared again without interrupting the first comparison. Each comparison does not interfere with other comparisons. The second comparison takes 3 to 6s, the third comparison takes 5 to 8s...
[0077] If it is determined that the target data to be detected is a fault arc, the fault arc is located on the target data to be detected.
[0078] The arc fault location process is as follows:
[0079] S01, performing discrete wavelet transform on the curve data corresponding to the first assignment γ in the target volt-ampere characteristic curve, calculating and obtaining a discrete signal, and generating a discrete wavelet coefficient sequence according to the discrete signal,
[0080] Before generating the discrete wavelet coefficient sequence, the discrete signal needs to be smoothed. The smoothing formula is as follows:
[0081] s(n)=[g(n+1)+g(n-1)-2g(n)] / [[g(n)] 2 -g(n+1)·g(n-1)]
[0082] Among them, g(n) is the nth discrete signal, and s(n) is the signal after smoothing the nth discrete signal.
[0083] S02, performing singular value decomposition on the discrete wavelet coefficient sequence, obtaining and determining characteristic parameters of the current signal, and constructing a characteristic matrix;
[0084] Before constructing the feature matrix, the feature parameters need to be processed by triple exponential smoothing to achieve filtering enhancement. The calculation formula of triple exponential smoothing is as follows:
[0085] t1(n)=β[g(n)-t1(n-1)]+t1(n-1)
[0086] t2(n)=β[t(n)-t2(n-1)]+t2(n-1)
[0087] t3(n)=β[t2(n)-t3(n-1)]+t3(n-1)
[0088] Among them, t i (n) is the result of the i-th exponential smoothing, i=1, 2, 3, β is the smoothing coefficient, 0<β<1.
[0089] S03. Use the LSTM-Attention regression model to predict faults on the feature matrix and obtain the fault location results.
[0090] Example 2: Figure 2 As shown, a load series arc detection system comprises:
[0091] A signal acquisition module, used to collect target data to be detected and circuit signals of normal operation of the line under various loads;
[0092] A curve generating module, used for calculating a standard volt-ampere characteristic curve according to a circuit signal, wherein the volt-ampere characteristic curve includes a standard volt-ampere characteristic curve and a target volt-ampere characteristic curve of target data to be detected;
[0093] The calculation formula of the volt-ampere characteristic curve is as follows:
[0094] U a =U i +bK r / I q a
[0095] U a is the voltage across the arc, U i is the voltage drop, I q a is the arc current, K is the distance between the two ends of the electrode, and b, q, and r are constant coefficients.
[0096] A fault judgment module is used to match and compare the target volt-ampere characteristic curve of the target data to be detected with the volt-ampere characteristic standard curve to determine whether the target data to be detected is a fault arc;
[0097] The process of matching and comparing the target volt-ampere characteristic curve with the volt-ampere characteristic standard curve is as follows:
[0098] Align the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve on the timeline according to the timestamp, and evenly set multiple detection time points on the timeline;
[0099] A time window is preset, and the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve are passed through the time window along with the time line;
[0100] Whenever a detection time point enters the time window, the comparison between the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve entering the time window begins, and when the detection time point leaves the time window, the comparison ends;
[0101] A judgment is made based on all the comparison results to determine whether the target volt-ampere characteristic curve is a fault arc.
[0102] The judgment process based on all the comparison results is as follows:
[0103] First, a first ratio threshold θ and a second ratio threshold δ are preset;
[0104] Each time the comparison result is abnormal, the first assignment γ is obtained;
[0105] Each time the comparison result is normal, the second assignment λ is obtained;
[0106] When the proportion of the first assigned value γ in all assigned values is greater than or equal to the first proportion threshold θ, it is determined that the target volt-ampere characteristic curve is a fault arc;
[0107] When the proportion of the second assigned value λ in all assigned values is greater than or equal to the second proportion threshold δ, it is determined that the target volt-ampere characteristic curve is a normal arc;
[0108] Otherwise, the target volt-ampere characteristic curve is judged to be a suspected arc, and whether the suspected arc is a fault arc is manually judged.
[0109] A fault location module is used to locate the fault arc for the target data to be detected that is the fault arc;
[0110] If it is determined that the target data to be detected is a fault arc, the fault arc is located on the target data to be detected.
[0111] The arc fault location process is as follows:
[0112] S01, performing discrete wavelet transform on the curve data corresponding to the first assignment γ in the target volt-ampere characteristic curve, calculating and obtaining a discrete signal, and generating a discrete wavelet coefficient sequence according to the discrete signal,
[0113] Before generating the discrete wavelet coefficient sequence, the discrete signal needs to be smoothed. The smoothing formula is as follows:
[0114] s(n)=[g(n+1)+g(n-1)-2g(n)] / [[g(n)] 2 -g(n+1)·g(n-1)]
[0115] Among them, g(n) is the nth discrete signal, and s(n) is the signal after smoothing the nth discrete signal.
[0116] S02, performing singular value decomposition on the discrete wavelet coefficient sequence, obtaining and determining characteristic parameters of the current signal, and constructing a characteristic matrix;
[0117] Before constructing the feature matrix, the feature parameters need to be processed by triple exponential smoothing to achieve filtering enhancement. The calculation formula of triple exponential smoothing is as follows:
[0118] t1(n)=β[g(n)-t1(n-1)]+t1(n-1)
[0119] t2(n)=β[t(n)-t2(n-1)]+t2(n-1)
[0120] t3(n)=β[t2(n)-t3(n-1)]+t3(n-1)
[0121] Among them, t i (n) is the result of the i-th exponential smoothing, i=1, 2, 3, β is the smoothing coefficient, 0<β<1.
[0122] S03. Use the LSTM-Attention regression model to predict faults on the feature matrix and obtain the fault location results.
[0123] Example 3: Figure 3 As shown, a load series arc detection device includes a service processor and a distributed memory, wherein the service processor is connected to the memory, wherein the distributed memory stores a service self-management program configured to store machine-readable instructions, and the service processor executes the service self-management program, and when the instructions are executed by the processor, the method described in Example 1 is implemented.
[0124] The above-mentioned embodiments only express the preferred implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for ordinary technicians in this field, several modifications, improvements and substitutions can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application.
Claims
1. A load series arc detection method, characterized in that: The following steps are involved: Step 1, collecting target data to be detected and circuit signals of normal operation of the line under various loads, including load current waveform, load voltage, current, and fault arc voltage signal; Step 2: Calculate and obtain a volt-ampere characteristic curve according to the circuit signal, wherein the volt-ampere characteristic curve includes a volt-ampere characteristic standard curve and a target volt-ampere characteristic curve of the target data to be detected; Step 3: Match and compare the target volt-ampere characteristic curve of the target data to be detected with the volt-ampere characteristic standard curve to determine whether the target data to be detected is a fault arc. The matching and comparison process is as follows: Align the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve on the timeline according to the timestamp, and evenly set multiple detection time points on the timeline; The time extent of the time window is preset, and the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve are passed through the time window along with the time line; Whenever a detection time point enters the time window, the comparison between the target volt-ampere characteristic curve and the volt-ampere characteristic standard curve entering the time window begins, and when the detection time point leaves the time window, the comparison ends; Combining all the comparison results to determine whether the target volt-ampere characteristic curve is a fault arc; The judgment process is as follows: First, a first ratio threshold θ and a second ratio threshold δ are preset; Each time the comparison result is abnormal, the first assignment γ is obtained; Each time the comparison result is normal, the second assignment λ is obtained; When the proportion of the first assigned value γ in all assigned values is greater than or equal to the first proportion threshold θ, it is determined that the target volt-ampere characteristic curve is a fault arc; When the proportion of the second assigned value λ in all assigned values is greater than or equal to the second proportion threshold δ, it is determined that the target volt-ampere characteristic curve is a normal arc; Otherwise, the target volt-ampere characteristic curve is judged to be a suspected arc, and the suspected arc is manually judged to be a fault arc; Step 4: If it is determined that the target data to be detected is a fault arc, the fault arc is located on the target data to be detected.
2. A load series arc detection method according to claim 1, characterized in that: The volt-ampere characteristic curve calculation formula is as follows: U a =U i +bK r / I q a U a is the voltage across the arc, U i is the voltage drop, I q a is the arc current, K is the distance between the two ends of the electrode, and b, q, and r are constant coefficients.
3. A load series arc detection method according to claim 1, characterized in that: The time length of the time window is greater than a time interval for detecting adjacent detection time points, and less than two time intervals for detecting adjacent detection time points.
4. A load series arc detection method according to claim 3, characterized in that: The arc fault location process is as follows: S01, performing discrete wavelet transform on the curve data corresponding to the first assignment γ in the target volt-ampere characteristic curve, calculating and obtaining a discrete signal, and generating a discrete wavelet coefficient sequence according to the discrete signal, S02, performing singular value decomposition on the discrete wavelet coefficient sequence, obtaining and determining characteristic parameters of the current signal, and constructing a characteristic matrix; S03. Use the LSTM-Attention regression model to predict faults on the feature matrix and obtain the fault location results.
5. A load series arc detection method according to claim 4, characterized in that: Before generating the discrete wavelet coefficient sequence, the discrete signal needs to be smoothed. The smoothing formula is as follows: s(n)=[g(n+1)+g(n-1)-2g(n)] / [[g(n)] 2 -g(n+1)·g(n-1)] Among them, g(n) is the nth discrete signal, and s(n) is the signal after smoothing the nth discrete signal.
6. A load series arc detection method according to claim 5, characterized in that: Before constructing the feature matrix, the feature parameters need to be processed by triple exponential smoothing to achieve filtering enhancement. The calculation formula of triple exponential smoothing is as follows: t1(n)=β[g(n)-t1(n-1)]+t1(n-1) t2(n)=β[t(n)-t2(n-1)]+t2(n-1) t3(n)=β[t2(n)-t3(n-1)]+t3(n-1) Among them, t i (n) is the result of the i-th exponential smoothing, i=1, 2, 3, β is the smoothing coefficient, 0<β<1.
7. A load series arc detection device, characterized in that: The device includes a service processor and a distributed memory, the service processor is connected to the memory, the distributed memory stores a service self-management program, and is configured to store machine-readable instructions. The service processor executes the service self-management program, and the instructions, when executed by the processor, implement the method described in claim 1.
Citation Information
Patent Citations
Fault arc detection method, device and equipment and readable storage medium
CN116520082A